diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -10,32213 +10,32213 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 5.6632,
+      "loss": 5.5802,
       "step": 1
     },
     {
       "epoch": 0.0,
       "learning_rate": 0.0,
-      "loss": 5.7062,
+      "loss": 5.589,
       "step": 2
     },
     {
       "epoch": 0.0,
-      "learning_rate": 6e-07,
-      "loss": 5.9497,
+      "learning_rate": 0.0,
+      "loss": 5.8413,
       "step": 3
     },
     {
       "epoch": 0.0,
-      "learning_rate": 1.2e-06,
-      "loss": 5.4786,
+      "learning_rate": 2e-06,
+      "loss": 5.2778,
       "step": 4
     },
     {
       "epoch": 0.0,
-      "learning_rate": 1.8e-06,
-      "loss": 5.6714,
+      "learning_rate": 4e-06,
+      "loss": 5.5308,
       "step": 5
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.4e-06,
-      "loss": 5.4817,
+      "learning_rate": 6e-06,
+      "loss": 5.3403,
       "step": 6
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.9999999999999997e-06,
-      "loss": 5.5102,
+      "learning_rate": 8e-06,
+      "loss": 5.3527,
       "step": 7
     },
     {
       "epoch": 0.0,
-      "learning_rate": 3.6e-06,
-      "loss": 5.3595,
+      "learning_rate": 1e-05,
+      "loss": 5.168,
       "step": 8
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.2e-06,
-      "loss": 5.5823,
+      "learning_rate": 1.2e-05,
+      "loss": 5.4157,
       "step": 9
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.8e-06,
-      "loss": 5.231,
+      "learning_rate": 1.4e-05,
+      "loss": 4.991,
       "step": 10
     },
     {
       "epoch": 0.01,
-      "learning_rate": 5.399999999999999e-06,
-      "loss": 5.1313,
+      "learning_rate": 1.6e-05,
+      "loss": 4.8795,
       "step": 11
     },
     {
       "epoch": 0.01,
-      "learning_rate": 5.999999999999999e-06,
-      "loss": 5.1219,
+      "learning_rate": 1.8e-05,
+      "loss": 4.8489,
       "step": 12
     },
     {
       "epoch": 0.01,
-      "learning_rate": 5.999999999999999e-06,
-      "loss": 5.2163,
+      "learning_rate": 2e-05,
+      "loss": 4.9404,
       "step": 13
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6.599999999999999e-06,
-      "loss": 4.9356,
+      "learning_rate": 2.2e-05,
+      "loss": 4.5799,
       "step": 14
     },
     {
       "epoch": 0.01,
-      "learning_rate": 7.2e-06,
-      "loss": 4.9701,
+      "learning_rate": 2.4e-05,
+      "loss": 4.7039,
       "step": 15
     },
     {
       "epoch": 0.01,
-      "learning_rate": 7.799999999999998e-06,
-      "loss": 5.0139,
+      "learning_rate": 2.6e-05,
+      "loss": 4.7245,
       "step": 16
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8.4e-06,
-      "loss": 5.069,
+      "learning_rate": 2.8e-05,
+      "loss": 4.7614,
       "step": 17
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8.999999999999999e-06,
-      "loss": 4.7759,
+      "learning_rate": 3e-05,
+      "loss": 4.5656,
       "step": 18
     },
     {
       "epoch": 0.01,
-      "learning_rate": 9.6e-06,
-      "loss": 4.7786,
+      "learning_rate": 3.2e-05,
+      "loss": 4.5534,
       "step": 19
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.02e-05,
-      "loss": 5.0655,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 4.8297,
       "step": 20
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.0799999999999998e-05,
-      "loss": 4.7218,
+      "learning_rate": 3.6e-05,
+      "loss": 4.5089,
       "step": 21
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.14e-05,
-      "loss": 5.1393,
+      "learning_rate": 3.8e-05,
+      "loss": 4.9362,
       "step": 22
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.1999999999999999e-05,
-      "loss": 4.8822,
+      "learning_rate": 4e-05,
+      "loss": 4.6566,
       "step": 23
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.26e-05,
-      "loss": 4.5621,
+      "learning_rate": 4.2000000000000004e-05,
+      "loss": 4.4095,
       "step": 24
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.3199999999999997e-05,
-      "loss": 4.7078,
+      "learning_rate": 4.4e-05,
+      "loss": 4.5129,
       "step": 25
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.3799999999999998e-05,
-      "loss": 4.8129,
+      "learning_rate": 4.6e-05,
+      "loss": 4.6861,
       "step": 26
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.44e-05,
-      "loss": 4.6493,
+      "learning_rate": 4.8e-05,
+      "loss": 4.4625,
       "step": 27
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.4999999999999999e-05,
-      "loss": 4.3486,
+      "learning_rate": 5e-05,
+      "loss": 4.1955,
       "step": 28
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.5599999999999996e-05,
-      "loss": 4.6841,
+      "learning_rate": 5.2e-05,
+      "loss": 4.542,
       "step": 29
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.6199999999999997e-05,
-      "loss": 4.581,
+      "learning_rate": 5.4e-05,
+      "loss": 4.4487,
       "step": 30
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.68e-05,
-      "loss": 4.4972,
+      "learning_rate": 5.6e-05,
+      "loss": 4.3978,
       "step": 31
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.74e-05,
-      "loss": 4.8282,
+      "learning_rate": 5.800000000000001e-05,
+      "loss": 4.7367,
       "step": 32
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.7999999999999997e-05,
-      "loss": 4.5163,
+      "learning_rate": 6e-05,
+      "loss": 4.4123,
       "step": 33
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.8599999999999998e-05,
-      "loss": 4.3629,
+      "learning_rate": 6.2e-05,
+      "loss": 4.2376,
       "step": 34
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.92e-05,
-      "loss": 4.6336,
+      "learning_rate": 6.4e-05,
+      "loss": 4.5116,
       "step": 35
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.98e-05,
-      "loss": 4.8354,
+      "learning_rate": 6.6e-05,
+      "loss": 4.7934,
       "step": 36
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.04e-05,
-      "loss": 4.5484,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 4.4381,
       "step": 37
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.1e-05,
-      "loss": 4.6374,
+      "learning_rate": 7.000000000000001e-05,
+      "loss": 4.5595,
       "step": 38
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.1599999999999996e-05,
-      "loss": 4.702,
+      "learning_rate": 7.2e-05,
+      "loss": 4.6284,
       "step": 39
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.2199999999999998e-05,
-      "loss": 4.6855,
+      "learning_rate": 7.4e-05,
+      "loss": 4.6479,
       "step": 40
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.28e-05,
-      "loss": 4.4869,
+      "learning_rate": 7.6e-05,
+      "loss": 4.4379,
       "step": 41
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.34e-05,
-      "loss": 4.4182,
+      "learning_rate": 7.8e-05,
+      "loss": 4.3186,
       "step": 42
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.3999999999999997e-05,
-      "loss": 4.6218,
+      "learning_rate": 8e-05,
+      "loss": 4.5889,
       "step": 43
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.4599999999999998e-05,
-      "loss": 4.3735,
+      "learning_rate": 8.2e-05,
+      "loss": 4.3214,
       "step": 44
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.52e-05,
-      "loss": 4.7456,
+      "learning_rate": 8.400000000000001e-05,
+      "loss": 4.7199,
       "step": 45
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.5799999999999997e-05,
-      "loss": 4.9274,
+      "learning_rate": 8.599999999999999e-05,
+      "loss": 4.7496,
       "step": 46
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.6399999999999995e-05,
-      "loss": 4.8967,
+      "learning_rate": 8.8e-05,
+      "loss": 4.7364,
       "step": 47
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.6999999999999996e-05,
-      "loss": 5.0151,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 4.8753,
       "step": 48
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.7599999999999997e-05,
-      "loss": 5.0408,
+      "learning_rate": 9.2e-05,
+      "loss": 4.949,
       "step": 49
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 4.9853,
+      "learning_rate": 9.400000000000001e-05,
+      "loss": 5.027,
       "step": 50
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.88e-05,
-      "loss": 4.4185,
+      "learning_rate": 9.6e-05,
+      "loss": 4.4543,
       "step": 51
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.94e-05,
-      "loss": 4.5339,
+      "learning_rate": 9.800000000000001e-05,
+      "loss": 4.6299,
       "step": 52
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.9999999999999997e-05,
-      "loss": 4.5095,
+      "learning_rate": 0.0001,
+      "loss": 4.5232,
       "step": 53
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.06e-05,
-      "loss": 4.384,
+      "learning_rate": 0.000102,
+      "loss": 4.3683,
       "step": 54
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.119999999999999e-05,
-      "loss": 4.4593,
+      "learning_rate": 0.000104,
+      "loss": 4.5343,
       "step": 55
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.1799999999999994e-05,
-      "loss": 4.5121,
+      "learning_rate": 0.000106,
+      "loss": 4.5478,
       "step": 56
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.2399999999999995e-05,
-      "loss": 4.4289,
+      "learning_rate": 0.000108,
+      "loss": 4.4856,
       "step": 57
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.2999999999999996e-05,
-      "loss": 4.3921,
+      "learning_rate": 0.00011,
+      "loss": 4.4394,
       "step": 58
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.36e-05,
-      "loss": 4.4353,
+      "learning_rate": 0.000112,
+      "loss": 4.4379,
       "step": 59
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.42e-05,
-      "loss": 4.4227,
+      "learning_rate": 0.000114,
+      "loss": 4.4407,
       "step": 60
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.48e-05,
-      "loss": 4.4845,
+      "learning_rate": 0.00011600000000000001,
+      "loss": 4.5579,
       "step": 61
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.539999999999999e-05,
-      "loss": 4.3991,
+      "learning_rate": 0.000118,
+      "loss": 4.5174,
       "step": 62
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.5999999999999994e-05,
-      "loss": 4.7688,
+      "learning_rate": 0.00012,
+      "loss": 4.8199,
       "step": 63
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.6599999999999995e-05,
-      "loss": 4.2236,
+      "learning_rate": 0.000122,
+      "loss": 4.2503,
       "step": 64
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.7199999999999996e-05,
-      "loss": 4.4618,
+      "learning_rate": 0.000124,
+      "loss": 4.4843,
       "step": 65
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.78e-05,
-      "loss": 4.5446,
+      "learning_rate": 0.000126,
+      "loss": 4.6071,
       "step": 66
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.84e-05,
-      "loss": 4.5508,
+      "learning_rate": 0.000128,
+      "loss": 4.5811,
       "step": 67
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.9e-05,
-      "loss": 4.5243,
+      "learning_rate": 0.00013000000000000002,
+      "loss": 4.5564,
       "step": 68
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.96e-05,
-      "loss": 4.5106,
+      "learning_rate": 0.000132,
+      "loss": 4.5395,
       "step": 69
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.02e-05,
-      "loss": 4.4084,
+      "learning_rate": 0.000134,
+      "loss": 4.4299,
       "step": 70
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.08e-05,
-      "loss": 4.4932,
+      "learning_rate": 0.00013600000000000003,
+      "loss": 4.5831,
       "step": 71
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.14e-05,
-      "loss": 4.2541,
+      "learning_rate": 0.00013800000000000002,
+      "loss": 4.2683,
       "step": 72
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.2e-05,
-      "loss": 4.5389,
+      "learning_rate": 0.00014000000000000001,
+      "loss": 4.6712,
       "step": 73
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.259999999999999e-05,
-      "loss": 4.2515,
+      "learning_rate": 0.00014199999999999998,
+      "loss": 4.3491,
       "step": 74
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.319999999999999e-05,
-      "loss": 4.7623,
+      "learning_rate": 0.000144,
+      "loss": 4.9766,
       "step": 75
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.3799999999999994e-05,
-      "loss": 4.5547,
+      "learning_rate": 0.000146,
+      "loss": 4.6173,
       "step": 76
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.4399999999999995e-05,
-      "loss": 4.3733,
+      "learning_rate": 0.000148,
+      "loss": 4.4911,
       "step": 77
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.4999999999999996e-05,
-      "loss": 4.3419,
+      "learning_rate": 0.00015,
+      "loss": 4.4158,
       "step": 78
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.56e-05,
-      "loss": 4.3568,
+      "learning_rate": 0.000152,
+      "loss": 4.5255,
       "step": 79
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.62e-05,
-      "loss": 4.6176,
+      "learning_rate": 0.000154,
+      "loss": 4.729,
       "step": 80
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.68e-05,
-      "loss": 4.3145,
+      "learning_rate": 0.000156,
+      "loss": 4.4216,
       "step": 81
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7399999999999993e-05,
-      "loss": 4.428,
+      "learning_rate": 0.000158,
+      "loss": 4.6069,
       "step": 82
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7999999999999994e-05,
-      "loss": 4.2687,
+      "learning_rate": 0.00016,
+      "loss": 4.4213,
       "step": 83
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.8599999999999995e-05,
-      "loss": 4.4546,
+      "learning_rate": 0.000162,
+      "loss": 4.7356,
       "step": 84
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.9199999999999997e-05,
-      "loss": 4.3374,
+      "learning_rate": 0.000164,
+      "loss": 4.5921,
       "step": 85
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.98e-05,
-      "loss": 4.2864,
+      "learning_rate": 0.00016600000000000002,
+      "loss": 4.4656,
       "step": 86
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.04e-05,
-      "loss": 4.6991,
+      "learning_rate": 0.00016800000000000002,
+      "loss": 4.8462,
       "step": 87
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.1e-05,
-      "loss": 4.274,
+      "learning_rate": 0.00017,
+      "loss": 4.4326,
       "step": 88
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.1599999999999994e-05,
-      "loss": 4.6373,
+      "learning_rate": 0.00017199999999999998,
+      "loss": 4.7751,
       "step": 89
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.2199999999999995e-05,
-      "loss": 4.2158,
+      "learning_rate": 0.000174,
+      "loss": 4.3423,
       "step": 90
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.279999999999999e-05,
-      "loss": 4.5819,
+      "learning_rate": 0.000176,
+      "loss": 4.7629,
       "step": 91
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.339999999999999e-05,
-      "loss": 4.265,
+      "learning_rate": 0.000178,
+      "loss": 4.4473,
       "step": 92
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.399999999999999e-05,
-      "loss": 4.1031,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 4.2477,
       "step": 93
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.459999999999999e-05,
-      "loss": 4.5446,
+      "learning_rate": 0.000182,
+      "loss": 4.9831,
       "step": 94
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.519999999999999e-05,
-      "loss": 4.7588,
+      "learning_rate": 0.000184,
+      "loss": 4.9647,
       "step": 95
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.5799999999999994e-05,
-      "loss": 4.5028,
+      "learning_rate": 0.000186,
+      "loss": 4.8039,
       "step": 96
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 4.7865,
+      "learning_rate": 0.00018800000000000002,
+      "loss": 4.8747,
       "step": 97
     },
     {
       "epoch": 0.05,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 4.4749,
+      "learning_rate": 0.00019,
+      "loss": 4.6304,
       "step": 98
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.76e-05,
-      "loss": 4.8534,
+      "learning_rate": 0.000192,
+      "loss": 4.9462,
       "step": 99
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.82e-05,
-      "loss": 4.797,
+      "learning_rate": 0.000194,
+      "loss": 4.9919,
       "step": 100
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.88e-05,
-      "loss": 4.4968,
+      "learning_rate": 0.00019600000000000002,
+      "loss": 5.0039,
       "step": 101
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.94e-05,
-      "loss": 4.4992,
+      "learning_rate": 0.00019800000000000002,
+      "loss": 4.7123,
       "step": 102
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.9999999999999995e-05,
-      "loss": 4.4298,
+      "learning_rate": 0.0002,
+      "loss": 5.1955,
       "step": 103
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.0599999999999996e-05,
-      "loss": 4.3874,
+      "learning_rate": 0.000202,
+      "loss": 4.7571,
       "step": 104
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.12e-05,
-      "loss": 4.4604,
+      "learning_rate": 0.000204,
+      "loss": 4.7477,
       "step": 105
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.18e-05,
-      "loss": 4.412,
+      "learning_rate": 0.000206,
+      "loss": 4.6276,
       "step": 106
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.239999999999999e-05,
-      "loss": 4.4223,
+      "learning_rate": 0.000208,
+      "loss": 4.754,
       "step": 107
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.299999999999999e-05,
-      "loss": 4.542,
+      "learning_rate": 0.00021,
+      "loss": 4.7858,
       "step": 108
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.359999999999999e-05,
-      "loss": 4.465,
+      "learning_rate": 0.000212,
+      "loss": 4.7258,
       "step": 109
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.419999999999999e-05,
-      "loss": 4.5173,
+      "learning_rate": 0.000214,
+      "loss": 4.7978,
       "step": 110
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.479999999999999e-05,
-      "loss": 4.3557,
+      "learning_rate": 0.000216,
+      "loss": 5.0629,
       "step": 111
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.539999999999999e-05,
-      "loss": 4.3036,
+      "learning_rate": 0.000218,
+      "loss": 4.547,
       "step": 112
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.599999999999999e-05,
-      "loss": 4.4658,
+      "learning_rate": 0.00022,
+      "loss": 4.7776,
       "step": 113
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.659999999999999e-05,
-      "loss": 4.2061,
+      "learning_rate": 0.000222,
+      "loss": 4.5737,
       "step": 114
     },
     {
       "epoch": 0.06,
-      "learning_rate": 6.72e-05,
-      "loss": 4.1916,
+      "learning_rate": 0.000224,
+      "loss": 4.4555,
       "step": 115
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.78e-05,
-      "loss": 4.3232,
+      "learning_rate": 0.00022600000000000002,
+      "loss": 4.569,
       "step": 116
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.84e-05,
-      "loss": 4.5664,
+      "learning_rate": 0.000228,
+      "loss": 4.9807,
       "step": 117
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.9e-05,
-      "loss": 4.2962,
+      "learning_rate": 0.00023,
+      "loss": 5.4174,
       "step": 118
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.96e-05,
-      "loss": 4.4515,
+      "learning_rate": 0.00023200000000000003,
+      "loss": 4.7848,
       "step": 119
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.02e-05,
-      "loss": 4.5881,
+      "learning_rate": 0.00023400000000000002,
+      "loss": 5.2558,
       "step": 120
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.079999999999999e-05,
-      "loss": 4.515,
+      "learning_rate": 0.000236,
+      "loss": 5.0696,
       "step": 121
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.139999999999999e-05,
-      "loss": 4.484,
+      "learning_rate": 0.00023799999999999998,
+      "loss": 4.7726,
       "step": 122
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.199999999999999e-05,
-      "loss": 4.4975,
+      "learning_rate": 0.00024,
+      "loss": 4.7591,
       "step": 123
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.259999999999999e-05,
-      "loss": 4.4655,
+      "learning_rate": 0.000242,
+      "loss": 4.9439,
       "step": 124
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.319999999999999e-05,
-      "loss": 4.1367,
+      "learning_rate": 0.000244,
+      "loss": 4.5524,
       "step": 125
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.379999999999999e-05,
-      "loss": 4.6877,
+      "learning_rate": 0.000246,
+      "loss": 4.9518,
       "step": 126
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.439999999999999e-05,
-      "loss": 4.465,
+      "learning_rate": 0.000248,
+      "loss": 4.674,
       "step": 127
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.5e-05,
-      "loss": 4.4351,
+      "learning_rate": 0.00025,
+      "loss": 4.8241,
       "step": 128
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.56e-05,
-      "loss": 4.3375,
+      "learning_rate": 0.000252,
+      "loss": 4.8413,
       "step": 129
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.62e-05,
-      "loss": 4.4541,
+      "learning_rate": 0.000254,
+      "loss": 5.1165,
       "step": 130
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.68e-05,
-      "loss": 4.6463,
+      "learning_rate": 0.000256,
+      "loss": 5.6705,
       "step": 131
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.74e-05,
-      "loss": 4.3778,
+      "learning_rate": 0.00025800000000000004,
+      "loss": 4.6968,
       "step": 132
     },
     {
       "epoch": 0.07,
-      "learning_rate": 7.8e-05,
-      "loss": 4.5082,
+      "learning_rate": 0.00026000000000000003,
+      "loss": 5.4421,
       "step": 133
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.86e-05,
-      "loss": 4.4627,
+      "learning_rate": 0.000262,
+      "loss": 5.273,
       "step": 134
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.92e-05,
-      "loss": 4.2918,
+      "learning_rate": 0.000264,
+      "loss": 4.6382,
       "step": 135
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.98e-05,
-      "loss": 4.6064,
+      "learning_rate": 0.000266,
+      "loss": 5.379,
       "step": 136
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.04e-05,
-      "loss": 4.284,
+      "learning_rate": 0.000268,
+      "loss": 5.0821,
       "step": 137
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.1e-05,
-      "loss": 4.5167,
+      "learning_rate": 0.00027,
+      "loss": 5.8035,
       "step": 138
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.16e-05,
-      "loss": 4.6984,
+      "learning_rate": 0.00027200000000000005,
+      "loss": 5.5749,
       "step": 139
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.22e-05,
-      "loss": 4.3706,
+      "learning_rate": 0.00027400000000000005,
+      "loss": 5.1669,
       "step": 140
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.28e-05,
-      "loss": 4.5401,
+      "learning_rate": 0.00027600000000000004,
+      "loss": 6.1823,
       "step": 141
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.34e-05,
-      "loss": 4.7277,
+      "learning_rate": 0.00027800000000000004,
+      "loss": 5.2155,
       "step": 142
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.4e-05,
-      "loss": 4.4341,
+      "learning_rate": 0.00028000000000000003,
+      "loss": 4.8095,
       "step": 143
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.459999999999998e-05,
-      "loss": 4.7501,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 6.1255,
       "step": 144
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.519999999999998e-05,
-      "loss": 4.7898,
+      "learning_rate": 0.00028399999999999996,
+      "loss": 5.2863,
       "step": 145
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.579999999999998e-05,
-      "loss": 4.7006,
+      "learning_rate": 0.00028599999999999996,
+      "loss": 5.1855,
       "step": 146
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.639999999999999e-05,
-      "loss": 4.7019,
+      "learning_rate": 0.000288,
+      "loss": 5.2812,
       "step": 147
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.699999999999999e-05,
-      "loss": 4.7463,
+      "learning_rate": 0.00029,
+      "loss": 5.1448,
       "step": 148
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.759999999999999e-05,
-      "loss": 4.9289,
+      "learning_rate": 0.000292,
+      "loss": 5.3333,
       "step": 149
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.819999999999999e-05,
-      "loss": 4.2683,
+      "learning_rate": 0.000294,
+      "loss": 4.6151,
       "step": 150
     },
     {
       "epoch": 0.08,
-      "learning_rate": 8.879999999999999e-05,
-      "loss": 4.3449,
+      "learning_rate": 0.000296,
+      "loss": 4.945,
       "step": 151
     },
     {
       "epoch": 0.09,
-      "learning_rate": 8.939999999999999e-05,
-      "loss": 4.7897,
+      "learning_rate": 0.000298,
+      "loss": 5.4402,
       "step": 152
     },
     {
       "epoch": 0.09,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 4.2006,
+      "learning_rate": 0.0003,
+      "loss": 4.599,
       "step": 153
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.059999999999999e-05,
-      "loss": 4.682,
+      "learning_rate": 0.000302,
+      "loss": 6.4004,
       "step": 154
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.12e-05,
-      "loss": 4.4867,
+      "learning_rate": 0.000304,
+      "loss": 4.9015,
       "step": 155
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.18e-05,
-      "loss": 4.8325,
+      "learning_rate": 0.000306,
+      "loss": 5.2098,
       "step": 156
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.24e-05,
-      "loss": 4.609,
+      "learning_rate": 0.000308,
+      "loss": 5.079,
       "step": 157
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.3e-05,
-      "loss": 4.5416,
+      "learning_rate": 0.00031,
+      "loss": 4.8843,
       "step": 158
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.36e-05,
-      "loss": 4.6764,
+      "learning_rate": 0.000312,
+      "loss": 5.4663,
       "step": 159
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.419999999999999e-05,
-      "loss": 4.7579,
+      "learning_rate": 0.000314,
+      "loss": 6.6276,
       "step": 160
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.479999999999999e-05,
-      "loss": 4.6248,
+      "learning_rate": 0.000316,
+      "loss": 5.2692,
       "step": 161
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.539999999999999e-05,
-      "loss": 4.8427,
+      "learning_rate": 0.00031800000000000003,
+      "loss": 5.4687,
       "step": 162
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.599999999999999e-05,
-      "loss": 4.6742,
+      "learning_rate": 0.00032,
+      "loss": 5.0059,
       "step": 163
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.659999999999999e-05,
-      "loss": 4.7829,
+      "learning_rate": 0.000322,
+      "loss": 5.4655,
       "step": 164
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.719999999999999e-05,
-      "loss": 4.5877,
+      "learning_rate": 0.000324,
+      "loss": 4.9151,
       "step": 165
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.779999999999999e-05,
-      "loss": 4.709,
+      "learning_rate": 0.000326,
+      "loss": 5.3121,
       "step": 166
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.839999999999999e-05,
-      "loss": 4.4537,
+      "learning_rate": 0.000328,
+      "loss": 4.9021,
       "step": 167
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.9e-05,
-      "loss": 4.3135,
+      "learning_rate": 0.00033,
+      "loss": 4.7903,
       "step": 168
     },
     {
       "epoch": 0.09,
-      "learning_rate": 9.96e-05,
-      "loss": 4.3335,
+      "learning_rate": 0.00033200000000000005,
+      "loss": 4.7505,
       "step": 169
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.0001002,
-      "loss": 4.4505,
+      "learning_rate": 0.00033400000000000004,
+      "loss": 4.8679,
       "step": 170
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.0001008,
-      "loss": 4.5844,
+      "learning_rate": 0.00033600000000000004,
+      "loss": 5.3312,
       "step": 171
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.0001014,
-      "loss": 4.9214,
+      "learning_rate": 0.00033800000000000003,
+      "loss": 7.5863,
       "step": 172
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000102,
-      "loss": 4.4454,
+      "learning_rate": 0.00034,
+      "loss": 4.9188,
       "step": 173
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.0001026,
-      "loss": 4.5723,
+      "learning_rate": 0.000342,
+      "loss": 5.1428,
       "step": 174
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010319999999999999,
-      "loss": 4.8336,
+      "learning_rate": 0.00034399999999999996,
+      "loss": 6.7474,
       "step": 175
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010379999999999999,
-      "loss": 4.4524,
+      "learning_rate": 0.000346,
+      "loss": 5.2806,
       "step": 176
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010439999999999999,
-      "loss": 4.4489,
+      "learning_rate": 0.000348,
+      "loss": 4.8877,
       "step": 177
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010499999999999999,
-      "loss": 4.415,
+      "learning_rate": 0.00035,
+      "loss": 4.8529,
       "step": 178
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010559999999999998,
-      "loss": 4.2687,
+      "learning_rate": 0.000352,
+      "loss": 4.8546,
       "step": 179
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010619999999999998,
-      "loss": 4.8465,
+      "learning_rate": 0.000354,
+      "loss": 5.6208,
       "step": 180
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010679999999999998,
-      "loss": 4.9842,
+      "learning_rate": 0.000356,
+      "loss": 5.7513,
       "step": 181
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010739999999999998,
-      "loss": 4.4354,
+      "learning_rate": 0.000358,
+      "loss": 4.8119,
       "step": 182
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010799999999999998,
-      "loss": 4.7134,
+      "learning_rate": 0.00035999999999999997,
+      "loss": 5.3898,
       "step": 183
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010859999999999998,
-      "loss": 4.6371,
+      "learning_rate": 0.000362,
+      "loss": 4.9066,
       "step": 184
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010919999999999998,
-      "loss": 4.6847,
+      "learning_rate": 0.000364,
+      "loss": 5.1167,
       "step": 185
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00010979999999999999,
-      "loss": 4.528,
+      "learning_rate": 0.000366,
+      "loss": 5.0465,
       "step": 186
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00011039999999999999,
-      "loss": 4.4762,
+      "learning_rate": 0.000368,
+      "loss": 4.8059,
       "step": 187
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00011099999999999999,
-      "loss": 4.7171,
+      "learning_rate": 0.00037,
+      "loss": 5.1748,
       "step": 188
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00011159999999999999,
-      "loss": 4.7589,
+      "learning_rate": 0.000372,
+      "loss": 5.2375,
       "step": 189
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00011219999999999999,
-      "loss": 4.6562,
+      "learning_rate": 0.000374,
+      "loss": 5.6544,
       "step": 190
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00011279999999999999,
-      "loss": 4.7215,
+      "learning_rate": 0.00037600000000000003,
+      "loss": 5.2274,
       "step": 191
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00011339999999999999,
-      "loss": 4.4537,
+      "learning_rate": 0.000378,
+      "loss": 5.153,
       "step": 192
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00011399999999999999,
-      "loss": 4.6936,
+      "learning_rate": 0.00038,
+      "loss": 6.7931,
       "step": 193
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001146,
-      "loss": 4.4776,
+      "learning_rate": 0.000382,
+      "loss": 5.2208,
       "step": 194
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001152,
-      "loss": 4.9033,
+      "learning_rate": 0.000384,
+      "loss": 5.1046,
       "step": 195
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001158,
-      "loss": 5.0891,
+      "learning_rate": 0.000386,
+      "loss": 6.0011,
       "step": 196
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001164,
-      "loss": 4.7964,
+      "learning_rate": 0.000388,
+      "loss": 6.4511,
       "step": 197
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000117,
-      "loss": 4.5244,
+      "learning_rate": 0.00039000000000000005,
+      "loss": 4.8761,
       "step": 198
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001176,
-      "loss": 4.6896,
+      "learning_rate": 0.00039200000000000004,
+      "loss": 5.1652,
       "step": 199
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001182,
-      "loss": 5.4979,
+      "learning_rate": 0.00039400000000000004,
+      "loss": 6.4481,
       "step": 200
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001188,
-      "loss": 4.7826,
+      "learning_rate": 0.00039600000000000003,
+      "loss": 5.4104,
       "step": 201
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0001194,
-      "loss": 4.5912,
+      "learning_rate": 0.000398,
+      "loss": 5.1596,
       "step": 202
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00011999999999999999,
-      "loss": 4.6793,
+      "learning_rate": 0.0004,
+      "loss": 5.3031,
       "step": 203
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00012059999999999999,
-      "loss": 4.6473,
+      "learning_rate": 0.000402,
+      "loss": 5.6399,
       "step": 204
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00012119999999999999,
-      "loss": 4.5137,
+      "learning_rate": 0.000404,
+      "loss": 5.4172,
       "step": 205
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012179999999999999,
-      "loss": 5.0252,
+      "learning_rate": 0.00040600000000000006,
+      "loss": 6.016,
       "step": 206
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.0001224,
-      "loss": 4.7233,
+      "learning_rate": 0.000408,
+      "loss": 5.3531,
       "step": 207
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012299999999999998,
-      "loss": 5.2148,
+      "learning_rate": 0.00041,
+      "loss": 6.4945,
       "step": 208
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.0001236,
-      "loss": 4.4845,
+      "learning_rate": 0.000412,
+      "loss": 5.2357,
       "step": 209
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012419999999999998,
-      "loss": 4.7004,
+      "learning_rate": 0.000414,
+      "loss": 5.5978,
       "step": 210
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012479999999999997,
-      "loss": 4.5284,
+      "learning_rate": 0.000416,
+      "loss": 5.059,
       "step": 211
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012539999999999999,
-      "loss": 4.6566,
+      "learning_rate": 0.00041799999999999997,
+      "loss": 5.702,
       "step": 212
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012599999999999997,
-      "loss": 4.7671,
+      "learning_rate": 0.00042,
+      "loss": 6.4733,
       "step": 213
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.0001266,
-      "loss": 4.4086,
+      "learning_rate": 0.000422,
+      "loss": 4.8425,
       "step": 214
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012719999999999997,
-      "loss": 5.1062,
+      "learning_rate": 0.000424,
+      "loss": 5.8915,
       "step": 215
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.0001278,
-      "loss": 4.1818,
+      "learning_rate": 0.000426,
+      "loss": 4.8864,
       "step": 216
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012839999999999998,
-      "loss": 4.6396,
+      "learning_rate": 0.000428,
+      "loss": 5.4242,
       "step": 217
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000129,
-      "loss": 4.4736,
+      "learning_rate": 0.00043,
+      "loss": 4.9671,
       "step": 218
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00012959999999999998,
-      "loss": 4.6794,
+      "learning_rate": 0.000432,
+      "loss": 5.0474,
       "step": 219
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.0001302,
-      "loss": 5.0791,
+      "learning_rate": 0.00043400000000000003,
+      "loss": 6.2105,
       "step": 220
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00013079999999999998,
-      "loss": 4.6162,
+      "learning_rate": 0.000436,
+      "loss": 5.2854,
       "step": 221
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.0001314,
-      "loss": 4.7603,
+      "learning_rate": 0.000438,
+      "loss": 5.8823,
       "step": 222
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00013199999999999998,
-      "loss": 4.3964,
+      "learning_rate": 0.00044,
+      "loss": 5.0082,
       "step": 223
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001326,
-      "loss": 4.5764,
+      "learning_rate": 0.000442,
+      "loss": 5.3606,
       "step": 224
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00013319999999999999,
-      "loss": 4.6934,
+      "learning_rate": 0.000444,
+      "loss": 5.3525,
       "step": 225
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001338,
-      "loss": 4.891,
+      "learning_rate": 0.000446,
+      "loss": 5.6586,
       "step": 226
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001344,
-      "loss": 4.7576,
+      "learning_rate": 0.000448,
+      "loss": 5.4038,
       "step": 227
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000135,
-      "loss": 4.4243,
+      "learning_rate": 0.00045000000000000004,
+      "loss": 5.3157,
       "step": 228
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001356,
-      "loss": 4.59,
+      "learning_rate": 0.00045200000000000004,
+      "loss": 5.154,
       "step": 229
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001362,
-      "loss": 4.5525,
+      "learning_rate": 0.00045400000000000003,
+      "loss": 5.1189,
       "step": 230
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001368,
-      "loss": 4.7133,
+      "learning_rate": 0.000456,
+      "loss": 4.9935,
       "step": 231
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001374,
-      "loss": 4.6903,
+      "learning_rate": 0.000458,
+      "loss": 5.7535,
       "step": 232
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000138,
-      "loss": 4.7626,
+      "learning_rate": 0.00046,
+      "loss": 5.4902,
       "step": 233
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001386,
-      "loss": 4.553,
+      "learning_rate": 0.000462,
+      "loss": 5.1133,
       "step": 234
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001392,
-      "loss": 4.4217,
+      "learning_rate": 0.00046400000000000006,
+      "loss": 4.858,
       "step": 235
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00013979999999999998,
-      "loss": 5.1364,
+      "learning_rate": 0.00046600000000000005,
+      "loss": 5.6959,
       "step": 236
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001404,
-      "loss": 4.6538,
+      "learning_rate": 0.00046800000000000005,
+      "loss": 6.2172,
       "step": 237
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00014099999999999998,
-      "loss": 5.2553,
+      "learning_rate": 0.00047,
+      "loss": 6.562,
       "step": 238
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00014159999999999997,
-      "loss": 5.2305,
+      "learning_rate": 0.000472,
+      "loss": 5.935,
       "step": 239
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.0001422,
-      "loss": 4.8764,
+      "learning_rate": 0.000474,
+      "loss": 6.6483,
       "step": 240
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00014279999999999997,
-      "loss": 4.6992,
+      "learning_rate": 0.00047599999999999997,
+      "loss": 5.2033,
       "step": 241
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001434,
-      "loss": 4.8932,
+      "learning_rate": 0.00047799999999999996,
+      "loss": 5.6109,
       "step": 242
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00014399999999999998,
-      "loss": 6.5728,
+      "learning_rate": 0.00048,
+      "loss": 8.1627,
       "step": 243
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001446,
-      "loss": 4.974,
+      "learning_rate": 0.000482,
+      "loss": 5.6307,
       "step": 244
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00014519999999999998,
-      "loss": 4.7734,
+      "learning_rate": 0.000484,
+      "loss": 5.3183,
       "step": 245
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001458,
-      "loss": 4.748,
+      "learning_rate": 0.000486,
+      "loss": 5.3182,
       "step": 246
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00014639999999999998,
-      "loss": 4.6014,
+      "learning_rate": 0.000488,
+      "loss": 4.9353,
       "step": 247
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000147,
-      "loss": 4.7732,
+      "learning_rate": 0.00049,
+      "loss": 5.4198,
       "step": 248
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00014759999999999998,
-      "loss": 4.6782,
+      "learning_rate": 0.000492,
+      "loss": 5.2011,
       "step": 249
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001482,
-      "loss": 5.2136,
+      "learning_rate": 0.000494,
+      "loss": 6.221,
       "step": 250
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00014879999999999998,
-      "loss": 4.6211,
+      "learning_rate": 0.000496,
+      "loss": 5.7471,
       "step": 251
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001494,
-      "loss": 5.1229,
+      "learning_rate": 0.000498,
+      "loss": 6.7848,
       "step": 252
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00015,
-      "loss": 5.4691,
+      "learning_rate": 0.0005,
+      "loss": 6.1404,
       "step": 253
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00015059999999999997,
-      "loss": 4.8309,
+      "learning_rate": 0.0005020000000000001,
+      "loss": 5.6192,
       "step": 254
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001512,
-      "loss": 4.7792,
+      "learning_rate": 0.000504,
+      "loss": 5.2452,
       "step": 255
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00015179999999999998,
-      "loss": 4.6088,
+      "learning_rate": 0.000506,
+      "loss": 5.4712,
       "step": 256
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0001524,
-      "loss": 4.6558,
+      "learning_rate": 0.000508,
+      "loss": 5.3885,
       "step": 257
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00015299999999999998,
-      "loss": 5.2097,
+      "learning_rate": 0.00051,
+      "loss": 6.309,
       "step": 258
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001536,
-      "loss": 4.8985,
+      "learning_rate": 0.000512,
+      "loss": 5.5993,
       "step": 259
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.00015419999999999998,
-      "loss": 4.4116,
+      "learning_rate": 0.000514,
+      "loss": 5.0639,
       "step": 260
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001548,
-      "loss": 5.2036,
+      "learning_rate": 0.0005160000000000001,
+      "loss": 6.3613,
       "step": 261
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.00015539999999999998,
-      "loss": 5.0657,
+      "learning_rate": 0.000518,
+      "loss": 5.6503,
       "step": 262
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000156,
-      "loss": 5.2332,
+      "learning_rate": 0.0005200000000000001,
+      "loss": 6.2474,
       "step": 263
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.00015659999999999998,
-      "loss": 4.6791,
+      "learning_rate": 0.000522,
+      "loss": 5.4449,
       "step": 264
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001572,
-      "loss": 4.876,
+      "learning_rate": 0.000524,
+      "loss": 5.4925,
       "step": 265
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001578,
-      "loss": 4.3383,
+      "learning_rate": 0.000526,
+      "loss": 4.9186,
       "step": 266
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001584,
-      "loss": 4.6682,
+      "learning_rate": 0.000528,
+      "loss": 5.8049,
       "step": 267
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000159,
-      "loss": 4.8132,
+      "learning_rate": 0.0005300000000000001,
+      "loss": 5.524,
       "step": 268
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001596,
-      "loss": 4.7957,
+      "learning_rate": 0.000532,
+      "loss": 5.6952,
       "step": 269
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001602,
-      "loss": 4.3345,
+      "learning_rate": 0.0005340000000000001,
+      "loss": 5.0467,
       "step": 270
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001608,
-      "loss": 4.616,
+      "learning_rate": 0.000536,
+      "loss": 5.2591,
       "step": 271
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001614,
-      "loss": 4.5979,
+      "learning_rate": 0.0005380000000000001,
+      "loss": 5.2361,
       "step": 272
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000162,
-      "loss": 4.9527,
+      "learning_rate": 0.00054,
+      "loss": 5.6284,
       "step": 273
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001626,
-      "loss": 6.8094,
+      "learning_rate": 0.0005420000000000001,
+      "loss": 7.7446,
       "step": 274
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001632,
-      "loss": 4.7034,
+      "learning_rate": 0.0005440000000000001,
+      "loss": 5.3482,
       "step": 275
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0001638,
-      "loss": 5.99,
+      "learning_rate": 0.000546,
+      "loss": 6.5872,
       "step": 276
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0001644,
-      "loss": 4.5877,
+      "learning_rate": 0.0005480000000000001,
+      "loss": 6.1554,
       "step": 277
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000165,
-      "loss": 4.8687,
+      "learning_rate": 0.00055,
+      "loss": 5.5187,
       "step": 278
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0001656,
-      "loss": 4.4936,
+      "learning_rate": 0.0005520000000000001,
+      "loss": 5.2059,
       "step": 279
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0001662,
-      "loss": 4.5845,
+      "learning_rate": 0.000554,
+      "loss": 5.1271,
       "step": 280
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0001668,
-      "loss": 5.168,
+      "learning_rate": 0.0005560000000000001,
+      "loss": 6.0853,
       "step": 281
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0001674,
-      "loss": 4.7432,
+      "learning_rate": 0.000558,
+      "loss": 5.5123,
       "step": 282
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000168,
-      "loss": 5.1332,
+      "learning_rate": 0.0005600000000000001,
+      "loss": 5.9598,
       "step": 283
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0001686,
-      "loss": 4.547,
+      "learning_rate": 0.0005620000000000001,
+      "loss": 4.9705,
       "step": 284
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00016919999999999997,
-      "loss": 5.0918,
+      "learning_rate": 0.0005639999999999999,
+      "loss": 5.5854,
       "step": 285
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00016979999999999998,
-      "loss": 4.878,
+      "learning_rate": 0.000566,
+      "loss": 5.4473,
       "step": 286
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017039999999999997,
-      "loss": 4.869,
+      "learning_rate": 0.0005679999999999999,
+      "loss": 5.4138,
       "step": 287
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017099999999999998,
-      "loss": 4.9144,
+      "learning_rate": 0.00057,
+      "loss": 5.807,
       "step": 288
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017159999999999997,
-      "loss": 4.6275,
+      "learning_rate": 0.0005719999999999999,
+      "loss": 5.2322,
       "step": 289
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017219999999999998,
-      "loss": 5.413,
+      "learning_rate": 0.000574,
+      "loss": 6.1189,
       "step": 290
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017279999999999997,
-      "loss": 4.659,
+      "learning_rate": 0.000576,
+      "loss": 5.35,
       "step": 291
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017339999999999996,
-      "loss": 5.6418,
+      "learning_rate": 0.000578,
+      "loss": 6.651,
       "step": 292
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017399999999999997,
-      "loss": 5.054,
+      "learning_rate": 0.00058,
+      "loss": 5.4333,
       "step": 293
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00017459999999999996,
-      "loss": 4.8574,
+      "learning_rate": 0.0005819999999999999,
+      "loss": 5.5214,
       "step": 294
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017519999999999998,
-      "loss": 4.7022,
+      "learning_rate": 0.000584,
+      "loss": 5.2903,
       "step": 295
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017579999999999996,
-      "loss": 6.8064,
+      "learning_rate": 0.0005859999999999999,
+      "loss": 11.0393,
       "step": 296
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017639999999999998,
-      "loss": 5.4316,
+      "learning_rate": 0.000588,
+      "loss": 6.1775,
       "step": 297
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017699999999999997,
-      "loss": 4.9651,
+      "learning_rate": 0.00059,
+      "loss": 5.7617,
       "step": 298
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017759999999999998,
-      "loss": 4.6225,
+      "learning_rate": 0.000592,
+      "loss": 6.5483,
       "step": 299
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017819999999999997,
-      "loss": 4.6815,
+      "learning_rate": 0.000594,
+      "loss": 5.1852,
       "step": 300
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017879999999999998,
-      "loss": 5.0166,
+      "learning_rate": 0.000596,
+      "loss": 6.1581,
       "step": 301
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017939999999999997,
-      "loss": 4.8534,
+      "learning_rate": 0.000598,
+      "loss": 5.3975,
       "step": 302
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00017999999999999998,
-      "loss": 4.8499,
+      "learning_rate": 0.0006,
+      "loss": 5.429,
       "step": 303
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00018059999999999997,
-      "loss": 4.6247,
+      "learning_rate": 0.000602,
+      "loss": 5.1957,
       "step": 304
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00018119999999999999,
-      "loss": 5.4012,
+      "learning_rate": 0.000604,
+      "loss": 6.3192,
       "step": 305
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00018179999999999997,
-      "loss": 4.8328,
+      "learning_rate": 0.000606,
+      "loss": 5.3676,
       "step": 306
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.0001824,
-      "loss": 4.77,
+      "learning_rate": 0.000608,
+      "loss": 5.339,
       "step": 307
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00018299999999999998,
-      "loss": 6.1595,
+      "learning_rate": 0.00061,
+      "loss": 6.7886,
       "step": 308
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.0001836,
-      "loss": 4.9547,
+      "learning_rate": 0.000612,
+      "loss": 5.5512,
       "step": 309
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00018419999999999998,
-      "loss": 5.2612,
+      "learning_rate": 0.000614,
+      "loss": 5.9811,
       "step": 310
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.0001848,
-      "loss": 4.7641,
+      "learning_rate": 0.000616,
+      "loss": 5.3142,
       "step": 311
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00018539999999999998,
-      "loss": 4.8442,
+      "learning_rate": 0.0006180000000000001,
+      "loss": 5.8051,
       "step": 312
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000186,
-      "loss": 4.9182,
+      "learning_rate": 0.00062,
+      "loss": 5.3857,
       "step": 313
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00018659999999999998,
-      "loss": 5.0422,
+      "learning_rate": 0.000622,
+      "loss": 6.0494,
       "step": 314
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0001872,
-      "loss": 4.7754,
+      "learning_rate": 0.000624,
+      "loss": 6.1703,
       "step": 315
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00018779999999999998,
-      "loss": 4.6309,
+      "learning_rate": 0.000626,
+      "loss": 5.26,
       "step": 316
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00018839999999999997,
-      "loss": 4.7187,
+      "learning_rate": 0.000628,
+      "loss": 5.3323,
       "step": 317
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00018899999999999999,
-      "loss": 4.3824,
+      "learning_rate": 0.00063,
+      "loss": 5.1775,
       "step": 318
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00018959999999999997,
-      "loss": 4.9719,
+      "learning_rate": 0.000632,
+      "loss": 6.4257,
       "step": 319
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0001902,
-      "loss": 5.3327,
+      "learning_rate": 0.000634,
+      "loss": 6.4121,
       "step": 320
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00019079999999999998,
-      "loss": 5.2786,
+      "learning_rate": 0.0006360000000000001,
+      "loss": 6.3211,
       "step": 321
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0001914,
-      "loss": 5.1248,
+      "learning_rate": 0.000638,
+      "loss": 6.3609,
       "step": 322
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00019199999999999998,
-      "loss": 5.1787,
+      "learning_rate": 0.00064,
+      "loss": 5.843,
       "step": 323
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0001926,
-      "loss": 4.8597,
+      "learning_rate": 0.000642,
+      "loss": 6.328,
       "step": 324
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00019319999999999998,
-      "loss": 5.3247,
+      "learning_rate": 0.000644,
+      "loss": 5.4115,
       "step": 325
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0001938,
-      "loss": 5.0439,
+      "learning_rate": 0.000646,
+      "loss": 5.5552,
       "step": 326
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00019439999999999998,
-      "loss": 5.2188,
+      "learning_rate": 0.000648,
+      "loss": 5.9054,
       "step": 327
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000195,
-      "loss": 5.4713,
+      "learning_rate": 0.0006500000000000001,
+      "loss": 5.8717,
       "step": 328
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00019559999999999998,
-      "loss": 4.7846,
+      "learning_rate": 0.000652,
+      "loss": 5.2736,
       "step": 329
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0001962,
-      "loss": 4.9201,
+      "learning_rate": 0.0006540000000000001,
+      "loss": 5.742,
       "step": 330
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.00019679999999999999,
-      "loss": 4.7976,
+      "learning_rate": 0.000656,
+      "loss": 5.6059,
       "step": 331
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0001974,
-      "loss": 5.1048,
+      "learning_rate": 0.0006580000000000001,
+      "loss": 5.8953,
       "step": 332
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000198,
-      "loss": 5.7319,
+      "learning_rate": 0.00066,
+      "loss": 6.9948,
       "step": 333
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0001986,
-      "loss": 5.4133,
+      "learning_rate": 0.000662,
+      "loss": 5.9059,
       "step": 334
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0001992,
-      "loss": 5.0639,
+      "learning_rate": 0.0006640000000000001,
+      "loss": 6.7979,
       "step": 335
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0001998,
-      "loss": 4.9932,
+      "learning_rate": 0.000666,
+      "loss": 6.4717,
       "step": 336
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0002004,
-      "loss": 4.6783,
+      "learning_rate": 0.0006680000000000001,
+      "loss": 6.2319,
       "step": 337
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000201,
-      "loss": 4.563,
+      "learning_rate": 0.00067,
+      "loss": 5.3135,
       "step": 338
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0002016,
-      "loss": 4.7717,
+      "learning_rate": 0.0006720000000000001,
+      "loss": 5.2328,
       "step": 339
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0002022,
-      "loss": 4.9817,
+      "learning_rate": 0.000674,
+      "loss": 5.8928,
       "step": 340
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0002028,
-      "loss": 4.6539,
+      "learning_rate": 0.0006760000000000001,
+      "loss": 5.3474,
       "step": 341
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.00020339999999999998,
-      "loss": 4.7897,
+      "learning_rate": 0.0006780000000000001,
+      "loss": 5.6893,
       "step": 342
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000204,
-      "loss": 4.948,
+      "learning_rate": 0.00068,
+      "loss": 5.5719,
       "step": 343
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.00020459999999999999,
-      "loss": 4.4071,
+      "learning_rate": 0.0006820000000000001,
+      "loss": 5.1262,
       "step": 344
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0002052,
-      "loss": 5.4885,
+      "learning_rate": 0.000684,
+      "loss": 6.4271,
       "step": 345
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0002058,
-      "loss": 4.5249,
+      "learning_rate": 0.0006860000000000001,
+      "loss": 5.284,
       "step": 346
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.00020639999999999998,
-      "loss": 5.4014,
+      "learning_rate": 0.0006879999999999999,
+      "loss": 5.8936,
       "step": 347
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00020699999999999996,
-      "loss": 4.989,
+      "learning_rate": 0.00069,
+      "loss": 5.7215,
       "step": 348
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00020759999999999998,
-      "loss": 4.705,
+      "learning_rate": 0.000692,
+      "loss": 7.5502,
       "step": 349
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00020819999999999996,
-      "loss": 4.8833,
+      "learning_rate": 0.000694,
+      "loss": 5.2064,
       "step": 350
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00020879999999999998,
-      "loss": 5.2783,
+      "learning_rate": 0.000696,
+      "loss": 6.4653,
       "step": 351
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00020939999999999997,
-      "loss": 4.9268,
+      "learning_rate": 0.0006979999999999999,
+      "loss": 5.8233,
       "step": 352
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00020999999999999998,
-      "loss": 4.8817,
+      "learning_rate": 0.0007,
+      "loss": 5.6414,
       "step": 353
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021059999999999997,
-      "loss": 4.6966,
+      "learning_rate": 0.0007019999999999999,
+      "loss": 5.335,
       "step": 354
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021119999999999996,
-      "loss": 4.4055,
+      "learning_rate": 0.000704,
+      "loss": 5.2106,
       "step": 355
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021179999999999997,
-      "loss": 4.8156,
+      "learning_rate": 0.0007059999999999999,
+      "loss": 5.675,
       "step": 356
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021239999999999996,
-      "loss": 5.1126,
+      "learning_rate": 0.000708,
+      "loss": 5.7361,
       "step": 357
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021299999999999997,
-      "loss": 4.9441,
+      "learning_rate": 0.00071,
+      "loss": 6.1298,
       "step": 358
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021359999999999996,
-      "loss": 5.816,
+      "learning_rate": 0.000712,
+      "loss": 6.9985,
       "step": 359
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021419999999999998,
-      "loss": 4.7336,
+      "learning_rate": 0.000714,
+      "loss": 5.4424,
       "step": 360
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021479999999999996,
-      "loss": 4.7377,
+      "learning_rate": 0.000716,
+      "loss": 5.3975,
       "step": 361
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021539999999999998,
-      "loss": 4.4326,
+      "learning_rate": 0.000718,
+      "loss": 5.057,
       "step": 362
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021599999999999996,
-      "loss": 4.7565,
+      "learning_rate": 0.0007199999999999999,
+      "loss": 5.4258,
       "step": 363
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021659999999999998,
-      "loss": 5.5854,
+      "learning_rate": 0.000722,
+      "loss": 6.0874,
       "step": 364
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00021719999999999997,
-      "loss": 5.3786,
+      "learning_rate": 0.000724,
+      "loss": 8.2428,
       "step": 365
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00021779999999999998,
-      "loss": 5.3139,
+      "learning_rate": 0.000726,
+      "loss": 7.0248,
       "step": 366
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00021839999999999997,
-      "loss": 5.5806,
+      "learning_rate": 0.000728,
+      "loss": 5.8983,
       "step": 367
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00021899999999999998,
-      "loss": 5.5451,
+      "learning_rate": 0.00073,
+      "loss": 5.5039,
       "step": 368
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00021959999999999997,
-      "loss": 4.9983,
+      "learning_rate": 0.000732,
+      "loss": 6.0179,
       "step": 369
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022019999999999999,
-      "loss": 5.1341,
+      "learning_rate": 0.000734,
+      "loss": 5.5334,
       "step": 370
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022079999999999997,
-      "loss": 4.7078,
+      "learning_rate": 0.000736,
+      "loss": 5.3457,
       "step": 371
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.0002214,
-      "loss": 5.6713,
+      "learning_rate": 0.000738,
+      "loss": 6.2481,
       "step": 372
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022199999999999998,
-      "loss": 4.6925,
+      "learning_rate": 0.00074,
+      "loss": 5.4737,
       "step": 373
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.0002226,
-      "loss": 5.3345,
+      "learning_rate": 0.000742,
+      "loss": 6.2183,
       "step": 374
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022319999999999998,
-      "loss": 4.4707,
+      "learning_rate": 0.000744,
+      "loss": 5.1563,
       "step": 375
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.0002238,
-      "loss": 4.9737,
+      "learning_rate": 0.000746,
+      "loss": 5.6365,
       "step": 376
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022439999999999998,
-      "loss": 4.7309,
+      "learning_rate": 0.000748,
+      "loss": 5.403,
       "step": 377
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000225,
-      "loss": 4.6912,
+      "learning_rate": 0.00075,
+      "loss": 5.3533,
       "step": 378
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022559999999999998,
-      "loss": 4.6685,
+      "learning_rate": 0.0007520000000000001,
+      "loss": 5.3927,
       "step": 379
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022619999999999997,
-      "loss": 5.6542,
+      "learning_rate": 0.000754,
+      "loss": 5.939,
       "step": 380
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022679999999999998,
-      "loss": 4.3565,
+      "learning_rate": 0.000756,
+      "loss": 5.0415,
       "step": 381
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022739999999999997,
-      "loss": 4.917,
+      "learning_rate": 0.000758,
+      "loss": 5.493,
       "step": 382
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00022799999999999999,
-      "loss": 4.7542,
+      "learning_rate": 0.00076,
+      "loss": 5.7137,
       "step": 383
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00022859999999999997,
-      "loss": 5.1225,
+      "learning_rate": 0.000762,
+      "loss": 6.2573,
       "step": 384
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002292,
-      "loss": 5.0217,
+      "learning_rate": 0.000764,
+      "loss": 5.7064,
       "step": 385
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00022979999999999997,
-      "loss": 4.802,
+      "learning_rate": 0.0007660000000000001,
+      "loss": 5.6286,
       "step": 386
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002304,
-      "loss": 6.2126,
+      "learning_rate": 0.000768,
+      "loss": 6.3579,
       "step": 387
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00023099999999999998,
-      "loss": 4.4239,
+      "learning_rate": 0.0007700000000000001,
+      "loss": 5.2865,
       "step": 388
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002316,
-      "loss": 5.1384,
+      "learning_rate": 0.000772,
+      "loss": 5.2695,
       "step": 389
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00023219999999999998,
-      "loss": 6.373,
+      "learning_rate": 0.0007740000000000001,
+      "loss": 6.6205,
       "step": 390
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002328,
-      "loss": 4.9988,
+      "learning_rate": 0.000776,
+      "loss": 5.2116,
       "step": 391
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00023339999999999998,
-      "loss": 5.6505,
+      "learning_rate": 0.000778,
+      "loss": 6.2093,
       "step": 392
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000234,
-      "loss": 5.0591,
+      "learning_rate": 0.0007800000000000001,
+      "loss": 5.5573,
       "step": 393
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00023459999999999998,
-      "loss": 5.0622,
+      "learning_rate": 0.000782,
+      "loss": 5.7461,
       "step": 394
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002352,
-      "loss": 5.0191,
+      "learning_rate": 0.0007840000000000001,
+      "loss": 5.8815,
       "step": 395
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00023579999999999999,
-      "loss": 5.8461,
+      "learning_rate": 0.000786,
+      "loss": 7.0836,
       "step": 396
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002364,
-      "loss": 5.4494,
+      "learning_rate": 0.0007880000000000001,
+      "loss": 5.584,
       "step": 397
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000237,
-      "loss": 5.7292,
+      "learning_rate": 0.00079,
+      "loss": 9.7145,
       "step": 398
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002376,
-      "loss": 7.3261,
+      "learning_rate": 0.0007920000000000001,
+      "loss": 11.0987,
       "step": 399
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002382,
-      "loss": 5.5844,
+      "learning_rate": 0.0007940000000000001,
+      "loss": 5.9893,
       "step": 400
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0002388,
-      "loss": 5.0367,
+      "learning_rate": 0.000796,
+      "loss": 5.8344,
       "step": 401
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0002394,
-      "loss": 5.2496,
+      "learning_rate": 0.0007980000000000001,
+      "loss": 5.8467,
       "step": 402
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00023999999999999998,
-      "loss": 4.9508,
+      "learning_rate": 0.0008,
+      "loss": 5.403,
       "step": 403
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0002406,
-      "loss": 5.0579,
+      "learning_rate": 0.0008020000000000001,
+      "loss": 5.8787,
       "step": 404
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024119999999999998,
-      "loss": 4.8829,
+      "learning_rate": 0.000804,
+      "loss": 5.47,
       "step": 405
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0002418,
-      "loss": 6.0058,
+      "learning_rate": 0.0008060000000000001,
+      "loss": 6.4202,
       "step": 406
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024239999999999998,
-      "loss": 4.8735,
+      "learning_rate": 0.000808,
+      "loss": 5.7586,
       "step": 407
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000243,
-      "loss": 4.7284,
+      "learning_rate": 0.0008100000000000001,
+      "loss": 5.5644,
       "step": 408
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024359999999999999,
-      "loss": 5.8016,
+      "learning_rate": 0.0008120000000000001,
+      "loss": 5.9939,
       "step": 409
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024419999999999997,
-      "loss": 5.0079,
+      "learning_rate": 0.0008139999999999999,
+      "loss": 5.7594,
       "step": 410
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0002448,
-      "loss": 4.9166,
+      "learning_rate": 0.000816,
+      "loss": 6.3656,
       "step": 411
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024539999999999995,
-      "loss": 5.2124,
+      "learning_rate": 0.0008179999999999999,
+      "loss": 5.9035,
       "step": 412
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024599999999999996,
-      "loss": 5.1572,
+      "learning_rate": 0.00082,
+      "loss": 6.1055,
       "step": 413
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0002466,
-      "loss": 5.2936,
+      "learning_rate": 0.0008219999999999999,
+      "loss": 5.7212,
       "step": 414
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0002472,
-      "loss": 5.2027,
+      "learning_rate": 0.000824,
+      "loss": 6.1417,
       "step": 415
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024779999999999995,
-      "loss": 4.8693,
+      "learning_rate": 0.000826,
+      "loss": 5.8061,
       "step": 416
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024839999999999997,
-      "loss": 5.7226,
+      "learning_rate": 0.000828,
+      "loss": 6.2892,
       "step": 417
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000249,
-      "loss": 5.4737,
+      "learning_rate": 0.00083,
+      "loss": 5.7129,
       "step": 418
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00024959999999999994,
-      "loss": 5.0095,
+      "learning_rate": 0.000832,
+      "loss": 6.6271,
       "step": 419
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025019999999999996,
-      "loss": 4.9612,
+      "learning_rate": 0.000834,
+      "loss": 5.8522,
       "step": 420
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025079999999999997,
-      "loss": 4.7594,
+      "learning_rate": 0.0008359999999999999,
+      "loss": 5.6503,
       "step": 421
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002514,
-      "loss": 4.9897,
+      "learning_rate": 0.000838,
+      "loss": 5.6825,
       "step": 422
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025199999999999995,
-      "loss": 4.9528,
+      "learning_rate": 0.00084,
+      "loss": 5.4941,
       "step": 423
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025259999999999996,
-      "loss": 5.5202,
+      "learning_rate": 0.000842,
+      "loss": 5.8587,
       "step": 424
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002532,
-      "loss": 4.9447,
+      "learning_rate": 0.000844,
+      "loss": 5.6034,
       "step": 425
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002538,
-      "loss": 4.8729,
+      "learning_rate": 0.000846,
+      "loss": 5.5805,
       "step": 426
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025439999999999995,
-      "loss": 5.8871,
+      "learning_rate": 0.000848,
+      "loss": 6.4935,
       "step": 427
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025499999999999996,
-      "loss": 6.8518,
+      "learning_rate": 0.00085,
+      "loss": 8.7125,
       "step": 428
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002556,
-      "loss": 4.737,
+      "learning_rate": 0.000852,
+      "loss": 5.5379,
       "step": 429
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002562,
-      "loss": 4.9193,
+      "learning_rate": 0.000854,
+      "loss": 6.774,
       "step": 430
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025679999999999995,
-      "loss": 5.2678,
+      "learning_rate": 0.000856,
+      "loss": 6.1877,
       "step": 431
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025739999999999997,
-      "loss": 4.9773,
+      "learning_rate": 0.000858,
+      "loss": 5.7596,
       "step": 432
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000258,
-      "loss": 4.8193,
+      "learning_rate": 0.00086,
+      "loss": 5.5451,
       "step": 433
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002586,
-      "loss": 4.9082,
+      "learning_rate": 0.000862,
+      "loss": 5.5449,
       "step": 434
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025919999999999996,
-      "loss": 4.9528,
+      "learning_rate": 0.000864,
+      "loss": 5.557,
       "step": 435
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00025979999999999997,
-      "loss": 4.657,
+      "learning_rate": 0.000866,
+      "loss": 5.2794,
       "step": 436
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0002604,
-      "loss": 4.8676,
+      "learning_rate": 0.0008680000000000001,
+      "loss": 5.6317,
       "step": 437
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000261,
-      "loss": 4.9797,
+      "learning_rate": 0.00087,
+      "loss": 6.0562,
       "step": 438
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00026159999999999996,
-      "loss": 4.6369,
+      "learning_rate": 0.000872,
+      "loss": 5.3547,
       "step": 439
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002622,
-      "loss": 4.8951,
+      "learning_rate": 0.000874,
+      "loss": 5.578,
       "step": 440
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002628,
-      "loss": 5.3579,
+      "learning_rate": 0.000876,
+      "loss": 5.8526,
       "step": 441
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00026339999999999995,
-      "loss": 5.6086,
+      "learning_rate": 0.000878,
+      "loss": 6.8698,
       "step": 442
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00026399999999999997,
-      "loss": 4.7691,
+      "learning_rate": 0.00088,
+      "loss": 5.4054,
       "step": 443
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002646,
-      "loss": 6.3931,
+      "learning_rate": 0.000882,
+      "loss": 5.7684,
       "step": 444
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002652,
-      "loss": 4.9209,
+      "learning_rate": 0.000884,
+      "loss": 5.5407,
       "step": 445
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00026579999999999996,
-      "loss": 7.5351,
+      "learning_rate": 0.0008860000000000001,
+      "loss": 7.7708,
       "step": 446
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00026639999999999997,
-      "loss": 6.881,
+      "learning_rate": 0.000888,
+      "loss": 6.5105,
       "step": 447
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000267,
-      "loss": 5.6229,
+      "learning_rate": 0.0008900000000000001,
+      "loss": 6.0654,
       "step": 448
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002676,
-      "loss": 6.2351,
+      "learning_rate": 0.000892,
+      "loss": 10.2675,
       "step": 449
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00026819999999999996,
-      "loss": 4.9889,
+      "learning_rate": 0.000894,
+      "loss": 6.4337,
       "step": 450
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002688,
-      "loss": 4.7304,
+      "learning_rate": 0.000896,
+      "loss": 5.6065,
       "step": 451
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0002694,
-      "loss": 5.0795,
+      "learning_rate": 0.000898,
+      "loss": 6.0042,
       "step": 452
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00027,
-      "loss": 5.574,
+      "learning_rate": 0.0009000000000000001,
+      "loss": 7.2876,
       "step": 453
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00027059999999999996,
-      "loss": 4.7855,
+      "learning_rate": 0.000902,
+      "loss": 5.697,
       "step": 454
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002712,
-      "loss": 4.81,
+      "learning_rate": 0.0009040000000000001,
+      "loss": 6.536,
       "step": 455
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002718,
-      "loss": 4.8733,
+      "learning_rate": 0.000906,
+      "loss": 5.861,
       "step": 456
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002724,
-      "loss": 5.1732,
+      "learning_rate": 0.0009080000000000001,
+      "loss": 5.9541,
       "step": 457
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00027299999999999997,
-      "loss": 5.3624,
+      "learning_rate": 0.00091,
+      "loss": 6.7376,
       "step": 458
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002736,
-      "loss": 4.603,
+      "learning_rate": 0.000912,
+      "loss": 5.5201,
       "step": 459
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002742,
-      "loss": 5.301,
+      "learning_rate": 0.0009140000000000001,
+      "loss": 6.2328,
       "step": 460
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002748,
-      "loss": 5.6724,
+      "learning_rate": 0.000916,
+      "loss": 6.746,
       "step": 461
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00027539999999999997,
-      "loss": 5.1587,
+      "learning_rate": 0.0009180000000000001,
+      "loss": 5.5854,
       "step": 462
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.000276,
-      "loss": 5.4181,
+      "learning_rate": 0.00092,
+      "loss": 6.1128,
       "step": 463
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002766,
-      "loss": 4.9622,
+      "learning_rate": 0.0009220000000000001,
+      "loss": 5.6173,
       "step": 464
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002772,
-      "loss": 4.6857,
+      "learning_rate": 0.000924,
+      "loss": 5.556,
       "step": 465
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002778,
-      "loss": 5.3064,
+      "learning_rate": 0.0009260000000000001,
+      "loss": 6.0281,
       "step": 466
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002784,
-      "loss": 5.0287,
+      "learning_rate": 0.0009280000000000001,
+      "loss": 5.6352,
       "step": 467
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.000279,
-      "loss": 5.1764,
+      "learning_rate": 0.00093,
+      "loss": 6.1857,
       "step": 468
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00027959999999999997,
-      "loss": 5.0388,
+      "learning_rate": 0.0009320000000000001,
+      "loss": 5.3981,
       "step": 469
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002802,
-      "loss": 5.0773,
+      "learning_rate": 0.000934,
+      "loss": 5.7451,
       "step": 470
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0002808,
-      "loss": 5.1667,
+      "learning_rate": 0.0009360000000000001,
+      "loss": 6.0019,
       "step": 471
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00028139999999999996,
-      "loss": 4.8111,
+      "learning_rate": 0.0009379999999999999,
+      "loss": 5.4895,
       "step": 472
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 4.8168,
+      "learning_rate": 0.00094,
+      "loss": 5.5495,
       "step": 473
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0002826,
-      "loss": 5.8241,
+      "learning_rate": 0.000942,
+      "loss": 6.0448,
       "step": 474
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028319999999999994,
-      "loss": 4.9812,
+      "learning_rate": 0.000944,
+      "loss": 5.8212,
       "step": 475
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028379999999999996,
-      "loss": 5.3898,
+      "learning_rate": 0.000946,
+      "loss": 6.0345,
       "step": 476
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0002844,
-      "loss": 5.5009,
+      "learning_rate": 0.000948,
+      "loss": 5.9918,
       "step": 477
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000285,
-      "loss": 4.773,
+      "learning_rate": 0.00095,
+      "loss": 5.8179,
       "step": 478
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028559999999999995,
-      "loss": 4.9837,
+      "learning_rate": 0.0009519999999999999,
+      "loss": 5.8431,
       "step": 479
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028619999999999996,
-      "loss": 5.8951,
+      "learning_rate": 0.000954,
+      "loss": 7.2752,
       "step": 480
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0002868,
-      "loss": 4.9592,
+      "learning_rate": 0.0009559999999999999,
+      "loss": 5.6978,
       "step": 481
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028739999999999994,
-      "loss": 5.6106,
+      "learning_rate": 0.000958,
+      "loss": 7.397,
       "step": 482
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028799999999999995,
-      "loss": 5.3505,
+      "learning_rate": 0.00096,
+      "loss": 6.5424,
       "step": 483
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028859999999999997,
-      "loss": 5.2123,
+      "learning_rate": 0.000962,
+      "loss": 5.8226,
       "step": 484
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0002892,
-      "loss": 5.226,
+      "learning_rate": 0.000964,
+      "loss": 6.5297,
       "step": 485
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00028979999999999994,
-      "loss": 5.9357,
+      "learning_rate": 0.000966,
+      "loss": 6.5972,
       "step": 486
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00029039999999999996,
-      "loss": 5.048,
+      "learning_rate": 0.000968,
+      "loss": 5.6045,
       "step": 487
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00029099999999999997,
-      "loss": 5.245,
+      "learning_rate": 0.0009699999999999999,
+      "loss": 6.1922,
       "step": 488
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0002916,
-      "loss": 5.6778,
+      "learning_rate": 0.000972,
+      "loss": 6.0956,
       "step": 489
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00029219999999999995,
-      "loss": 5.0631,
+      "learning_rate": 0.000974,
+      "loss": 5.7588,
       "step": 490
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029279999999999996,
-      "loss": 5.2207,
+      "learning_rate": 0.000976,
+      "loss": 5.5526,
       "step": 491
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002934,
-      "loss": 5.7228,
+      "learning_rate": 0.000978,
+      "loss": 6.2006,
       "step": 492
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000294,
-      "loss": 5.0073,
+      "learning_rate": 0.00098,
+      "loss": 5.8593,
       "step": 493
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029459999999999995,
-      "loss": 6.3613,
+      "learning_rate": 0.000982,
+      "loss": 6.4826,
       "step": 494
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029519999999999997,
-      "loss": 5.6793,
+      "learning_rate": 0.000984,
+      "loss": 6.0277,
       "step": 495
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002958,
-      "loss": 5.0977,
+      "learning_rate": 0.0009860000000000001,
+      "loss": 6.0805,
       "step": 496
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002964,
-      "loss": 4.8535,
+      "learning_rate": 0.000988,
+      "loss": 5.9525,
       "step": 497
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029699999999999996,
-      "loss": 5.6654,
+      "learning_rate": 0.00099,
+      "loss": 7.538,
       "step": 498
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029759999999999997,
-      "loss": 5.2398,
+      "learning_rate": 0.000992,
+      "loss": 5.6965,
       "step": 499
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002982,
-      "loss": 5.1263,
+      "learning_rate": 0.000994,
+      "loss": 5.4068,
       "step": 500
     },
     {
       "epoch": 0.28,
-      "eval_loss": 5.209839344024658,
-      "eval_runtime": 1248.7575,
-      "eval_samples_per_second": 2.116,
+      "eval_loss": 6.093479156494141,
+      "eval_runtime": 1249.6964,
+      "eval_samples_per_second": 2.114,
       "eval_steps_per_second": 0.265,
-      "eval_wer": 1.955672352241174,
+      "eval_wer": 1.9428203094010312,
       "step": 500
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002988,
-      "loss": 5.853,
+      "learning_rate": 0.000996,
+      "loss": 6.9039,
       "step": 501
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029939999999999996,
-      "loss": 4.9734,
+      "learning_rate": 0.000998,
+      "loss": 7.2881,
       "step": 502
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0003,
-      "loss": 4.9882,
+      "learning_rate": 0.001,
+      "loss": 6.454,
       "step": 503
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002999381698268755,
-      "loss": 4.9975,
+      "learning_rate": 0.0009997938994229184,
+      "loss": 5.8195,
       "step": 504
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000299876339653751,
-      "loss": 5.5896,
+      "learning_rate": 0.0009995877988458368,
+      "loss": 6.1896,
       "step": 505
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0002998145094806265,
-      "loss": 5.1359,
+      "learning_rate": 0.0009993816982687552,
+      "loss": 9.0881,
       "step": 506
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029975267930750204,
-      "loss": 4.8183,
+      "learning_rate": 0.0009991755976916736,
+      "loss": 5.5124,
       "step": 507
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00029969084913437753,
-      "loss": 4.9363,
+      "learning_rate": 0.000998969497114592,
+      "loss": 5.7427,
       "step": 508
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002996290189612531,
-      "loss": 5.7693,
+      "learning_rate": 0.0009987633965375105,
+      "loss": 6.6938,
       "step": 509
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029956718878812856,
-      "loss": 5.5971,
+      "learning_rate": 0.0009985572959604287,
+      "loss": 5.789,
       "step": 510
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002995053586150041,
-      "loss": 4.9724,
+      "learning_rate": 0.000998351195383347,
+      "loss": 5.7052,
       "step": 511
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002994435284418796,
-      "loss": 5.2413,
+      "learning_rate": 0.0009981450948062655,
+      "loss": 6.1583,
       "step": 512
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029938169826875514,
-      "loss": 5.7321,
+      "learning_rate": 0.0009979389942291839,
+      "loss": 6.2109,
       "step": 513
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029931986809563063,
-      "loss": 4.6161,
+      "learning_rate": 0.0009977328936521023,
+      "loss": 5.2647,
       "step": 514
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002992580379225061,
-      "loss": 4.8212,
+      "learning_rate": 0.0009975267930750207,
+      "loss": 5.5793,
       "step": 515
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029919620774938166,
-      "loss": 5.8654,
+      "learning_rate": 0.000997320692497939,
+      "loss": 7.4105,
       "step": 516
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002991343775762572,
-      "loss": 5.1853,
+      "learning_rate": 0.0009971145919208575,
+      "loss": 6.5091,
       "step": 517
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002990725474031327,
-      "loss": 5.2604,
+      "learning_rate": 0.000996908491343776,
+      "loss": 6.1413,
       "step": 518
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029901071723000824,
-      "loss": 4.9796,
+      "learning_rate": 0.0009967023907666943,
+      "loss": 5.7145,
       "step": 519
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029894888705688373,
-      "loss": 5.12,
+      "learning_rate": 0.0009964962901896125,
+      "loss": 6.0263,
       "step": 520
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029888705688375927,
-      "loss": 4.6269,
+      "learning_rate": 0.000996290189612531,
+      "loss": 5.3656,
       "step": 521
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029882522671063476,
-      "loss": 5.1426,
+      "learning_rate": 0.0009960840890354493,
+      "loss": 6.0043,
       "step": 522
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002987633965375103,
-      "loss": 5.6056,
+      "learning_rate": 0.0009958779884583677,
+      "loss": 6.4915,
       "step": 523
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002987015663643858,
-      "loss": 5.434,
+      "learning_rate": 0.0009956718878812861,
+      "loss": 5.9178,
       "step": 524
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0002986397361912613,
-      "loss": 6.672,
+      "learning_rate": 0.0009954657873042045,
+      "loss": 7.8161,
       "step": 525
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.00029857790601813683,
-      "loss": 5.4697,
+      "learning_rate": 0.000995259686727123,
+      "loss": 5.6401,
       "step": 526
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029851607584501237,
-      "loss": 5.572,
+      "learning_rate": 0.0009950535861500413,
+      "loss": 6.6302,
       "step": 527
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029845424567188786,
-      "loss": 4.6859,
+      "learning_rate": 0.0009948474855729598,
+      "loss": 5.6324,
       "step": 528
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029839241549876335,
-      "loss": 5.3184,
+      "learning_rate": 0.000994641384995878,
+      "loss": 6.2261,
       "step": 529
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0002983305853256389,
-      "loss": 5.503,
+      "learning_rate": 0.0009944352844187964,
+      "loss": 7.1911,
       "step": 530
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0002982687551525144,
-      "loss": 4.7562,
+      "learning_rate": 0.0009942291838417148,
+      "loss": 5.6159,
       "step": 531
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029820692497938993,
-      "loss": 4.776,
+      "learning_rate": 0.0009940230832646332,
+      "loss": 5.4483,
       "step": 532
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0002981450948062654,
-      "loss": 4.8236,
+      "learning_rate": 0.0009938169826875516,
+      "loss": 5.6703,
       "step": 533
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029808326463314096,
-      "loss": 4.6489,
+      "learning_rate": 0.00099361088211047,
+      "loss": 6.1458,
       "step": 534
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029802143446001645,
-      "loss": 5.2644,
+      "learning_rate": 0.0009934047815333884,
+      "loss": 6.1687,
       "step": 535
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.000297959604286892,
-      "loss": 4.9271,
+      "learning_rate": 0.0009931986809563068,
+      "loss": 5.733,
       "step": 536
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0002978977741137675,
-      "loss": 5.0318,
+      "learning_rate": 0.0009929925803792252,
+      "loss": 5.6151,
       "step": 537
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.000297835943940643,
-      "loss": 5.3545,
+      "learning_rate": 0.0009927864798021434,
+      "loss": 5.9331,
       "step": 538
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0002977741137675185,
-      "loss": 5.1369,
+      "learning_rate": 0.0009925803792250618,
+      "loss": 5.6894,
       "step": 539
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029771228359439406,
-      "loss": 5.0709,
+      "learning_rate": 0.0009923742786479802,
+      "loss": 5.7838,
       "step": 540
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029765045342126955,
-      "loss": 5.4444,
+      "learning_rate": 0.0009921681780708986,
+      "loss": 7.2282,
       "step": 541
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029758862324814504,
-      "loss": 5.036,
+      "learning_rate": 0.000991962077493817,
+      "loss": 6.2396,
       "step": 542
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0002975267930750206,
-      "loss": 5.8727,
+      "learning_rate": 0.0009917559769167354,
+      "loss": 7.2881,
       "step": 543
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00029746496290189607,
-      "loss": 4.9163,
+      "learning_rate": 0.0009915498763396538,
+      "loss": 5.969,
       "step": 544
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002974031327287716,
-      "loss": 4.8211,
+      "learning_rate": 0.0009913437757625722,
+      "loss": 5.6602,
       "step": 545
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029734130255564716,
-      "loss": 5.5999,
+      "learning_rate": 0.0009911376751854906,
+      "loss": 7.742,
       "step": 546
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029727947238252265,
-      "loss": 5.7078,
+      "learning_rate": 0.0009909315746084088,
+      "loss": 6.4301,
       "step": 547
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029721764220939814,
-      "loss": 5.4751,
+      "learning_rate": 0.0009907254740313272,
+      "loss": 7.2254,
       "step": 548
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002971558120362737,
-      "loss": 5.4486,
+      "learning_rate": 0.0009905193734542456,
+      "loss": 7.9274,
       "step": 549
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002970939818631492,
-      "loss": 5.0512,
+      "learning_rate": 0.000990313272877164,
+      "loss": 6.0151,
       "step": 550
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002970321516900247,
-      "loss": 5.595,
+      "learning_rate": 0.0009901071723000825,
+      "loss": 6.6425,
       "step": 551
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002969703215169002,
-      "loss": 5.7528,
+      "learning_rate": 0.0009899010717230009,
+      "loss": 6.7184,
       "step": 552
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029690849134377575,
-      "loss": 6.7085,
+      "learning_rate": 0.0009896949711459193,
+      "loss": 6.3715,
       "step": 553
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029684666117065124,
-      "loss": 4.9963,
+      "learning_rate": 0.0009894888705688377,
+      "loss": 5.9396,
       "step": 554
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002967848309975268,
-      "loss": 4.8754,
+      "learning_rate": 0.000989282769991756,
+      "loss": 5.7681,
       "step": 555
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029672300082440227,
-      "loss": 5.5662,
+      "learning_rate": 0.0009890766694146743,
+      "loss": 6.3127,
       "step": 556
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002966611706512778,
-      "loss": 5.6163,
+      "learning_rate": 0.0009888705688375927,
+      "loss": 6.206,
       "step": 557
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0002965993404781533,
-      "loss": 5.0489,
+      "learning_rate": 0.000988664468260511,
+      "loss": 6.8175,
       "step": 558
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029653751030502885,
-      "loss": 4.7127,
+      "learning_rate": 0.0009884583676834295,
+      "loss": 5.5251,
       "step": 559
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029647568013190434,
-      "loss": 4.6543,
+      "learning_rate": 0.000988252267106348,
+      "loss": 5.6403,
       "step": 560
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.00029641384995877983,
-      "loss": 5.1812,
+      "learning_rate": 0.0009880461665292663,
+      "loss": 5.882,
       "step": 561
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00029635201978565537,
-      "loss": 5.435,
+      "learning_rate": 0.0009878400659521847,
+      "loss": 5.891,
       "step": 562
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002962901896125309,
-      "loss": 4.6278,
+      "learning_rate": 0.0009876339653751031,
+      "loss": 5.5482,
       "step": 563
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002962283594394064,
-      "loss": 5.3283,
+      "learning_rate": 0.0009874278647980215,
+      "loss": 5.9301,
       "step": 564
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002961665292662819,
-      "loss": 5.4768,
+      "learning_rate": 0.0009872217642209397,
+      "loss": 5.9511,
       "step": 565
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00029610469909315744,
-      "loss": 4.9921,
+      "learning_rate": 0.0009870156636438581,
+      "loss": 5.6083,
       "step": 566
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002960428689200329,
-      "loss": 5.1908,
+      "learning_rate": 0.0009868095630667765,
+      "loss": 6.5232,
       "step": 567
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00029598103874690847,
-      "loss": 5.7388,
+      "learning_rate": 0.000986603462489695,
+      "loss": 5.9832,
       "step": 568
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.000295919208573784,
-      "loss": 4.8959,
+      "learning_rate": 0.0009863973619126133,
+      "loss": 5.7986,
       "step": 569
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002958573784006595,
-      "loss": 5.1751,
+      "learning_rate": 0.0009861912613355318,
+      "loss": 5.7474,
       "step": 570
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.000295795548227535,
-      "loss": 5.5148,
+      "learning_rate": 0.0009859851607584502,
+      "loss": 6.1929,
       "step": 571
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00029573371805441054,
-      "loss": 5.0948,
+      "learning_rate": 0.0009857790601813686,
+      "loss": 5.623,
       "step": 572
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.000295671887881286,
-      "loss": 5.4853,
+      "learning_rate": 0.000985572959604287,
+      "loss": 6.2775,
       "step": 573
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002956100577081615,
-      "loss": 5.0688,
+      "learning_rate": 0.0009853668590272052,
+      "loss": 6.6127,
       "step": 574
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00029554822753503706,
-      "loss": 4.9018,
+      "learning_rate": 0.0009851607584501236,
+      "loss": 5.7773,
       "step": 575
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002954863973619126,
-      "loss": 5.1228,
+      "learning_rate": 0.000984954657873042,
+      "loss": 5.8412,
       "step": 576
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002954245671887881,
-      "loss": 4.9913,
+      "learning_rate": 0.0009847485572959604,
+      "loss": 5.8474,
       "step": 577
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.00029536273701566364,
-      "loss": 4.856,
+      "learning_rate": 0.0009845424567188788,
+      "loss": 5.5494,
       "step": 578
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0002953009068425391,
-      "loss": 5.0493,
+      "learning_rate": 0.0009843363561417972,
+      "loss": 5.8369,
       "step": 579
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029523907666941467,
-      "loss": 5.7153,
+      "learning_rate": 0.0009841302555647156,
+      "loss": 6.1954,
       "step": 580
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029517724649629016,
-      "loss": 4.7659,
+      "learning_rate": 0.000983924154987634,
+      "loss": 5.5802,
       "step": 581
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002951154163231657,
-      "loss": 5.034,
+      "learning_rate": 0.0009837180544105524,
+      "loss": 5.6219,
       "step": 582
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002950535861500412,
-      "loss": 4.9882,
+      "learning_rate": 0.0009835119538334708,
+      "loss": 5.7712,
       "step": 583
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002949917559769167,
-      "loss": 5.2563,
+      "learning_rate": 0.000983305853256389,
+      "loss": 5.9541,
       "step": 584
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002949299258037922,
-      "loss": 6.8796,
+      "learning_rate": 0.0009830997526793074,
+      "loss": 5.8814,
       "step": 585
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029486809563066777,
-      "loss": 4.6826,
+      "learning_rate": 0.0009828936521022258,
+      "loss": 5.3727,
       "step": 586
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029480626545754326,
-      "loss": 5.5488,
+      "learning_rate": 0.0009826875515251442,
+      "loss": 5.9545,
       "step": 587
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029474443528441875,
-      "loss": 4.9582,
+      "learning_rate": 0.0009824814509480626,
+      "loss": 5.6051,
       "step": 588
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002946826051112943,
-      "loss": 5.7305,
+      "learning_rate": 0.000982275350370981,
+      "loss": 6.2995,
       "step": 589
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002946207749381698,
-      "loss": 5.1785,
+      "learning_rate": 0.0009820692497938995,
+      "loss": 6.0274,
       "step": 590
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002945589447650453,
-      "loss": 4.8395,
+      "learning_rate": 0.0009818631492168179,
+      "loss": 5.613,
       "step": 591
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029449711459192087,
-      "loss": 5.2085,
+      "learning_rate": 0.0009816570486397363,
+      "loss": 5.6562,
       "step": 592
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029443528441879636,
-      "loss": 5.6777,
+      "learning_rate": 0.0009814509480626545,
+      "loss": 5.5516,
       "step": 593
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029437345424567185,
-      "loss": 5.1756,
+      "learning_rate": 0.0009812448474855729,
+      "loss": 5.7627,
       "step": 594
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002943116240725474,
-      "loss": 4.9555,
+      "learning_rate": 0.0009810387469084913,
+      "loss": 5.5813,
       "step": 595
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0002942497938994229,
-      "loss": 5.4113,
+      "learning_rate": 0.0009808326463314099,
+      "loss": 5.636,
       "step": 596
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.00029418796372629837,
-      "loss": 7.6273,
+      "learning_rate": 0.000980626545754328,
+      "loss": 10.2422,
       "step": 597
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0002941261335531739,
-      "loss": 8.7745,
+      "learning_rate": 0.0009804204451772465,
+      "loss": 8.1904,
       "step": 598
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029406430338004946,
-      "loss": 5.3392,
+      "learning_rate": 0.000980214344600165,
+      "loss": 5.8852,
       "step": 599
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029400247320692495,
-      "loss": 5.5342,
+      "learning_rate": 0.0009800082440230833,
+      "loss": 6.0553,
       "step": 600
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0002939406430338005,
-      "loss": 4.9175,
+      "learning_rate": 0.0009798021434460017,
+      "loss": 6.3463,
       "step": 601
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.000293878812860676,
-      "loss": 5.4193,
+      "learning_rate": 0.0009795960428689201,
+      "loss": 6.1136,
       "step": 602
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029381698268755147,
-      "loss": 4.7408,
+      "learning_rate": 0.0009793899422918385,
+      "loss": 5.5926,
       "step": 603
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.000293755152514427,
-      "loss": 4.7971,
+      "learning_rate": 0.000979183841714757,
+      "loss": 5.7269,
       "step": 604
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029369332234130256,
-      "loss": 5.2998,
+      "learning_rate": 0.0009789777411376753,
+      "loss": 6.3031,
       "step": 605
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029363149216817805,
-      "loss": 4.7482,
+      "learning_rate": 0.0009787716405605935,
+      "loss": 5.541,
       "step": 606
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029356966199505354,
-      "loss": 5.6009,
+      "learning_rate": 0.000978565539983512,
+      "loss": 6.0213,
       "step": 607
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0002935078318219291,
-      "loss": 5.2161,
+      "learning_rate": 0.0009783594394064303,
+      "loss": 6.3093,
       "step": 608
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029344600164880457,
-      "loss": 5.3307,
+      "learning_rate": 0.0009781533388293487,
+      "loss": 6.9093,
       "step": 609
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0002933841714756801,
-      "loss": 5.0827,
+      "learning_rate": 0.0009779472382522672,
+      "loss": 5.7258,
       "step": 610
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0002933223413025556,
-      "loss": 5.4163,
+      "learning_rate": 0.0009777411376751856,
+      "loss": 6.208,
       "step": 611
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029326051112943115,
-      "loss": 4.7821,
+      "learning_rate": 0.000977535037098104,
+      "loss": 5.7205,
       "step": 612
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.00029319868095630664,
-      "loss": 4.9356,
+      "learning_rate": 0.0009773289365210224,
+      "loss": 6.4296,
       "step": 613
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0002931368507831822,
-      "loss": 4.996,
+      "learning_rate": 0.0009771228359439408,
+      "loss": 5.8435,
       "step": 614
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0002930750206100577,
-      "loss": 5.7388,
+      "learning_rate": 0.0009769167353668592,
+      "loss": 6.3646,
       "step": 615
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002930131904369332,
-      "loss": 4.8439,
+      "learning_rate": 0.0009767106347897774,
+      "loss": 5.683,
       "step": 616
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002929513602638087,
-      "loss": 5.6029,
+      "learning_rate": 0.0009765045342126958,
+      "loss": 5.9436,
       "step": 617
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029288953009068425,
-      "loss": 4.7227,
+      "learning_rate": 0.0009762984336356142,
+      "loss": 5.8275,
       "step": 618
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029282769991755974,
-      "loss": 5.2727,
+      "learning_rate": 0.0009760923330585326,
+      "loss": 6.4333,
       "step": 619
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002927658697444352,
-      "loss": 5.9921,
+      "learning_rate": 0.0009758862324814509,
+      "loss": 6.5196,
       "step": 620
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029270403957131077,
-      "loss": 5.0419,
+      "learning_rate": 0.0009756801319043693,
+      "loss": 5.7596,
       "step": 621
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002926422093981863,
-      "loss": 5.6111,
+      "learning_rate": 0.0009754740313272877,
+      "loss": 5.7417,
       "step": 622
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002925803792250618,
-      "loss": 5.7807,
+      "learning_rate": 0.0009752679307502061,
+      "loss": 5.7944,
       "step": 623
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029251854905193735,
-      "loss": 7.0621,
+      "learning_rate": 0.0009750618301731245,
+      "loss": 6.7521,
       "step": 624
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029245671887881283,
-      "loss": 5.2582,
+      "learning_rate": 0.0009748557295960428,
+      "loss": 5.6838,
       "step": 625
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002923948887056883,
-      "loss": 4.8489,
+      "learning_rate": 0.0009746496290189612,
+      "loss": 5.5583,
       "step": 626
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029233305853256387,
-      "loss": 4.9177,
+      "learning_rate": 0.0009744435284418796,
+      "loss": 5.5746,
       "step": 627
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002922712283594394,
-      "loss": 6.8271,
+      "learning_rate": 0.000974237427864798,
+      "loss": 7.0217,
       "step": 628
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002922093981863149,
-      "loss": 6.5427,
+      "learning_rate": 0.0009740313272877163,
+      "loss": 7.019,
       "step": 629
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002921475680131904,
-      "loss": 4.7744,
+      "learning_rate": 0.0009738252267106347,
+      "loss": 5.555,
       "step": 630
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029208573784006593,
-      "loss": 4.9416,
+      "learning_rate": 0.0009736191261335531,
+      "loss": 5.9149,
       "step": 631
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0002920239076669414,
-      "loss": 5.315,
+      "learning_rate": 0.0009734130255564716,
+      "loss": 5.7774,
       "step": 632
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00029196207749381697,
-      "loss": 4.9249,
+      "learning_rate": 0.00097320692497939,
+      "loss": 5.7222,
       "step": 633
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.00029190024732069246,
-      "loss": 5.1619,
+      "learning_rate": 0.0009730008244023083,
+      "loss": 5.9334,
       "step": 634
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.000291838417147568,
-      "loss": 4.8749,
+      "learning_rate": 0.0009727947238252267,
+      "loss": 5.6841,
       "step": 635
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002917765869744435,
-      "loss": 5.1713,
+      "learning_rate": 0.0009725886232481451,
+      "loss": 5.8093,
       "step": 636
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.00029171475680131903,
-      "loss": 4.8124,
+      "learning_rate": 0.0009723825226710635,
+      "loss": 5.5973,
       "step": 637
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002916529266281945,
-      "loss": 4.6045,
+      "learning_rate": 0.0009721764220939819,
+      "loss": 5.3166,
       "step": 638
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.00029159109645507,
-      "loss": 5.0988,
+      "learning_rate": 0.0009719703215169002,
+      "loss": 5.7042,
       "step": 639
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.00029152926628194556,
-      "loss": 5.9654,
+      "learning_rate": 0.0009717642209398186,
+      "loss": 6.4014,
       "step": 640
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002914674361088211,
-      "loss": 5.337,
+      "learning_rate": 0.000971558120362737,
+      "loss": 5.773,
       "step": 641
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002914056059356966,
-      "loss": 5.4031,
+      "learning_rate": 0.0009713520197856554,
+      "loss": 6.1479,
       "step": 642
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002913437757625721,
-      "loss": 5.1032,
+      "learning_rate": 0.0009711459192085737,
+      "loss": 9.3503,
       "step": 643
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002912819455894476,
-      "loss": 5.0263,
+      "learning_rate": 0.0009709398186314922,
+      "loss": 6.1811,
       "step": 644
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.00029122011541632317,
-      "loss": 4.8476,
+      "learning_rate": 0.0009707337180544106,
+      "loss": 5.3168,
       "step": 645
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.00029115828524319866,
-      "loss": 5.9561,
+      "learning_rate": 0.000970527617477329,
+      "loss": 5.5771,
       "step": 646
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002910964550700742,
-      "loss": 5.7647,
+      "learning_rate": 0.0009703215169002474,
+      "loss": 8.9244,
       "step": 647
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002910346248969497,
-      "loss": 6.3159,
+      "learning_rate": 0.0009701154163231657,
+      "loss": 9.2924,
       "step": 648
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002909727947238252,
-      "loss": 5.2757,
+      "learning_rate": 0.0009699093157460841,
+      "loss": 6.0032,
       "step": 649
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0002909109645507007,
-      "loss": 5.833,
+      "learning_rate": 0.0009697032151690026,
+      "loss": 5.9065,
       "step": 650
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.00029084913437757627,
-      "loss": 5.262,
+      "learning_rate": 0.000969497114591921,
+      "loss": 5.7178,
       "step": 651
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00029078730420445176,
-      "loss": 4.9441,
+      "learning_rate": 0.0009692910140148393,
+      "loss": 5.5791,
       "step": 652
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00029072547403132724,
-      "loss": 4.9675,
+      "learning_rate": 0.0009690849134377577,
+      "loss": 5.8403,
       "step": 653
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002906636438582028,
-      "loss": 4.91,
+      "learning_rate": 0.0009688788128606761,
+      "loss": 5.8589,
       "step": 654
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002906018136850783,
-      "loss": 5.1493,
+      "learning_rate": 0.0009686727122835945,
+      "loss": 5.8483,
       "step": 655
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002905399835119538,
-      "loss": 4.7846,
+      "learning_rate": 0.0009684666117065129,
+      "loss": 5.6194,
       "step": 656
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002904781533388293,
-      "loss": 4.8601,
+      "learning_rate": 0.0009682605111294312,
+      "loss": 5.5117,
       "step": 657
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00029041632316570485,
-      "loss": 5.8555,
+      "learning_rate": 0.0009680544105523496,
+      "loss": 8.8586,
       "step": 658
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00029035449299258034,
-      "loss": 5.5562,
+      "learning_rate": 0.000967848309975268,
+      "loss": 5.7808,
       "step": 659
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002902926628194559,
-      "loss": 5.0028,
+      "learning_rate": 0.0009676422093981864,
+      "loss": 5.7761,
       "step": 660
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002902308326463314,
-      "loss": 5.0542,
+      "learning_rate": 0.0009674361088211047,
+      "loss": 6.0116,
       "step": 661
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00029016900247320687,
-      "loss": 5.0884,
+      "learning_rate": 0.0009672300082440231,
+      "loss": 5.6008,
       "step": 662
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002901071723000824,
-      "loss": 4.7777,
+      "learning_rate": 0.0009670239076669415,
+      "loss": 6.5432,
       "step": 663
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00029004534212695795,
-      "loss": 5.192,
+      "learning_rate": 0.0009668178070898599,
+      "loss": 5.8096,
       "step": 664
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00028998351195383344,
-      "loss": 5.2064,
+      "learning_rate": 0.0009666117065127783,
+      "loss": 5.4905,
       "step": 665
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00028992168178070893,
-      "loss": 5.5733,
+      "learning_rate": 0.0009664056059356966,
+      "loss": 6.0899,
       "step": 666
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0002898598516075845,
-      "loss": 5.1989,
+      "learning_rate": 0.000966199505358615,
+      "loss": 5.7675,
       "step": 667
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.00028979802143445997,
-      "loss": 5.2734,
+      "learning_rate": 0.0009659934047815334,
+      "loss": 6.1804,
       "step": 668
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002897361912613355,
-      "loss": 4.7041,
+      "learning_rate": 0.0009657873042044518,
+      "loss": 5.5204,
       "step": 669
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028967436108821105,
-      "loss": 5.0419,
+      "learning_rate": 0.0009655812036273703,
+      "loss": 5.6149,
       "step": 670
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028961253091508654,
-      "loss": 4.9719,
+      "learning_rate": 0.0009653751030502886,
+      "loss": 6.0129,
       "step": 671
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028955070074196203,
-      "loss": 5.8946,
+      "learning_rate": 0.000965169002473207,
+      "loss": 6.4508,
       "step": 672
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002894888705688376,
-      "loss": 4.8756,
+      "learning_rate": 0.0009649629018961254,
+      "loss": 5.4699,
       "step": 673
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002894270403957131,
-      "loss": 4.7725,
+      "learning_rate": 0.0009647568013190438,
+      "loss": 5.4832,
       "step": 674
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002893652102225886,
-      "loss": 5.4394,
+      "learning_rate": 0.0009645507007419621,
+      "loss": 5.7112,
       "step": 675
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002893033800494641,
-      "loss": 4.8552,
+      "learning_rate": 0.0009643446001648805,
+      "loss": 5.6204,
       "step": 676
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028924154987633964,
-      "loss": 4.7082,
+      "learning_rate": 0.0009641384995877989,
+      "loss": 5.4164,
       "step": 677
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028917971970321513,
-      "loss": 4.7634,
+      "learning_rate": 0.0009639323990107173,
+      "loss": 5.6556,
       "step": 678
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002891178895300907,
-      "loss": 5.3176,
+      "learning_rate": 0.0009637262984336357,
+      "loss": 6.1959,
       "step": 679
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028905605935696617,
-      "loss": 5.6402,
+      "learning_rate": 0.000963520197856554,
+      "loss": 5.9093,
       "step": 680
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002889942291838417,
-      "loss": 4.9472,
+      "learning_rate": 0.0009633140972794724,
+      "loss": 5.9239,
       "step": 681
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002889323990107172,
-      "loss": 4.8645,
+      "learning_rate": 0.0009631079967023908,
+      "loss": 5.5869,
       "step": 682
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028887056883759274,
-      "loss": 5.421,
+      "learning_rate": 0.0009629018961253092,
+      "loss": 6.027,
       "step": 683
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028880873866446823,
-      "loss": 5.1797,
+      "learning_rate": 0.0009626957955482275,
+      "loss": 5.6705,
       "step": 684
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0002887469084913437,
-      "loss": 5.2275,
+      "learning_rate": 0.0009624896949711459,
+      "loss": 6.0041,
       "step": 685
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.00028868507831821927,
-      "loss": 5.4143,
+      "learning_rate": 0.0009622835943940643,
+      "loss": 5.9309,
       "step": 686
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002886232481450948,
-      "loss": 4.864,
+      "learning_rate": 0.0009620774938169827,
+      "loss": 6.27,
       "step": 687
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002885614179719703,
-      "loss": 4.7454,
+      "learning_rate": 0.0009618713932399011,
+      "loss": 5.6478,
       "step": 688
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002884995877988458,
-      "loss": 5.0972,
+      "learning_rate": 0.0009616652926628194,
+      "loss": 7.2664,
       "step": 689
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028843775762572133,
-      "loss": 4.8467,
+      "learning_rate": 0.0009614591920857378,
+      "loss": 5.6514,
       "step": 690
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002883759274525968,
-      "loss": 4.7513,
+      "learning_rate": 0.0009612530915086563,
+      "loss": 5.4726,
       "step": 691
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028831409727947236,
-      "loss": 5.2937,
+      "learning_rate": 0.0009610469909315747,
+      "loss": 6.023,
       "step": 692
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028825226710634785,
-      "loss": 5.4897,
+      "learning_rate": 0.000960840890354493,
+      "loss": 5.8477,
       "step": 693
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002881904369332234,
-      "loss": 4.8261,
+      "learning_rate": 0.0009606347897774114,
+      "loss": 5.6152,
       "step": 694
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002881286067600989,
-      "loss": 4.947,
+      "learning_rate": 0.0009604286892003298,
+      "loss": 5.6259,
       "step": 695
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028806677658697443,
-      "loss": 5.0018,
+      "learning_rate": 0.0009602225886232482,
+      "loss": 5.6599,
       "step": 696
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002880049464138499,
-      "loss": 4.5913,
+      "learning_rate": 0.0009600164880461666,
+      "loss": 5.4909,
       "step": 697
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002879431162407254,
-      "loss": 4.961,
+      "learning_rate": 0.0009598103874690849,
+      "loss": 5.7744,
       "step": 698
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028788128606760095,
-      "loss": 5.4193,
+      "learning_rate": 0.0009596042868920033,
+      "loss": 5.7638,
       "step": 699
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0002878194558944765,
-      "loss": 5.066,
+      "learning_rate": 0.0009593981863149217,
+      "loss": 5.7855,
       "step": 700
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.000287757625721352,
-      "loss": 4.6608,
+      "learning_rate": 0.0009591920857378401,
+      "loss": 5.6992,
       "step": 701
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028769579554822753,
-      "loss": 4.7681,
+      "learning_rate": 0.0009589859851607585,
+      "loss": 5.5817,
       "step": 702
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.000287633965375103,
-      "loss": 5.2043,
+      "learning_rate": 0.0009587798845836768,
+      "loss": 6.1692,
       "step": 703
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.00028757213520197856,
-      "loss": 6.5327,
+      "learning_rate": 0.0009585737840065952,
+      "loss": 7.3786,
       "step": 704
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00028751030502885405,
-      "loss": 6.29,
+      "learning_rate": 0.0009583676834295136,
+      "loss": 8.9289,
       "step": 705
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002874484748557296,
-      "loss": 5.3758,
+      "learning_rate": 0.000958161582852432,
+      "loss": 6.1429,
       "step": 706
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002873866446826051,
-      "loss": 5.5577,
+      "learning_rate": 0.0009579554822753503,
+      "loss": 7.9422,
       "step": 707
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002873248145094806,
-      "loss": 5.1977,
+      "learning_rate": 0.0009577493816982687,
+      "loss": 6.3416,
       "step": 708
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002872629843363561,
-      "loss": 4.8048,
+      "learning_rate": 0.0009575432811211871,
+      "loss": 5.7308,
       "step": 709
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00028720115416323166,
-      "loss": 4.9544,
+      "learning_rate": 0.0009573371805441055,
+      "loss": 6.0273,
       "step": 710
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00028713932399010715,
-      "loss": 5.2296,
+      "learning_rate": 0.000957131079967024,
+      "loss": 5.9464,
       "step": 711
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00028707749381698264,
-      "loss": 4.8695,
+      "learning_rate": 0.0009569249793899422,
+      "loss": 5.7031,
       "step": 712
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002870156636438582,
-      "loss": 4.8944,
+      "learning_rate": 0.0009567188788128607,
+      "loss": 6.1895,
       "step": 713
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002869538334707337,
-      "loss": 5.3106,
+      "learning_rate": 0.0009565127782357791,
+      "loss": 6.2069,
       "step": 714
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002868920032976092,
-      "loss": 5.604,
+      "learning_rate": 0.0009563066776586975,
+      "loss": 6.162,
       "step": 715
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002868301731244847,
-      "loss": 4.6407,
+      "learning_rate": 0.0009561005770816158,
+      "loss": 5.4371,
       "step": 716
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00028676834295136025,
-      "loss": 5.1252,
+      "learning_rate": 0.0009558944765045342,
+      "loss": 5.5628,
       "step": 717
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00028670651277823574,
-      "loss": 5.1701,
+      "learning_rate": 0.0009556883759274526,
+      "loss": 6.9056,
       "step": 718
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002866446826051113,
-      "loss": 4.7738,
+      "learning_rate": 0.000955482275350371,
+      "loss": 5.5642,
       "step": 719
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002865828524319868,
-      "loss": 4.8216,
+      "learning_rate": 0.0009552761747732894,
+      "loss": 5.9009,
       "step": 720
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.00028652102225886226,
-      "loss": 5.1518,
+      "learning_rate": 0.0009550700741962077,
+      "loss": 6.071,
       "step": 721
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0002864591920857378,
-      "loss": 5.0214,
+      "learning_rate": 0.0009548639736191261,
+      "loss": 6.0097,
       "step": 722
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028639736191261335,
-      "loss": 4.9003,
+      "learning_rate": 0.0009546578730420445,
+      "loss": 5.6494,
       "step": 723
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028633553173948884,
-      "loss": 6.2172,
+      "learning_rate": 0.0009544517724649629,
+      "loss": 6.4281,
       "step": 724
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002862737015663644,
-      "loss": 4.9487,
+      "learning_rate": 0.0009542456718878813,
+      "loss": 5.8058,
       "step": 725
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002862118713932399,
-      "loss": 5.3658,
+      "learning_rate": 0.0009540395713107996,
+      "loss": 5.9138,
       "step": 726
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028615004122011536,
-      "loss": 4.4905,
+      "learning_rate": 0.000953833470733718,
+      "loss": 5.5967,
       "step": 727
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002860882110469909,
-      "loss": 5.2696,
+      "learning_rate": 0.0009536273701566364,
+      "loss": 6.0279,
       "step": 728
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028602638087386645,
-      "loss": 4.796,
+      "learning_rate": 0.0009534212695795548,
+      "loss": 5.4258,
       "step": 729
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028596455070074194,
-      "loss": 4.8293,
+      "learning_rate": 0.0009532151690024731,
+      "loss": 5.5545,
       "step": 730
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028590272052761743,
-      "loss": 4.7833,
+      "learning_rate": 0.0009530090684253915,
+      "loss": 5.568,
       "step": 731
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.000285840890354493,
-      "loss": 4.7773,
+      "learning_rate": 0.00095280296784831,
+      "loss": 5.6758,
       "step": 732
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002857790601813685,
-      "loss": 6.026,
+      "learning_rate": 0.0009525968672712284,
+      "loss": 7.9411,
       "step": 733
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.000285717230008244,
-      "loss": 4.8748,
+      "learning_rate": 0.0009523907666941468,
+      "loss": 5.7769,
       "step": 734
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002856553998351195,
-      "loss": 5.1061,
+      "learning_rate": 0.0009521846661170651,
+      "loss": 5.7837,
       "step": 735
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028559356966199504,
-      "loss": 4.8654,
+      "learning_rate": 0.0009519785655399835,
+      "loss": 5.6159,
       "step": 736
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028553173948887053,
-      "loss": 4.7432,
+      "learning_rate": 0.000951772464962902,
+      "loss": 5.4329,
       "step": 737
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002854699093157461,
-      "loss": 4.7289,
+      "learning_rate": 0.0009515663643858204,
+      "loss": 5.5327,
       "step": 738
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00028540807914262156,
-      "loss": 5.7506,
+      "learning_rate": 0.0009513602638087387,
+      "loss": 7.4568,
       "step": 739
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0002853462489694971,
-      "loss": 4.9486,
+      "learning_rate": 0.0009511541632316571,
+      "loss": 5.4145,
       "step": 740
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002852844187963726,
-      "loss": 5.5122,
+      "learning_rate": 0.0009509480626545755,
+      "loss": 6.4719,
       "step": 741
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028522258862324814,
-      "loss": 4.9497,
+      "learning_rate": 0.0009507419620774939,
+      "loss": 5.7968,
       "step": 742
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028516075845012363,
-      "loss": 6.6022,
+      "learning_rate": 0.0009505358615004123,
+      "loss": 9.9958,
       "step": 743
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002850989282769991,
-      "loss": 5.0641,
+      "learning_rate": 0.0009503297609233306,
+      "loss": 6.1593,
       "step": 744
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028503709810387466,
-      "loss": 5.975,
+      "learning_rate": 0.000950123660346249,
+      "loss": 7.4034,
       "step": 745
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002849752679307502,
-      "loss": 5.087,
+      "learning_rate": 0.0009499175597691674,
+      "loss": 5.7294,
       "step": 746
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002849134377576257,
-      "loss": 5.1108,
+      "learning_rate": 0.0009497114591920858,
+      "loss": 5.9461,
       "step": 747
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002848516075845012,
-      "loss": 6.1419,
+      "learning_rate": 0.0009495053586150041,
+      "loss": 8.9013,
       "step": 748
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028478977741137673,
-      "loss": 4.7772,
+      "learning_rate": 0.0009492992580379225,
+      "loss": 5.4214,
       "step": 749
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002847279472382522,
-      "loss": 5.0598,
+      "learning_rate": 0.0009490931574608409,
+      "loss": 5.6228,
       "step": 750
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028466611706512776,
-      "loss": 5.2531,
+      "learning_rate": 0.0009488870568837594,
+      "loss": 6.6634,
       "step": 751
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002846042868920033,
-      "loss": 5.0756,
+      "learning_rate": 0.0009486809563066778,
+      "loss": 5.9863,
       "step": 752
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002845424567188788,
-      "loss": 5.026,
+      "learning_rate": 0.0009484748557295961,
+      "loss": 5.6832,
       "step": 753
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002844806265457543,
-      "loss": 5.3462,
+      "learning_rate": 0.0009482687551525145,
+      "loss": 6.0156,
       "step": 754
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028441879637262983,
-      "loss": 4.7894,
+      "learning_rate": 0.0009480626545754329,
+      "loss": 5.986,
       "step": 755
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0002843569661995053,
-      "loss": 5.229,
+      "learning_rate": 0.0009478565539983513,
+      "loss": 5.996,
       "step": 756
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028429513602638086,
-      "loss": 5.3994,
+      "learning_rate": 0.0009476504534212697,
+      "loss": 5.9839,
       "step": 757
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.00028423330585325635,
-      "loss": 5.0722,
+      "learning_rate": 0.000947444352844188,
+      "loss": 5.7588,
       "step": 758
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002841714756801319,
-      "loss": 4.8998,
+      "learning_rate": 0.0009472382522671064,
+      "loss": 5.4257,
       "step": 759
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002841096455070074,
-      "loss": 4.943,
+      "learning_rate": 0.0009470321516900248,
+      "loss": 5.5268,
       "step": 760
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028404781533388293,
-      "loss": 5.08,
+      "learning_rate": 0.0009468260511129432,
+      "loss": 5.7849,
       "step": 761
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002839859851607584,
-      "loss": 4.9773,
+      "learning_rate": 0.0009466199505358615,
+      "loss": 6.3809,
       "step": 762
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028392415498763396,
-      "loss": 4.8733,
+      "learning_rate": 0.0009464138499587799,
+      "loss": 5.6375,
       "step": 763
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028386232481450945,
-      "loss": 4.6858,
+      "learning_rate": 0.0009462077493816983,
+      "loss": 5.8323,
       "step": 764
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.000283800494641385,
-      "loss": 4.6454,
+      "learning_rate": 0.0009460016488046167,
+      "loss": 5.6393,
       "step": 765
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002837386644682605,
-      "loss": 5.4697,
+      "learning_rate": 0.0009457955482275351,
+      "loss": 6.0319,
       "step": 766
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.000283676834295136,
-      "loss": 5.4983,
+      "learning_rate": 0.0009455894476504534,
+      "loss": 6.3085,
       "step": 767
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002836150041220115,
-      "loss": 4.6195,
+      "learning_rate": 0.0009453833470733718,
+      "loss": 5.9428,
       "step": 768
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028355317394888706,
-      "loss": 5.2511,
+      "learning_rate": 0.0009451772464962902,
+      "loss": 6.3527,
       "step": 769
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028349134377576255,
-      "loss": 4.8107,
+      "learning_rate": 0.0009449711459192086,
+      "loss": 5.9388,
       "step": 770
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028342951360263804,
-      "loss": 5.0851,
+      "learning_rate": 0.0009447650453421269,
+      "loss": 5.9063,
       "step": 771
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002833676834295136,
-      "loss": 4.7895,
+      "learning_rate": 0.0009445589447650453,
+      "loss": 5.6981,
       "step": 772
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028330585325638907,
-      "loss": 5.0976,
+      "learning_rate": 0.0009443528441879638,
+      "loss": 5.8204,
       "step": 773
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0002832440230832646,
-      "loss": 5.2343,
+      "learning_rate": 0.0009441467436108822,
+      "loss": 6.3106,
       "step": 774
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028318219291014016,
-      "loss": 4.8991,
+      "learning_rate": 0.0009439406430338006,
+      "loss": 5.7092,
       "step": 775
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.00028312036273701565,
-      "loss": 5.0805,
+      "learning_rate": 0.0009437345424567189,
+      "loss": 6.1343,
       "step": 776
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028305853256389114,
-      "loss": 4.6783,
+      "learning_rate": 0.0009435284418796373,
+      "loss": 6.0211,
       "step": 777
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0002829967023907667,
-      "loss": 5.2826,
+      "learning_rate": 0.0009433223413025557,
+      "loss": 5.7363,
       "step": 778
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028293487221764217,
-      "loss": 4.9449,
+      "learning_rate": 0.0009431162407254741,
+      "loss": 5.7902,
       "step": 779
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028287304204451766,
-      "loss": 4.9863,
+      "learning_rate": 0.0009429101401483924,
+      "loss": 5.7293,
       "step": 780
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0002828112118713932,
-      "loss": 5.2948,
+      "learning_rate": 0.0009427040395713108,
+      "loss": 5.8983,
       "step": 781
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028274938169826875,
-      "loss": 5.0827,
+      "learning_rate": 0.0009424979389942292,
+      "loss": 5.7028,
       "step": 782
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028268755152514424,
-      "loss": 4.9482,
+      "learning_rate": 0.0009422918384171476,
+      "loss": 5.5494,
       "step": 783
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0002826257213520198,
-      "loss": 4.7732,
+      "learning_rate": 0.000942085737840066,
+      "loss": 5.6193,
       "step": 784
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028256389117889527,
-      "loss": 4.8009,
+      "learning_rate": 0.0009418796372629843,
+      "loss": 5.9452,
       "step": 785
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028250206100577076,
-      "loss": 5.3838,
+      "learning_rate": 0.0009416735366859027,
+      "loss": 6.301,
       "step": 786
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0002824402308326463,
-      "loss": 5.6277,
+      "learning_rate": 0.0009414674361088211,
+      "loss": 5.9136,
       "step": 787
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028237840065952185,
-      "loss": 5.1932,
+      "learning_rate": 0.0009412613355317395,
+      "loss": 5.9246,
       "step": 788
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028231657048639734,
-      "loss": 5.2417,
+      "learning_rate": 0.0009410552349546579,
+      "loss": 6.4458,
       "step": 789
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028225474031327283,
-      "loss": 5.0448,
+      "learning_rate": 0.0009408491343775762,
+      "loss": 6.2647,
       "step": 790
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028219291014014837,
-      "loss": 5.2999,
+      "learning_rate": 0.0009406430338004946,
+      "loss": 5.8964,
       "step": 791
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.00028213107996702386,
-      "loss": 5.3956,
+      "learning_rate": 0.000940436933223413,
+      "loss": 5.638,
       "step": 792
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0002820692497938994,
-      "loss": 5.0752,
+      "learning_rate": 0.0009402308326463315,
+      "loss": 5.8369,
       "step": 793
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002820074196207749,
-      "loss": 4.8383,
+      "learning_rate": 0.0009400247320692498,
+      "loss": 5.8889,
       "step": 794
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.00028194558944765044,
-      "loss": 5.8318,
+      "learning_rate": 0.0009398186314921682,
+      "loss": 6.2636,
       "step": 795
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002818837592745259,
-      "loss": 5.6796,
+      "learning_rate": 0.0009396125309150866,
+      "loss": 7.3261,
       "step": 796
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.00028182192910140147,
-      "loss": 5.4293,
+      "learning_rate": 0.000939406430338005,
+      "loss": 6.69,
       "step": 797
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.000281760098928277,
-      "loss": 5.0637,
+      "learning_rate": 0.0009392003297609234,
+      "loss": 5.7478,
       "step": 798
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002816982687551525,
-      "loss": 5.0931,
+      "learning_rate": 0.0009389942291838417,
+      "loss": 5.9056,
       "step": 799
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.000281636438582028,
-      "loss": 5.2544,
+      "learning_rate": 0.0009387881286067601,
+      "loss": 5.7171,
       "step": 800
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.00028157460840890354,
-      "loss": 5.231,
+      "learning_rate": 0.0009385820280296785,
+      "loss": 7.3809,
       "step": 801
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.000281512778235779,
-      "loss": 5.1647,
+      "learning_rate": 0.0009383759274525969,
+      "loss": 5.8822,
       "step": 802
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002814509480626545,
-      "loss": 5.6684,
+      "learning_rate": 0.0009381698268755152,
+      "loss": 6.8818,
       "step": 803
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.00028138911788953006,
-      "loss": 4.9627,
+      "learning_rate": 0.0009379637262984336,
+      "loss": 5.5852,
       "step": 804
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002813272877164056,
-      "loss": 5.0181,
+      "learning_rate": 0.000937757625721352,
+      "loss": 7.3812,
       "step": 805
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002812654575432811,
-      "loss": 5.2562,
+      "learning_rate": 0.0009375515251442704,
+      "loss": 6.2432,
       "step": 806
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.00028120362737015664,
-      "loss": 4.8401,
+      "learning_rate": 0.0009373454245671888,
+      "loss": 5.7977,
       "step": 807
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002811417971970321,
-      "loss": 4.8153,
+      "learning_rate": 0.0009371393239901071,
+      "loss": 5.4737,
       "step": 808
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002810799670239076,
-      "loss": 5.0648,
+      "learning_rate": 0.0009369332234130255,
+      "loss": 10.2999,
       "step": 809
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.00028101813685078316,
-      "loss": 5.68,
+      "learning_rate": 0.0009367271228359439,
+      "loss": 6.8698,
       "step": 810
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0002809563066776587,
-      "loss": 4.94,
+      "learning_rate": 0.0009365210222588623,
+      "loss": 5.6809,
       "step": 811
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002808944765045342,
-      "loss": 5.3121,
+      "learning_rate": 0.0009363149216817806,
+      "loss": 6.2355,
       "step": 812
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002808326463314097,
-      "loss": 5.141,
+      "learning_rate": 0.000936108821104699,
+      "loss": 6.2455,
       "step": 813
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002807708161582852,
-      "loss": 4.9714,
+      "learning_rate": 0.0009359027205276175,
+      "loss": 5.5169,
       "step": 814
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002807089859851607,
-      "loss": 4.8498,
+      "learning_rate": 0.0009356966199505359,
+      "loss": 5.7886,
       "step": 815
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.00028064715581203626,
-      "loss": 5.0005,
+      "learning_rate": 0.0009354905193734543,
+      "loss": 5.8093,
       "step": 816
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.00028058532563891175,
-      "loss": 5.4404,
+      "learning_rate": 0.0009352844187963726,
+      "loss": 6.4745,
       "step": 817
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002805234954657873,
-      "loss": 5.1587,
+      "learning_rate": 0.000935078318219291,
+      "loss": 5.9947,
       "step": 818
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002804616652926628,
-      "loss": 5.0171,
+      "learning_rate": 0.0009348722176422094,
+      "loss": 5.7777,
       "step": 819
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002803998351195383,
-      "loss": 5.0557,
+      "learning_rate": 0.0009346661170651278,
+      "loss": 5.8253,
       "step": 820
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002803380049464138,
-      "loss": 5.1384,
+      "learning_rate": 0.0009344600164880462,
+      "loss": 5.978,
       "step": 821
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002802761747732893,
-      "loss": 5.0333,
+      "learning_rate": 0.0009342539159109645,
+      "loss": 5.7328,
       "step": 822
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.00028021434460016485,
-      "loss": 6.1588,
+      "learning_rate": 0.0009340478153338829,
+      "loss": 7.2603,
       "step": 823
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002801525144270404,
-      "loss": 5.1075,
+      "learning_rate": 0.0009338417147568013,
+      "loss": 5.9169,
       "step": 824
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002800906842539159,
-      "loss": 5.4776,
+      "learning_rate": 0.0009336356141797197,
+      "loss": 6.128,
       "step": 825
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.00028002885408079137,
-      "loss": 4.913,
+      "learning_rate": 0.000933429513602638,
+      "loss": 5.7228,
       "step": 826
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0002799670239076669,
-      "loss": 5.6346,
+      "learning_rate": 0.0009332234130255564,
+      "loss": 6.1841,
       "step": 827
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.00027990519373454246,
-      "loss": 4.8558,
+      "learning_rate": 0.0009330173124484748,
+      "loss": 6.1751,
       "step": 828
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.00027984336356141795,
-      "loss": 5.7263,
+      "learning_rate": 0.0009328112118713932,
+      "loss": 6.1,
       "step": 829
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0002797815333882935,
-      "loss": 5.0485,
+      "learning_rate": 0.0009326051112943117,
+      "loss": 5.7972,
       "step": 830
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.000279719703215169,
-      "loss": 5.2009,
+      "learning_rate": 0.00093239901071723,
+      "loss": 6.0542,
       "step": 831
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027965787304204447,
-      "loss": 5.5453,
+      "learning_rate": 0.0009321929101401484,
+      "loss": 6.0632,
       "step": 832
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027959604286892,
-      "loss": 5.3323,
+      "learning_rate": 0.0009319868095630669,
+      "loss": 5.9715,
       "step": 833
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027953421269579556,
-      "loss": 4.4498,
+      "learning_rate": 0.0009317807089859853,
+      "loss": 5.2951,
       "step": 834
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027947238252267105,
-      "loss": 4.9972,
+      "learning_rate": 0.0009315746084089036,
+      "loss": 5.4006,
       "step": 835
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027941055234954654,
-      "loss": 4.8833,
+      "learning_rate": 0.000931368507831822,
+      "loss": 5.7872,
       "step": 836
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0002793487221764221,
-      "loss": 5.1122,
+      "learning_rate": 0.0009311624072547404,
+      "loss": 5.9822,
       "step": 837
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027928689200329757,
-      "loss": 5.078,
+      "learning_rate": 0.0009309563066776588,
+      "loss": 5.4688,
       "step": 838
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0002792250618301731,
-      "loss": 4.7121,
+      "learning_rate": 0.0009307502061005772,
+      "loss": 6.0008,
       "step": 839
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0002791632316570486,
-      "loss": 5.4634,
+      "learning_rate": 0.0009305441055234955,
+      "loss": 6.3812,
       "step": 840
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027910140148392415,
-      "loss": 5.3404,
+      "learning_rate": 0.0009303380049464139,
+      "loss": 7.813,
       "step": 841
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027903957131079964,
-      "loss": 4.7014,
+      "learning_rate": 0.0009301319043693323,
+      "loss": 5.4515,
       "step": 842
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0002789777411376752,
-      "loss": 4.7585,
+      "learning_rate": 0.0009299258037922507,
+      "loss": 6.1169,
       "step": 843
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027891591096455067,
-      "loss": 5.0795,
+      "learning_rate": 0.0009297197032151691,
+      "loss": 6.059,
       "step": 844
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027885408079142616,
-      "loss": 4.7597,
+      "learning_rate": 0.0009295136026380874,
+      "loss": 5.5557,
       "step": 845
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0002787922506183017,
-      "loss": 4.6819,
+      "learning_rate": 0.0009293075020610058,
+      "loss": 5.468,
       "step": 846
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00027873042044517725,
-      "loss": 4.8754,
+      "learning_rate": 0.0009291014014839242,
+      "loss": 5.4248,
       "step": 847
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027866859027205274,
-      "loss": 5.1731,
+      "learning_rate": 0.0009288953009068426,
+      "loss": 6.0027,
       "step": 848
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002786067600989282,
-      "loss": 5.0501,
+      "learning_rate": 0.0009286892003297609,
+      "loss": 5.6318,
       "step": 849
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027854492992580377,
-      "loss": 6.454,
+      "learning_rate": 0.0009284830997526793,
+      "loss": 9.163,
       "step": 850
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027848309975267926,
-      "loss": 5.0942,
+      "learning_rate": 0.0009282769991755977,
+      "loss": 5.8012,
       "step": 851
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002784212695795548,
-      "loss": 4.8875,
+      "learning_rate": 0.0009280708985985161,
+      "loss": 5.7268,
       "step": 852
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027835943940643035,
-      "loss": 5.8274,
+      "learning_rate": 0.0009278647980214346,
+      "loss": 6.8307,
       "step": 853
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027829760923330583,
-      "loss": 4.9718,
+      "learning_rate": 0.0009276586974443529,
+      "loss": 5.7482,
       "step": 854
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002782357790601813,
-      "loss": 5.068,
+      "learning_rate": 0.0009274525968672713,
+      "loss": 5.7373,
       "step": 855
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027817394888705687,
-      "loss": 5.3884,
+      "learning_rate": 0.0009272464962901897,
+      "loss": 6.0101,
       "step": 856
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002781121187139324,
-      "loss": 4.9956,
+      "learning_rate": 0.0009270403957131081,
+      "loss": 5.6004,
       "step": 857
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002780502885408079,
-      "loss": 5.2126,
+      "learning_rate": 0.0009268342951360264,
+      "loss": 5.7602,
       "step": 858
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002779884583676834,
-      "loss": 5.111,
+      "learning_rate": 0.0009266281945589448,
+      "loss": 5.771,
       "step": 859
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027792662819455893,
-      "loss": 5.2589,
+      "learning_rate": 0.0009264220939818632,
+      "loss": 6.2244,
       "step": 860
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002778647980214344,
-      "loss": 4.663,
+      "learning_rate": 0.0009262159934047816,
+      "loss": 5.3677,
       "step": 861
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027780296784830997,
-      "loss": 5.3915,
+      "learning_rate": 0.0009260098928277,
+      "loss": 5.4577,
       "step": 862
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.00027774113767518546,
-      "loss": 5.4341,
+      "learning_rate": 0.0009258037922506183,
+      "loss": 6.3185,
       "step": 863
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.000277679307502061,
-      "loss": 4.7871,
+      "learning_rate": 0.0009255976916735367,
+      "loss": 5.4712,
       "step": 864
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0002776174773289365,
-      "loss": 5.1811,
+      "learning_rate": 0.0009253915910964551,
+      "loss": 5.9647,
       "step": 865
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.00027755564715581203,
-      "loss": 4.8813,
+      "learning_rate": 0.0009251854905193735,
+      "loss": 5.6616,
       "step": 866
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002774938169826875,
-      "loss": 5.0584,
+      "learning_rate": 0.0009249793899422918,
+      "loss": 5.525,
       "step": 867
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.000277431986809563,
-      "loss": 5.322,
+      "learning_rate": 0.0009247732893652102,
+      "loss": 5.972,
       "step": 868
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.00027737015663643856,
-      "loss": 4.9582,
+      "learning_rate": 0.0009245671887881286,
+      "loss": 5.5719,
       "step": 869
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002773083264633141,
-      "loss": 4.815,
+      "learning_rate": 0.000924361088211047,
+      "loss": 5.5685,
       "step": 870
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002772464962901896,
-      "loss": 4.8825,
+      "learning_rate": 0.0009241549876339654,
+      "loss": 5.5242,
       "step": 871
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002771846661170651,
-      "loss": 4.9803,
+      "learning_rate": 0.0009239488870568837,
+      "loss": 5.6932,
       "step": 872
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002771228359439406,
-      "loss": 5.0011,
+      "learning_rate": 0.0009237427864798021,
+      "loss": 5.6847,
       "step": 873
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002770610057708161,
-      "loss": 5.0284,
+      "learning_rate": 0.0009235366859027206,
+      "loss": 5.6326,
       "step": 874
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.00027699917559769166,
-      "loss": 5.4297,
+      "learning_rate": 0.000923330585325639,
+      "loss": 5.8906,
       "step": 875
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002769373454245672,
-      "loss": 5.2466,
+      "learning_rate": 0.0009231244847485574,
+      "loss": 5.7659,
       "step": 876
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002768755152514427,
-      "loss": 5.1956,
+      "learning_rate": 0.0009229183841714757,
+      "loss": 6.3473,
       "step": 877
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002768136850783182,
-      "loss": 4.7665,
+      "learning_rate": 0.0009227122835943941,
+      "loss": 5.5877,
       "step": 878
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002767518549051937,
-      "loss": 4.7455,
+      "learning_rate": 0.0009225061830173125,
+      "loss": 5.4683,
       "step": 879
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002766900247320692,
-      "loss": 4.9378,
+      "learning_rate": 0.0009223000824402309,
+      "loss": 5.643,
       "step": 880
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002766281945589447,
-      "loss": 5.3688,
+      "learning_rate": 0.0009220939818631492,
+      "loss": 5.8368,
       "step": 881
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.00027656636438582024,
-      "loss": 4.7659,
+      "learning_rate": 0.0009218878812860676,
+      "loss": 5.3025,
       "step": 882
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0002765045342126958,
-      "loss": 5.1146,
+      "learning_rate": 0.000921681780708986,
+      "loss": 5.9461,
       "step": 883
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002764427040395713,
-      "loss": 5.0752,
+      "learning_rate": 0.0009214756801319044,
+      "loss": 7.3355,
       "step": 884
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002763808738664468,
-      "loss": 5.4455,
+      "learning_rate": 0.0009212695795548228,
+      "loss": 6.2517,
       "step": 885
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002763190436933223,
-      "loss": 5.0876,
+      "learning_rate": 0.0009210634789777411,
+      "loss": 5.9753,
       "step": 886
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027625721352019786,
-      "loss": 5.5744,
+      "learning_rate": 0.0009208573784006595,
+      "loss": 6.046,
       "step": 887
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027619538334707334,
-      "loss": 4.676,
+      "learning_rate": 0.0009206512778235779,
+      "loss": 5.4745,
       "step": 888
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002761335531739489,
-      "loss": 5.2856,
+      "learning_rate": 0.0009204451772464963,
+      "loss": 5.7831,
       "step": 889
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002760717230008244,
-      "loss": 5.8008,
+      "learning_rate": 0.0009202390766694146,
+      "loss": 6.2167,
       "step": 890
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027600989282769987,
-      "loss": 4.8922,
+      "learning_rate": 0.000920032976092333,
+      "loss": 5.7482,
       "step": 891
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002759480626545754,
-      "loss": 4.8999,
+      "learning_rate": 0.0009198268755152514,
+      "loss": 5.7933,
       "step": 892
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027588623248145095,
-      "loss": 6.0545,
+      "learning_rate": 0.0009196207749381698,
+      "loss": 6.4929,
       "step": 893
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027582440230832644,
-      "loss": 4.9214,
+      "learning_rate": 0.0009194146743610883,
+      "loss": 5.5808,
       "step": 894
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027576257213520193,
-      "loss": 4.8219,
+      "learning_rate": 0.0009192085737840066,
+      "loss": 5.5091,
       "step": 895
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002757007419620775,
-      "loss": 5.6675,
+      "learning_rate": 0.000919002473206925,
+      "loss": 5.7982,
       "step": 896
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027563891178895297,
-      "loss": 4.907,
+      "learning_rate": 0.0009187963726298434,
+      "loss": 5.6103,
       "step": 897
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0002755770816158285,
-      "loss": 6.9768,
+      "learning_rate": 0.0009185902720527618,
+      "loss": 7.3187,
       "step": 898
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.000275515251442704,
-      "loss": 6.5554,
+      "learning_rate": 0.0009183841714756801,
+      "loss": 6.7444,
       "step": 899
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.00027545342126957954,
-      "loss": 10.1171,
+      "learning_rate": 0.0009181780708985985,
+      "loss": 7.599,
       "step": 900
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027539159109645503,
-      "loss": 4.8533,
+      "learning_rate": 0.0009179719703215169,
+      "loss": 5.7065,
       "step": 901
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002753297609233306,
-      "loss": 5.1983,
+      "learning_rate": 0.0009177658697444353,
+      "loss": 5.9893,
       "step": 902
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027526793075020607,
-      "loss": 5.3872,
+      "learning_rate": 0.0009175597691673537,
+      "loss": 6.0056,
       "step": 903
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027520610057708156,
-      "loss": 4.8618,
+      "learning_rate": 0.000917353668590272,
+      "loss": 5.8115,
       "step": 904
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002751442704039571,
-      "loss": 5.1607,
+      "learning_rate": 0.0009171475680131904,
+      "loss": 5.9716,
       "step": 905
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027508244023083264,
-      "loss": 5.2816,
+      "learning_rate": 0.0009169414674361088,
+      "loss": 5.988,
       "step": 906
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027502061005770813,
-      "loss": 5.0874,
+      "learning_rate": 0.0009167353668590272,
+      "loss": 6.2609,
       "step": 907
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002749587798845837,
-      "loss": 4.8244,
+      "learning_rate": 0.0009165292662819456,
+      "loss": 5.4386,
       "step": 908
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027489694971145917,
-      "loss": 5.07,
+      "learning_rate": 0.0009163231657048639,
+      "loss": 6.1184,
       "step": 909
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027483511953833466,
-      "loss": 4.7451,
+      "learning_rate": 0.0009161170651277823,
+      "loss": 5.5782,
       "step": 910
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002747732893652102,
-      "loss": 5.2707,
+      "learning_rate": 0.0009159109645507007,
+      "loss": 6.4379,
       "step": 911
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027471145919208574,
-      "loss": 5.0286,
+      "learning_rate": 0.0009157048639736191,
+      "loss": 5.57,
       "step": 912
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027464962901896123,
-      "loss": 5.2597,
+      "learning_rate": 0.0009154987633965374,
+      "loss": 6.7615,
       "step": 913
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002745877988458367,
-      "loss": 4.8834,
+      "learning_rate": 0.0009152926628194558,
+      "loss": 5.6496,
       "step": 914
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.00027452596867271227,
-      "loss": 5.5641,
+      "learning_rate": 0.0009150865622423743,
+      "loss": 6.6595,
       "step": 915
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002744641384995878,
-      "loss": 5.3105,
+      "learning_rate": 0.0009148804616652927,
+      "loss": 5.81,
       "step": 916
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002744023083264633,
-      "loss": 4.6726,
+      "learning_rate": 0.0009146743610882111,
+      "loss": 5.5627,
       "step": 917
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0002743404781533388,
-      "loss": 5.7179,
+      "learning_rate": 0.0009144682605111294,
+      "loss": 6.9672,
       "step": 918
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.00027427864798021433,
-      "loss": 5.202,
+      "learning_rate": 0.0009142621599340478,
+      "loss": 6.2755,
       "step": 919
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002742168178070898,
-      "loss": 4.48,
+      "learning_rate": 0.0009140560593569662,
+      "loss": 5.3061,
       "step": 920
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.00027415498763396536,
-      "loss": 4.776,
+      "learning_rate": 0.0009138499587798846,
+      "loss": 5.728,
       "step": 921
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.00027409315746084085,
-      "loss": 5.4766,
+      "learning_rate": 0.0009136438582028029,
+      "loss": 5.9284,
       "step": 922
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002740313272877164,
-      "loss": 5.0076,
+      "learning_rate": 0.0009134377576257213,
+      "loss": 5.4885,
       "step": 923
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002739694971145919,
-      "loss": 5.0358,
+      "learning_rate": 0.0009132316570486398,
+      "loss": 6.7174,
       "step": 924
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.00027390766694146743,
-      "loss": 5.1695,
+      "learning_rate": 0.0009130255564715582,
+      "loss": 5.909,
       "step": 925
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002738458367683429,
-      "loss": 5.5351,
+      "learning_rate": 0.0009128194558944766,
+      "loss": 6.3805,
       "step": 926
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002737840065952184,
-      "loss": 4.6783,
+      "learning_rate": 0.0009126133553173949,
+      "loss": 5.4519,
       "step": 927
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.00027372217642209395,
-      "loss": 4.7698,
+      "learning_rate": 0.0009124072547403133,
+      "loss": 5.5004,
       "step": 928
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002736603462489695,
-      "loss": 4.6951,
+      "learning_rate": 0.0009122011541632317,
+      "loss": 5.3908,
       "step": 929
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.000273598516075845,
-      "loss": 5.0038,
+      "learning_rate": 0.0009119950535861501,
+      "loss": 5.737,
       "step": 930
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002735366859027205,
-      "loss": 5.1458,
+      "learning_rate": 0.0009117889530090684,
+      "loss": 5.9619,
       "step": 931
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.000273474855729596,
-      "loss": 5.0728,
+      "learning_rate": 0.0009115828524319868,
+      "loss": 5.8696,
       "step": 932
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002734130255564715,
-      "loss": 5.1142,
+      "learning_rate": 0.0009113767518549052,
+      "loss": 6.2842,
       "step": 933
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.00027335119538334705,
-      "loss": 4.9986,
+      "learning_rate": 0.0009111706512778237,
+      "loss": 5.6991,
       "step": 934
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002732893652102226,
-      "loss": 5.2688,
+      "learning_rate": 0.0009109645507007421,
+      "loss": 5.9549,
       "step": 935
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0002732275350370981,
-      "loss": 5.087,
+      "learning_rate": 0.0009107584501236604,
+      "loss": 5.7849,
       "step": 936
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002731657048639736,
-      "loss": 5.2715,
+      "learning_rate": 0.0009105523495465788,
+      "loss": 5.6284,
       "step": 937
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002731038746908491,
-      "loss": 4.5977,
+      "learning_rate": 0.0009103462489694972,
+      "loss": 5.375,
       "step": 938
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002730420445177246,
-      "loss": 4.7939,
+      "learning_rate": 0.0009101401483924156,
+      "loss": 5.2918,
       "step": 939
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027298021434460015,
-      "loss": 5.2829,
+      "learning_rate": 0.000909934047815334,
+      "loss": 5.8196,
       "step": 940
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027291838417147564,
-      "loss": 4.9306,
+      "learning_rate": 0.0009097279472382523,
+      "loss": 5.8173,
       "step": 941
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002728565539983512,
-      "loss": 4.8539,
+      "learning_rate": 0.0009095218466611707,
+      "loss": 5.5793,
       "step": 942
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002727947238252267,
-      "loss": 6.0389,
+      "learning_rate": 0.0009093157460840891,
+      "loss": 6.2229,
       "step": 943
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002727328936521022,
-      "loss": 5.2289,
+      "learning_rate": 0.0009091096455070075,
+      "loss": 5.5593,
       "step": 944
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002726710634789777,
-      "loss": 4.9719,
+      "learning_rate": 0.0009089035449299258,
+      "loss": 5.9367,
       "step": 945
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027260923330585325,
-      "loss": 4.9032,
+      "learning_rate": 0.0009086974443528442,
+      "loss": 5.749,
       "step": 946
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027254740313272874,
-      "loss": 4.8283,
+      "learning_rate": 0.0009084913437757626,
+      "loss": 5.2255,
       "step": 947
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002724855729596043,
-      "loss": 4.9207,
+      "learning_rate": 0.000908285243198681,
+      "loss": 5.6387,
       "step": 948
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002724237427864798,
-      "loss": 5.9941,
+      "learning_rate": 0.0009080791426215994,
+      "loss": 7.885,
       "step": 949
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027236191261335526,
-      "loss": 4.3609,
+      "learning_rate": 0.0009078730420445177,
+      "loss": 4.9916,
       "step": 950
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0002723000824402308,
-      "loss": 4.8942,
+      "learning_rate": 0.0009076669414674361,
+      "loss": 5.6046,
       "step": 951
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027223825226710635,
-      "loss": 5.1492,
+      "learning_rate": 0.0009074608408903545,
+      "loss": 5.6585,
       "step": 952
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027217642209398184,
-      "loss": 5.1693,
+      "learning_rate": 0.000907254740313273,
+      "loss": 5.683,
       "step": 953
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.00027211459192085733,
-      "loss": 4.7744,
+      "learning_rate": 0.0009070486397361912,
+      "loss": 5.2602,
       "step": 954
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0002720527617477329,
-      "loss": 5.013,
+      "learning_rate": 0.0009068425391591097,
+      "loss": 5.645,
       "step": 955
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027199093157460836,
-      "loss": 4.9034,
+      "learning_rate": 0.0009066364385820281,
+      "loss": 5.4953,
       "step": 956
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0002719291014014839,
-      "loss": 5.1621,
+      "learning_rate": 0.0009064303380049465,
+      "loss": 5.6824,
       "step": 957
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027186727122835945,
-      "loss": 4.7574,
+      "learning_rate": 0.0009062242374278649,
+      "loss": 5.3621,
       "step": 958
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027180544105523494,
-      "loss": 5.6389,
+      "learning_rate": 0.0009060181368507832,
+      "loss": 6.2408,
       "step": 959
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027174361088211043,
-      "loss": 5.6339,
+      "learning_rate": 0.0009058120362737016,
+      "loss": 5.3227,
       "step": 960
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.000271681780708986,
-      "loss": 4.6895,
+      "learning_rate": 0.00090560593569662,
+      "loss": 5.3431,
       "step": 961
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027161995053586146,
-      "loss": 4.9136,
+      "learning_rate": 0.0009053998351195384,
+      "loss": 5.5201,
       "step": 962
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.000271558120362737,
-      "loss": 5.5033,
+      "learning_rate": 0.0009051937345424568,
+      "loss": 5.9274,
       "step": 963
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0002714962901896125,
-      "loss": 4.9679,
+      "learning_rate": 0.0009049876339653751,
+      "loss": 5.6072,
       "step": 964
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027143446001648804,
-      "loss": 4.7838,
+      "learning_rate": 0.0009047815333882935,
+      "loss": 5.419,
       "step": 965
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027137262984336353,
-      "loss": 4.7596,
+      "learning_rate": 0.0009045754328112119,
+      "loss": 5.424,
       "step": 966
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0002713107996702391,
-      "loss": 5.1027,
+      "learning_rate": 0.0009043693322341303,
+      "loss": 5.7684,
       "step": 967
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027124896949711456,
-      "loss": 4.8395,
+      "learning_rate": 0.0009041632316570486,
+      "loss": 5.5606,
       "step": 968
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027118713932399005,
-      "loss": 5.1351,
+      "learning_rate": 0.000903957131079967,
+      "loss": 5.6427,
       "step": 969
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0002711253091508656,
-      "loss": 4.6402,
+      "learning_rate": 0.0009037510305028854,
+      "loss": 5.5729,
       "step": 970
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027106347897774114,
-      "loss": 4.9308,
+      "learning_rate": 0.0009035449299258038,
+      "loss": 6.3085,
       "step": 971
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00027100164880461663,
-      "loss": 4.9653,
+      "learning_rate": 0.0009033388293487222,
+      "loss": 5.432,
       "step": 972
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002709398186314921,
-      "loss": 5.0278,
+      "learning_rate": 0.0009031327287716405,
+      "loss": 5.7285,
       "step": 973
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00027087798845836766,
-      "loss": 5.6573,
+      "learning_rate": 0.000902926628194559,
+      "loss": 6.4965,
       "step": 974
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00027081615828524315,
-      "loss": 5.0273,
+      "learning_rate": 0.0009027205276174774,
+      "loss": 5.6621,
       "step": 975
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002707543281121187,
-      "loss": 5.3102,
+      "learning_rate": 0.0009025144270403958,
+      "loss": 5.7753,
       "step": 976
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002706924979389942,
-      "loss": 5.6042,
+      "learning_rate": 0.0009023083264633141,
+      "loss": 6.1264,
       "step": 977
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00027063066776586973,
-      "loss": 4.8852,
+      "learning_rate": 0.0009021022258862325,
+      "loss": 5.5589,
       "step": 978
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002705688375927452,
-      "loss": 5.4215,
+      "learning_rate": 0.0009018961253091509,
+      "loss": 6.0841,
       "step": 979
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00027050700741962076,
-      "loss": 4.7782,
+      "learning_rate": 0.0009016900247320693,
+      "loss": 5.5604,
       "step": 980
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002704451772464963,
-      "loss": 4.5646,
+      "learning_rate": 0.0009014839241549877,
+      "loss": 5.5459,
       "step": 981
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002703833470733718,
-      "loss": 4.4924,
+      "learning_rate": 0.000901277823577906,
+      "loss": 5.2837,
       "step": 982
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002703215169002473,
-      "loss": 5.2813,
+      "learning_rate": 0.0009010717230008244,
+      "loss": 6.0762,
       "step": 983
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00027025968672712283,
-      "loss": 5.3463,
+      "learning_rate": 0.0009008656224237428,
+      "loss": 5.8828,
       "step": 984
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002701978565539983,
-      "loss": 4.5305,
+      "learning_rate": 0.0009006595218466612,
+      "loss": 5.3435,
       "step": 985
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002701360263808738,
-      "loss": 4.9637,
+      "learning_rate": 0.0009004534212695795,
+      "loss": 6.0708,
       "step": 986
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00027007419620774935,
-      "loss": 5.6083,
+      "learning_rate": 0.0009002473206924979,
+      "loss": 6.2976,
       "step": 987
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002700123660346249,
-      "loss": 4.788,
+      "learning_rate": 0.0009000412201154163,
+      "loss": 5.588,
       "step": 988
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0002699505358615004,
-      "loss": 5.1677,
+      "learning_rate": 0.0008998351195383347,
+      "loss": 5.9371,
       "step": 989
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.00026988870568837593,
-      "loss": 5.1004,
+      "learning_rate": 0.0008996290189612531,
+      "loss": 5.5429,
       "step": 990
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002698268755152514,
-      "loss": 5.5673,
+      "learning_rate": 0.0008994229183841714,
+      "loss": 5.7105,
       "step": 991
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002697650453421269,
-      "loss": 5.1991,
+      "learning_rate": 0.0008992168178070898,
+      "loss": 5.9454,
       "step": 992
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00026970321516900245,
-      "loss": 5.6295,
+      "learning_rate": 0.0008990107172300082,
+      "loss": 5.7403,
       "step": 993
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.000269641384995878,
-      "loss": 5.3505,
+      "learning_rate": 0.0008988046166529266,
+      "loss": 6.7284,
       "step": 994
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002695795548227535,
-      "loss": 4.7926,
+      "learning_rate": 0.000898598516075845,
+      "loss": 5.5366,
       "step": 995
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.000269517724649629,
-      "loss": 4.2548,
+      "learning_rate": 0.0008983924154987634,
+      "loss": 4.9742,
       "step": 996
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002694558944765045,
-      "loss": 4.6709,
+      "learning_rate": 0.0008981863149216818,
+      "loss": 5.5503,
       "step": 997
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00026939406430338,
-      "loss": 5.5215,
+      "learning_rate": 0.0008979802143446002,
+      "loss": 6.2854,
       "step": 998
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00026933223413025555,
-      "loss": 5.1313,
+      "learning_rate": 0.0008977741137675186,
+      "loss": 6.0931,
       "step": 999
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00026927040395713104,
-      "loss": 4.8677,
+      "learning_rate": 0.0008975680131904369,
+      "loss": 5.609,
       "step": 1000
     },
     {
       "epoch": 0.56,
-      "eval_loss": 5.3296637535095215,
-      "eval_runtime": 1244.5287,
-      "eval_samples_per_second": 2.123,
-      "eval_steps_per_second": 0.266,
-      "eval_wer": 1.9524593415311384,
+      "eval_loss": 6.030221462249756,
+      "eval_runtime": 1248.2223,
+      "eval_samples_per_second": 2.117,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.971717572391908,
       "step": 1000
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002692085737840066,
-      "loss": 5.2456,
+      "learning_rate": 0.0008973619126133553,
+      "loss": 6.0239,
       "step": 1001
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00026914674361088207,
-      "loss": 4.877,
+      "learning_rate": 0.0008971558120362737,
+      "loss": 5.5556,
       "step": 1002
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002690849134377576,
-      "loss": 5.0389,
+      "learning_rate": 0.0008969497114591921,
+      "loss": 5.6206,
       "step": 1003
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002690230832646331,
-      "loss": 5.3228,
+      "learning_rate": 0.0008967436108821105,
+      "loss": 5.7808,
       "step": 1004
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002689612530915086,
-      "loss": 5.095,
+      "learning_rate": 0.0008965375103050288,
+      "loss": 5.7128,
       "step": 1005
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.00026889942291838414,
-      "loss": 4.9869,
+      "learning_rate": 0.0008963314097279472,
+      "loss": 5.6755,
       "step": 1006
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0002688375927452597,
-      "loss": 5.8193,
+      "learning_rate": 0.0008961253091508656,
+      "loss": 6.3792,
       "step": 1007
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026877576257213517,
-      "loss": 5.3667,
+      "learning_rate": 0.000895919208573784,
+      "loss": 7.4117,
       "step": 1008
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026871393239901066,
-      "loss": 5.4724,
+      "learning_rate": 0.0008957131079967023,
+      "loss": 6.9529,
       "step": 1009
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0002686521022258862,
-      "loss": 4.8177,
+      "learning_rate": 0.0008955070074196207,
+      "loss": 5.8751,
       "step": 1010
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026859027205276175,
-      "loss": 4.9535,
+      "learning_rate": 0.0008953009068425391,
+      "loss": 5.7386,
       "step": 1011
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026852844187963724,
-      "loss": 5.0368,
+      "learning_rate": 0.0008950948062654575,
+      "loss": 5.8392,
       "step": 1012
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0002684666117065128,
-      "loss": 5.0102,
+      "learning_rate": 0.0008948887056883759,
+      "loss": 5.6897,
       "step": 1013
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026840478153338827,
-      "loss": 5.3243,
+      "learning_rate": 0.0008946826051112942,
+      "loss": 5.6693,
       "step": 1014
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026834295136026376,
-      "loss": 6.2077,
+      "learning_rate": 0.0008944765045342126,
+      "loss": 6.4492,
       "step": 1015
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0002682811211871393,
-      "loss": 4.7894,
+      "learning_rate": 0.000894270403957131,
+      "loss": 5.3406,
       "step": 1016
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026821929101401485,
-      "loss": 5.0627,
+      "learning_rate": 0.0008940643033800496,
+      "loss": 5.6562,
       "step": 1017
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026815746084089034,
-      "loss": 4.9263,
+      "learning_rate": 0.0008938582028029679,
+      "loss": 5.4921,
       "step": 1018
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026809563066776583,
-      "loss": 5.6975,
+      "learning_rate": 0.0008936521022258863,
+      "loss": 6.6062,
       "step": 1019
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026803380049464137,
-      "loss": 5.5463,
+      "learning_rate": 0.0008934460016488047,
+      "loss": 6.914,
       "step": 1020
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026797197032151686,
-      "loss": 5.2125,
+      "learning_rate": 0.0008932399010717231,
+      "loss": 5.7628,
       "step": 1021
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0002679101401483924,
-      "loss": 5.2008,
+      "learning_rate": 0.0008930338004946415,
+      "loss": 5.9083,
       "step": 1022
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0002678483099752679,
-      "loss": 5.0108,
+      "learning_rate": 0.0008928276999175598,
+      "loss": 5.6599,
       "step": 1023
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.00026778647980214344,
-      "loss": 4.9793,
+      "learning_rate": 0.0008926215993404782,
+      "loss": 5.6024,
       "step": 1024
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0002677246496290189,
-      "loss": 5.1991,
+      "learning_rate": 0.0008924154987633966,
+      "loss": 6.2904,
       "step": 1025
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026766281945589447,
-      "loss": 5.1512,
+      "learning_rate": 0.000892209398186315,
+      "loss": 5.6106,
       "step": 1026
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026760098928276996,
-      "loss": 4.9504,
+      "learning_rate": 0.0008920032976092334,
+      "loss": 5.659,
       "step": 1027
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026753915910964545,
-      "loss": 5.0719,
+      "learning_rate": 0.0008917971970321517,
+      "loss": 5.8208,
       "step": 1028
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.000267477328936521,
-      "loss": 4.9952,
+      "learning_rate": 0.0008915910964550701,
+      "loss": 5.5091,
       "step": 1029
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026741549876339654,
-      "loss": 5.1401,
+      "learning_rate": 0.0008913849958779885,
+      "loss": 5.752,
       "step": 1030
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.000267353668590272,
-      "loss": 5.1485,
+      "learning_rate": 0.0008911788953009069,
+      "loss": 5.7708,
       "step": 1031
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002672918384171475,
-      "loss": 5.0709,
+      "learning_rate": 0.0008909727947238252,
+      "loss": 5.5729,
       "step": 1032
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026723000824402306,
-      "loss": 4.7864,
+      "learning_rate": 0.0008907666941467436,
+      "loss": 5.5487,
       "step": 1033
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026716817807089855,
-      "loss": 4.6901,
+      "learning_rate": 0.000890560593569662,
+      "loss": 5.3115,
       "step": 1034
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002671063478977741,
-      "loss": 4.8362,
+      "learning_rate": 0.0008903544929925805,
+      "loss": 5.6185,
       "step": 1035
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026704451772464964,
-      "loss": 5.3691,
+      "learning_rate": 0.0008901483924154989,
+      "loss": 6.773,
       "step": 1036
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002669826875515251,
-      "loss": 5.1614,
+      "learning_rate": 0.0008899422918384172,
+      "loss": 5.5839,
       "step": 1037
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002669208573784006,
-      "loss": 4.859,
+      "learning_rate": 0.0008897361912613356,
+      "loss": 5.6614,
       "step": 1038
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.00026685902720527616,
-      "loss": 4.7919,
+      "learning_rate": 0.000889530090684254,
+      "loss": 6.7267,
       "step": 1039
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002667971970321517,
-      "loss": 5.845,
+      "learning_rate": 0.0008893239901071724,
+      "loss": 6.3976,
       "step": 1040
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002667353668590272,
-      "loss": 5.2092,
+      "learning_rate": 0.0008891178895300907,
+      "loss": 6.5246,
       "step": 1041
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002666735366859027,
-      "loss": 4.9459,
+      "learning_rate": 0.0008889117889530091,
+      "loss": 5.6608,
       "step": 1042
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0002666117065127782,
-      "loss": 5.284,
+      "learning_rate": 0.0008887056883759275,
+      "loss": 5.6547,
       "step": 1043
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002665498763396537,
-      "loss": 5.4155,
+      "learning_rate": 0.0008884995877988459,
+      "loss": 7.4612,
       "step": 1044
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00026648804616652926,
-      "loss": 5.4718,
+      "learning_rate": 0.0008882934872217643,
+      "loss": 5.9616,
       "step": 1045
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00026642621599340475,
-      "loss": 5.0093,
+      "learning_rate": 0.0008880873866446826,
+      "loss": 5.6414,
       "step": 1046
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002663643858202803,
-      "loss": 4.8802,
+      "learning_rate": 0.000887881286067601,
+      "loss": 5.7465,
       "step": 1047
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002663025556471558,
-      "loss": 4.8019,
+      "learning_rate": 0.0008876751854905194,
+      "loss": 5.4455,
       "step": 1048
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002662407254740313,
-      "loss": 4.8741,
+      "learning_rate": 0.0008874690849134378,
+      "loss": 5.9119,
       "step": 1049
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002661788953009068,
-      "loss": 5.6905,
+      "learning_rate": 0.0008872629843363561,
+      "loss": 6.497,
       "step": 1050
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002661170651277823,
-      "loss": 5.7908,
+      "learning_rate": 0.0008870568837592745,
+      "loss": 6.7875,
       "step": 1051
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00026605523495465785,
-      "loss": 4.6712,
+      "learning_rate": 0.0008868507831821929,
+      "loss": 5.4351,
       "step": 1052
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002659934047815334,
-      "loss": 5.2934,
+      "learning_rate": 0.0008866446826051113,
+      "loss": 5.9467,
       "step": 1053
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002659315746084089,
-      "loss": 4.7193,
+      "learning_rate": 0.0008864385820280297,
+      "loss": 5.3352,
       "step": 1054
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00026586974443528437,
-      "loss": 5.715,
+      "learning_rate": 0.000886232481450948,
+      "loss": 5.9635,
       "step": 1055
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002658079142621599,
-      "loss": 4.7603,
+      "learning_rate": 0.0008860263808738665,
+      "loss": 5.5758,
       "step": 1056
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002657460840890354,
-      "loss": 5.3447,
+      "learning_rate": 0.0008858202802967849,
+      "loss": 5.929,
       "step": 1057
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00026568425391591095,
-      "loss": 4.8764,
+      "learning_rate": 0.0008856141797197033,
+      "loss": 5.502,
       "step": 1058
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0002656224237427865,
-      "loss": 6.4631,
+      "learning_rate": 0.0008854080791426217,
+      "loss": 6.1336,
       "step": 1059
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.000265560593569662,
-      "loss": 4.7168,
+      "learning_rate": 0.00088520197856554,
+      "loss": 5.6216,
       "step": 1060
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00026549876339653747,
-      "loss": 4.8549,
+      "learning_rate": 0.0008849958779884584,
+      "loss": 5.7415,
       "step": 1061
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.000265436933223413,
-      "loss": 4.8071,
+      "learning_rate": 0.0008847897774113768,
+      "loss": 5.5518,
       "step": 1062
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002653751030502885,
-      "loss": 5.6072,
+      "learning_rate": 0.0008845836768342952,
+      "loss": 6.3973,
       "step": 1063
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.000265313272877164,
-      "loss": 4.6102,
+      "learning_rate": 0.0008843775762572135,
+      "loss": 5.2913,
       "step": 1064
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026525144270403954,
-      "loss": 5.1969,
+      "learning_rate": 0.0008841714756801319,
+      "loss": 5.9452,
       "step": 1065
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002651896125309151,
-      "loss": 4.6669,
+      "learning_rate": 0.0008839653751030503,
+      "loss": 5.3736,
       "step": 1066
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026512778235779057,
-      "loss": 5.1038,
+      "learning_rate": 0.0008837592745259687,
+      "loss": 5.7383,
       "step": 1067
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002650659521846661,
-      "loss": 5.1638,
+      "learning_rate": 0.0008835531739488871,
+      "loss": 5.4393,
       "step": 1068
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002650041220115416,
-      "loss": 5.1097,
+      "learning_rate": 0.0008833470733718054,
+      "loss": 5.8006,
       "step": 1069
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026494229183841715,
-      "loss": 4.6091,
+      "learning_rate": 0.0008831409727947238,
+      "loss": 5.4522,
       "step": 1070
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026488046166529264,
-      "loss": 4.759,
+      "learning_rate": 0.0008829348722176422,
+      "loss": 5.5037,
       "step": 1071
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002648186314921682,
-      "loss": 4.5411,
+      "learning_rate": 0.0008827287716405606,
+      "loss": 5.2662,
       "step": 1072
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026475680131904367,
-      "loss": 4.6699,
+      "learning_rate": 0.0008825226710634789,
+      "loss": 5.4944,
       "step": 1073
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026469497114591916,
-      "loss": 5.7103,
+      "learning_rate": 0.0008823165704863973,
+      "loss": 6.4176,
       "step": 1074
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002646331409727947,
-      "loss": 5.2251,
+      "learning_rate": 0.0008821104699093157,
+      "loss": 5.66,
       "step": 1075
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026457131079967025,
-      "loss": 5.5478,
+      "learning_rate": 0.0008819043693322342,
+      "loss": 6.5868,
       "step": 1076
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026450948062654574,
-      "loss": 5.3056,
+      "learning_rate": 0.0008816982687551526,
+      "loss": 5.6613,
       "step": 1077
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0002644476504534212,
-      "loss": 4.8029,
+      "learning_rate": 0.0008814921681780709,
+      "loss": 5.4285,
       "step": 1078
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.00026438582028029677,
-      "loss": 4.8439,
+      "learning_rate": 0.0008812860676009893,
+      "loss": 5.5742,
       "step": 1079
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026432399010717226,
-      "loss": 5.1021,
+      "learning_rate": 0.0008810799670239077,
+      "loss": 5.6072,
       "step": 1080
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0002642621599340478,
-      "loss": 5.4544,
+      "learning_rate": 0.0008808738664468261,
+      "loss": 6.9596,
       "step": 1081
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026420032976092335,
-      "loss": 5.4333,
+      "learning_rate": 0.0008806677658697445,
+      "loss": 6.0076,
       "step": 1082
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026413849958779883,
-      "loss": 4.7404,
+      "learning_rate": 0.0008804616652926628,
+      "loss": 5.7447,
       "step": 1083
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0002640766694146743,
-      "loss": 4.9258,
+      "learning_rate": 0.0008802555647155812,
+      "loss": 5.8632,
       "step": 1084
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026401483924154987,
-      "loss": 4.8587,
+      "learning_rate": 0.0008800494641384996,
+      "loss": 5.3555,
       "step": 1085
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026395300906842536,
-      "loss": 5.1042,
+      "learning_rate": 0.000879843363561418,
+      "loss": 6.2399,
       "step": 1086
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026389117889530085,
-      "loss": 5.1434,
+      "learning_rate": 0.0008796372629843363,
+      "loss": 6.6308,
       "step": 1087
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0002638293487221764,
-      "loss": 5.4743,
+      "learning_rate": 0.0008794311624072547,
+      "loss": 6.1564,
       "step": 1088
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026376751854905193,
-      "loss": 5.3319,
+      "learning_rate": 0.0008792250618301731,
+      "loss": 6.0152,
       "step": 1089
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0002637056883759274,
-      "loss": 4.8945,
+      "learning_rate": 0.0008790189612530915,
+      "loss": 5.4628,
       "step": 1090
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026364385820280297,
-      "loss": 4.9671,
+      "learning_rate": 0.0008788128606760099,
+      "loss": 5.6731,
       "step": 1091
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026358202802967846,
-      "loss": 4.9056,
+      "learning_rate": 0.0008786067600989282,
+      "loss": 5.7508,
       "step": 1092
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026352019785655395,
-      "loss": 4.9501,
+      "learning_rate": 0.0008784006595218466,
+      "loss": 5.6051,
       "step": 1093
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0002634583676834295,
-      "loss": 6.4356,
+      "learning_rate": 0.000878194558944765,
+      "loss": 6.0181,
       "step": 1094
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.00026339653751030503,
-      "loss": 4.9518,
+      "learning_rate": 0.0008779884583676834,
+      "loss": 5.6472,
       "step": 1095
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0002633347073371805,
-      "loss": 5.1875,
+      "learning_rate": 0.0008777823577906017,
+      "loss": 6.2686,
       "step": 1096
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.000263272877164056,
-      "loss": 5.287,
+      "learning_rate": 0.0008775762572135202,
+      "loss": 6.6742,
       "step": 1097
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.00026321104699093156,
-      "loss": 6.5556,
+      "learning_rate": 0.0008773701566364386,
+      "loss": 7.2678,
       "step": 1098
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002631492168178071,
-      "loss": 5.0733,
+      "learning_rate": 0.000877164056059357,
+      "loss": 12.335,
       "step": 1099
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002630873866446826,
-      "loss": 5.9056,
+      "learning_rate": 0.0008769579554822754,
+      "loss": 7.8698,
       "step": 1100
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002630255564715581,
-      "loss": 4.7588,
+      "learning_rate": 0.0008767518549051937,
+      "loss": 5.7204,
       "step": 1101
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002629637262984336,
-      "loss": 4.9718,
+      "learning_rate": 0.0008765457543281121,
+      "loss": 5.7686,
       "step": 1102
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002629018961253091,
-      "loss": 4.9324,
+      "learning_rate": 0.0008763396537510305,
+      "loss": 5.7966,
       "step": 1103
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.00026284006595218466,
-      "loss": 4.5766,
+      "learning_rate": 0.0008761335531739489,
+      "loss": 5.5083,
       "step": 1104
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.00026277823577906015,
-      "loss": 4.6335,
+      "learning_rate": 0.0008759274525968672,
+      "loss": 5.5322,
       "step": 1105
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002627164056059357,
-      "loss": 4.5401,
+      "learning_rate": 0.0008757213520197856,
+      "loss": 5.2557,
       "step": 1106
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002626545754328112,
-      "loss": 4.7469,
+      "learning_rate": 0.000875515251442704,
+      "loss": 5.7281,
       "step": 1107
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002625927452596867,
-      "loss": 4.5542,
+      "learning_rate": 0.0008753091508656224,
+      "loss": 5.4922,
       "step": 1108
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002625309150865622,
-      "loss": 4.8292,
+      "learning_rate": 0.0008751030502885409,
+      "loss": 5.5839,
       "step": 1109
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002624690849134377,
-      "loss": 5.0009,
+      "learning_rate": 0.0008748969497114591,
+      "loss": 5.6359,
       "step": 1110
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.00026240725474031325,
-      "loss": 4.8606,
+      "learning_rate": 0.0008746908491343776,
+      "loss": 5.2382,
       "step": 1111
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002623454245671888,
-      "loss": 4.9664,
+      "learning_rate": 0.000874484748557296,
+      "loss": 5.6426,
       "step": 1112
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002622835943940643,
-      "loss": 5.2102,
+      "learning_rate": 0.0008742786479802144,
+      "loss": 6.0734,
       "step": 1113
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002622217642209398,
-      "loss": 5.185,
+      "learning_rate": 0.0008740725474031328,
+      "loss": 6.0719,
       "step": 1114
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0002621599340478153,
-      "loss": 5.0649,
+      "learning_rate": 0.0008738664468260511,
+      "loss": 5.948,
       "step": 1115
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002620981038746908,
-      "loss": 4.9034,
+      "learning_rate": 0.0008736603462489696,
+      "loss": 5.7175,
       "step": 1116
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00026203627370156634,
-      "loss": 5.081,
+      "learning_rate": 0.000873454245671888,
+      "loss": 5.5157,
       "step": 1117
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002619744435284419,
-      "loss": 5.2825,
+      "learning_rate": 0.0008732481450948064,
+      "loss": 5.6237,
       "step": 1118
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002619126133553174,
-      "loss": 4.8339,
+      "learning_rate": 0.0008730420445177247,
+      "loss": 5.6129,
       "step": 1119
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00026185078318219287,
-      "loss": 4.9587,
+      "learning_rate": 0.0008728359439406431,
+      "loss": 5.8905,
       "step": 1120
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002617889530090684,
-      "loss": 4.8759,
+      "learning_rate": 0.0008726298433635615,
+      "loss": 5.6367,
       "step": 1121
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002617271228359439,
-      "loss": 4.8898,
+      "learning_rate": 0.0008724237427864799,
+      "loss": 5.5964,
       "step": 1122
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00026166529266281944,
-      "loss": 4.8448,
+      "learning_rate": 0.0008722176422093983,
+      "loss": 5.7005,
       "step": 1123
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00026160346248969493,
-      "loss": 4.558,
+      "learning_rate": 0.0008720115416323166,
+      "loss": 5.3189,
       "step": 1124
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002615416323165705,
-      "loss": 5.3784,
+      "learning_rate": 0.000871805441055235,
+      "loss": 5.712,
       "step": 1125
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00026147980214344597,
-      "loss": 4.9086,
+      "learning_rate": 0.0008715993404781534,
+      "loss": 5.4733,
       "step": 1126
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002614179719703215,
-      "loss": 4.8476,
+      "learning_rate": 0.0008713932399010718,
+      "loss": 5.2996,
       "step": 1127
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.000261356141797197,
-      "loss": 5.1946,
+      "learning_rate": 0.0008711871393239901,
+      "loss": 6.1,
       "step": 1128
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002612943116240725,
-      "loss": 4.4185,
+      "learning_rate": 0.0008709810387469085,
+      "loss": 5.1543,
       "step": 1129
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00026123248145094803,
-      "loss": 5.4736,
+      "learning_rate": 0.0008707749381698269,
+      "loss": 6.7132,
       "step": 1130
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0002611706512778236,
-      "loss": 4.9924,
+      "learning_rate": 0.0008705688375927453,
+      "loss": 5.7305,
       "step": 1131
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.00026110882110469907,
-      "loss": 4.8013,
+      "learning_rate": 0.0008703627370156637,
+      "loss": 6.1443,
       "step": 1132
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026104699093157456,
-      "loss": 5.0842,
+      "learning_rate": 0.000870156636438582,
+      "loss": 6.1438,
       "step": 1133
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002609851607584501,
-      "loss": 4.8512,
+      "learning_rate": 0.0008699505358615004,
+      "loss": 5.527,
       "step": 1134
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026092333058532564,
-      "loss": 5.2553,
+      "learning_rate": 0.0008697444352844188,
+      "loss": 6.0826,
       "step": 1135
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026086150041220113,
-      "loss": 4.7254,
+      "learning_rate": 0.0008695383347073373,
+      "loss": 5.2393,
       "step": 1136
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002607996702390766,
-      "loss": 5.0971,
+      "learning_rate": 0.0008693322341302556,
+      "loss": 6.2021,
       "step": 1137
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026073784006595217,
-      "loss": 4.8275,
+      "learning_rate": 0.000869126133553174,
+      "loss": 5.5513,
       "step": 1138
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026067600989282766,
-      "loss": 4.5282,
+      "learning_rate": 0.0008689200329760924,
+      "loss": 5.335,
       "step": 1139
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002606141797197032,
-      "loss": 4.7209,
+      "learning_rate": 0.0008687139323990108,
+      "loss": 5.3823,
       "step": 1140
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026055234954657874,
-      "loss": 4.7448,
+      "learning_rate": 0.0008685078318219292,
+      "loss": 5.5602,
       "step": 1141
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026049051937345423,
-      "loss": 5.1471,
+      "learning_rate": 0.0008683017312448475,
+      "loss": 6.554,
       "step": 1142
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002604286892003297,
-      "loss": 5.4783,
+      "learning_rate": 0.0008680956306677659,
+      "loss": 6.7542,
       "step": 1143
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026036685902720527,
-      "loss": 5.0044,
+      "learning_rate": 0.0008678895300906843,
+      "loss": 5.7537,
       "step": 1144
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026030502885408075,
-      "loss": 4.6673,
+      "learning_rate": 0.0008676834295136027,
+      "loss": 5.3978,
       "step": 1145
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002602431986809563,
-      "loss": 4.9502,
+      "learning_rate": 0.0008674773289365211,
+      "loss": 5.555,
       "step": 1146
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002601813685078318,
-      "loss": 4.8499,
+      "learning_rate": 0.0008672712283594394,
+      "loss": 5.6886,
       "step": 1147
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00026011953833470733,
-      "loss": 4.9898,
+      "learning_rate": 0.0008670651277823578,
+      "loss": 5.4934,
       "step": 1148
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0002600577081615828,
-      "loss": 7.0503,
+      "learning_rate": 0.0008668590272052762,
+      "loss": 6.3331,
       "step": 1149
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.00025999587798845836,
-      "loss": 4.7141,
+      "learning_rate": 0.0008666529266281946,
+      "loss": 5.794,
       "step": 1150
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.00025993404781533385,
-      "loss": 4.8496,
+      "learning_rate": 0.0008664468260511129,
+      "loss": 5.789,
       "step": 1151
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.00025987221764220934,
-      "loss": 5.1362,
+      "learning_rate": 0.0008662407254740313,
+      "loss": 5.9289,
       "step": 1152
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002598103874690849,
-      "loss": 4.7819,
+      "learning_rate": 0.0008660346248969497,
+      "loss": 5.6205,
       "step": 1153
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.00025974855729596043,
-      "loss": 4.8339,
+      "learning_rate": 0.0008658285243198681,
+      "loss": 5.849,
       "step": 1154
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002596867271228359,
-      "loss": 4.9788,
+      "learning_rate": 0.0008656224237427865,
+      "loss": 5.9778,
       "step": 1155
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002596248969497114,
-      "loss": 5.0352,
+      "learning_rate": 0.0008654163231657048,
+      "loss": 5.5916,
       "step": 1156
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.00025956306677658695,
-      "loss": 4.7121,
+      "learning_rate": 0.0008652102225886233,
+      "loss": 5.5011,
       "step": 1157
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.00025950123660346244,
-      "loss": 5.0317,
+      "learning_rate": 0.0008650041220115417,
+      "loss": 5.8971,
       "step": 1158
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.000259439406430338,
-      "loss": 4.5576,
+      "learning_rate": 0.0008647980214344601,
+      "loss": 5.3912,
       "step": 1159
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002593775762572135,
-      "loss": 4.9529,
+      "learning_rate": 0.0008645919208573784,
+      "loss": 5.3985,
       "step": 1160
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.000259315746084089,
-      "loss": 4.9488,
+      "learning_rate": 0.0008643858202802968,
+      "loss": 5.5576,
       "step": 1161
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002592539159109645,
-      "loss": 4.6826,
+      "learning_rate": 0.0008641797197032152,
+      "loss": 5.3502,
       "step": 1162
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.00025919208573784005,
-      "loss": 5.0933,
+      "learning_rate": 0.0008639736191261336,
+      "loss": 5.4327,
       "step": 1163
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002591302555647156,
-      "loss": 4.5428,
+      "learning_rate": 0.000863767518549052,
+      "loss": 5.2577,
       "step": 1164
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002590684253915911,
-      "loss": 4.7753,
+      "learning_rate": 0.0008635614179719703,
+      "loss": 5.3807,
       "step": 1165
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002590065952184666,
-      "loss": 4.9511,
+      "learning_rate": 0.0008633553173948887,
+      "loss": 5.5532,
       "step": 1166
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002589447650453421,
-      "loss": 5.023,
+      "learning_rate": 0.0008631492168178071,
+      "loss": 5.5996,
       "step": 1167
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0002588829348722176,
-      "loss": 4.7067,
+      "learning_rate": 0.0008629431162407255,
+      "loss": 4.9879,
       "step": 1168
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00025882110469909315,
-      "loss": 4.9157,
+      "learning_rate": 0.0008627370156636439,
+      "loss": 5.4891,
       "step": 1169
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00025875927452596864,
-      "loss": 5.0776,
+      "learning_rate": 0.0008625309150865622,
+      "loss": 5.8581,
       "step": 1170
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002586974443528442,
-      "loss": 5.353,
+      "learning_rate": 0.0008623248145094806,
+      "loss": 6.013,
       "step": 1171
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002586356141797197,
-      "loss": 4.4564,
+      "learning_rate": 0.000862118713932399,
+      "loss": 5.1599,
       "step": 1172
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002585737840065952,
-      "loss": 5.1127,
+      "learning_rate": 0.0008619126133553174,
+      "loss": 5.8124,
       "step": 1173
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002585119538334707,
-      "loss": 4.6252,
+      "learning_rate": 0.0008617065127782357,
+      "loss": 5.2817,
       "step": 1174
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002584501236603462,
-      "loss": 4.92,
+      "learning_rate": 0.0008615004122011541,
+      "loss": 6.0596,
       "step": 1175
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00025838829348722174,
-      "loss": 5.6644,
+      "learning_rate": 0.0008612943116240725,
+      "loss": 6.4745,
       "step": 1176
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002583264633140973,
-      "loss": 5.2197,
+      "learning_rate": 0.000861088211046991,
+      "loss": 5.5368,
       "step": 1177
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002582646331409728,
-      "loss": 5.0792,
+      "learning_rate": 0.0008608821104699094,
+      "loss": 5.5205,
       "step": 1178
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00025820280296784826,
-      "loss": 4.544,
+      "learning_rate": 0.0008606760098928277,
+      "loss": 5.3163,
       "step": 1179
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002581409727947238,
-      "loss": 4.789,
+      "learning_rate": 0.0008604699093157461,
+      "loss": 5.5853,
       "step": 1180
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002580791426215993,
-      "loss": 4.6347,
+      "learning_rate": 0.0008602638087386645,
+      "loss": 5.4125,
       "step": 1181
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00025801731244847484,
-      "loss": 5.0942,
+      "learning_rate": 0.0008600577081615829,
+      "loss": 5.9092,
       "step": 1182
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00025795548227535033,
-      "loss": 5.0565,
+      "learning_rate": 0.0008598516075845012,
+      "loss": 5.8494,
       "step": 1183
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002578936521022259,
-      "loss": 4.5785,
+      "learning_rate": 0.0008596455070074196,
+      "loss": 5.4406,
       "step": 1184
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.00025783182192910136,
-      "loss": 5.3616,
+      "learning_rate": 0.000859439406430338,
+      "loss": 6.0505,
       "step": 1185
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0002577699917559769,
-      "loss": 4.6774,
+      "learning_rate": 0.0008592333058532564,
+      "loss": 5.8445,
       "step": 1186
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0002577081615828524,
-      "loss": 5.2014,
+      "learning_rate": 0.0008590272052761748,
+      "loss": 5.8123,
       "step": 1187
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0002576463314097279,
-      "loss": 4.8648,
+      "learning_rate": 0.0008588211046990931,
+      "loss": 5.5666,
       "step": 1188
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025758450123660343,
-      "loss": 4.9073,
+      "learning_rate": 0.0008586150041220115,
+      "loss": 5.7071,
       "step": 1189
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.000257522671063479,
-      "loss": 4.8956,
+      "learning_rate": 0.0008584089035449299,
+      "loss": 6.5352,
       "step": 1190
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025746084089035446,
-      "loss": 4.9127,
+      "learning_rate": 0.0008582028029678483,
+      "loss": 5.4081,
       "step": 1191
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025739901071722995,
-      "loss": 5.1001,
+      "learning_rate": 0.0008579967023907666,
+      "loss": 6.5511,
       "step": 1192
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0002573371805441055,
-      "loss": 5.8124,
+      "learning_rate": 0.000857790601813685,
+      "loss": 5.934,
       "step": 1193
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025727535037098104,
-      "loss": 4.7253,
+      "learning_rate": 0.0008575845012366034,
+      "loss": 5.5478,
       "step": 1194
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025721352019785653,
-      "loss": 4.8271,
+      "learning_rate": 0.0008573784006595218,
+      "loss": 5.4186,
       "step": 1195
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0002571516900247321,
-      "loss": 4.4595,
+      "learning_rate": 0.0008571723000824402,
+      "loss": 5.3277,
       "step": 1196
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025708985985160756,
-      "loss": 4.8535,
+      "learning_rate": 0.0008569661995053585,
+      "loss": 5.6445,
       "step": 1197
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025702802967848305,
-      "loss": 4.9232,
+      "learning_rate": 0.000856760098928277,
+      "loss": 5.5742,
       "step": 1198
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0002569661995053586,
-      "loss": 4.8364,
+      "learning_rate": 0.0008565539983511954,
+      "loss": 5.4332,
       "step": 1199
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025690436933223414,
-      "loss": 4.3532,
+      "learning_rate": 0.0008563478977741138,
+      "loss": 5.0615,
       "step": 1200
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025684253915910963,
-      "loss": 4.9009,
+      "learning_rate": 0.0008561417971970322,
+      "loss": 5.667,
       "step": 1201
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0002567807089859851,
-      "loss": 4.8128,
+      "learning_rate": 0.0008559356966199505,
+      "loss": 5.4229,
       "step": 1202
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025671887881286066,
-      "loss": 4.8216,
+      "learning_rate": 0.000855729596042869,
+      "loss": 5.3872,
       "step": 1203
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.00025665704863973615,
-      "loss": 4.6244,
+      "learning_rate": 0.0008555234954657874,
+      "loss": 5.3896,
       "step": 1204
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002565952184666117,
-      "loss": 4.3943,
+      "learning_rate": 0.0008553173948887058,
+      "loss": 5.1914,
       "step": 1205
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002565333882934872,
-      "loss": 4.8696,
+      "learning_rate": 0.0008551112943116241,
+      "loss": 5.4791,
       "step": 1206
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025647155812036273,
-      "loss": 5.2853,
+      "learning_rate": 0.0008549051937345425,
+      "loss": 7.4053,
       "step": 1207
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002564097279472382,
-      "loss": 4.8942,
+      "learning_rate": 0.0008546990931574609,
+      "loss": 5.8224,
       "step": 1208
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025634789777411376,
-      "loss": 4.9078,
+      "learning_rate": 0.0008544929925803793,
+      "loss": 6.301,
       "step": 1209
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025628606760098925,
-      "loss": 4.8809,
+      "learning_rate": 0.0008542868920032977,
+      "loss": 5.4241,
       "step": 1210
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025622423742786474,
-      "loss": 5.0549,
+      "learning_rate": 0.000854080791426216,
+      "loss": 5.7755,
       "step": 1211
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002561624072547403,
-      "loss": 5.1304,
+      "learning_rate": 0.0008538746908491344,
+      "loss": 5.9801,
       "step": 1212
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025610057708161583,
-      "loss": 4.8621,
+      "learning_rate": 0.0008536685902720528,
+      "loss": 5.5397,
       "step": 1213
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002560387469084913,
-      "loss": 4.7155,
+      "learning_rate": 0.0008534624896949712,
+      "loss": 5.7723,
       "step": 1214
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002559769167353668,
-      "loss": 4.6437,
+      "learning_rate": 0.0008532563891178895,
+      "loss": 5.385,
       "step": 1215
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025591508656224235,
-      "loss": 4.811,
+      "learning_rate": 0.0008530502885408079,
+      "loss": 5.6722,
       "step": 1216
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025585325638911784,
-      "loss": 4.8233,
+      "learning_rate": 0.0008528441879637264,
+      "loss": 5.434,
       "step": 1217
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002557914262159934,
-      "loss": 4.3894,
+      "learning_rate": 0.0008526380873866448,
+      "loss": 5.1889,
       "step": 1218
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025572959604286893,
-      "loss": 4.6665,
+      "learning_rate": 0.0008524319868095632,
+      "loss": 5.3302,
       "step": 1219
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002556677658697444,
-      "loss": 5.0872,
+      "learning_rate": 0.0008522258862324815,
+      "loss": 5.6648,
       "step": 1220
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0002556059356966199,
-      "loss": 4.7209,
+      "learning_rate": 0.0008520197856553999,
+      "loss": 5.4091,
       "step": 1221
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.00025554410552349545,
-      "loss": 4.882,
+      "learning_rate": 0.0008518136850783183,
+      "loss": 6.3325,
       "step": 1222
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.000255482275350371,
-      "loss": 4.8798,
+      "learning_rate": 0.0008516075845012367,
+      "loss": 5.5858,
       "step": 1223
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002554204451772465,
-      "loss": 5.0543,
+      "learning_rate": 0.000851401483924155,
+      "loss": 6.14,
       "step": 1224
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.000255358615004122,
-      "loss": 5.241,
+      "learning_rate": 0.0008511953833470734,
+      "loss": 6.4742,
       "step": 1225
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002552967848309975,
-      "loss": 4.8791,
+      "learning_rate": 0.0008509892827699918,
+      "loss": 5.4455,
       "step": 1226
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.000255234954657873,
-      "loss": 4.6573,
+      "learning_rate": 0.0008507831821929102,
+      "loss": 5.4558,
       "step": 1227
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.00025517312448474855,
-      "loss": 4.6891,
+      "learning_rate": 0.0008505770816158286,
+      "loss": 5.6804,
       "step": 1228
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.00025511129431162404,
-      "loss": 4.6513,
+      "learning_rate": 0.0008503709810387469,
+      "loss": 5.3473,
       "step": 1229
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002550494641384996,
-      "loss": 5.6629,
+      "learning_rate": 0.0008501648804616653,
+      "loss": 6.164,
       "step": 1230
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.00025498763396537507,
-      "loss": 4.9699,
+      "learning_rate": 0.0008499587798845837,
+      "loss": 5.4779,
       "step": 1231
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002549258037922506,
-      "loss": 5.6454,
+      "learning_rate": 0.0008497526793075021,
+      "loss": 6.1431,
       "step": 1232
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002548639736191261,
-      "loss": 4.8326,
+      "learning_rate": 0.0008495465787304205,
+      "loss": 5.6048,
       "step": 1233
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002548021434460016,
-      "loss": 4.6138,
+      "learning_rate": 0.0008493404781533388,
+      "loss": 5.502,
       "step": 1234
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.00025474031327287714,
-      "loss": 5.1294,
+      "learning_rate": 0.0008491343775762572,
+      "loss": 5.7898,
       "step": 1235
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002546784830997527,
-      "loss": 4.9788,
+      "learning_rate": 0.0008489282769991756,
+      "loss": 5.3421,
       "step": 1236
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.00025461665292662817,
-      "loss": 5.3763,
+      "learning_rate": 0.000848722176422094,
+      "loss": 5.9001,
       "step": 1237
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.00025455482275350366,
-      "loss": 4.5893,
+      "learning_rate": 0.0008485160758450124,
+      "loss": 5.4223,
       "step": 1238
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0002544929925803792,
-      "loss": 5.0554,
+      "learning_rate": 0.0008483099752679308,
+      "loss": 5.9017,
       "step": 1239
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0002544311624072547,
-      "loss": 5.0587,
+      "learning_rate": 0.0008481038746908492,
+      "loss": 6.7111,
       "step": 1240
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025436933223413024,
-      "loss": 4.7333,
+      "learning_rate": 0.0008478977741137676,
+      "loss": 5.745,
       "step": 1241
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0002543075020610058,
-      "loss": 5.1705,
+      "learning_rate": 0.000847691673536686,
+      "loss": 6.1242,
       "step": 1242
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025424567188788127,
-      "loss": 5.2216,
+      "learning_rate": 0.0008474855729596043,
+      "loss": 5.5578,
       "step": 1243
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025418384171475676,
-      "loss": 4.6861,
+      "learning_rate": 0.0008472794723825227,
+      "loss": 5.5023,
       "step": 1244
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0002541220115416323,
-      "loss": 4.4021,
+      "learning_rate": 0.0008470733718054411,
+      "loss": 5.1116,
       "step": 1245
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0002540601813685078,
-      "loss": 5.0288,
+      "learning_rate": 0.0008468672712283595,
+      "loss": 5.657,
       "step": 1246
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0002539983511953833,
-      "loss": 4.8656,
+      "learning_rate": 0.0008466611706512778,
+      "loss": 5.6068,
       "step": 1247
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025393652102225883,
-      "loss": 5.0229,
+      "learning_rate": 0.0008464550700741962,
+      "loss": 5.4356,
       "step": 1248
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025387469084913437,
-      "loss": 4.8331,
+      "learning_rate": 0.0008462489694971146,
+      "loss": 5.6828,
       "step": 1249
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025381286067600986,
-      "loss": 5.0085,
+      "learning_rate": 0.000846042868920033,
+      "loss": 5.6051,
       "step": 1250
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0002537510305028854,
-      "loss": 4.8919,
+      "learning_rate": 0.0008458367683429514,
+      "loss": 6.0235,
       "step": 1251
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0002536892003297609,
-      "loss": 5.5628,
+      "learning_rate": 0.0008456306677658697,
+      "loss": 5.8462,
       "step": 1252
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025362737015663644,
-      "loss": 4.9047,
+      "learning_rate": 0.0008454245671887881,
+      "loss": 5.7775,
       "step": 1253
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025356553998351193,
-      "loss": 4.85,
+      "learning_rate": 0.0008452184666117065,
+      "loss": 5.587,
       "step": 1254
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025350370981038747,
-      "loss": 5.2581,
+      "learning_rate": 0.0008450123660346249,
+      "loss": 6.2628,
       "step": 1255
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025344187963726296,
-      "loss": 4.6677,
+      "learning_rate": 0.0008448062654575432,
+      "loss": 6.2243,
       "step": 1256
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.00025338004946413845,
-      "loss": 4.6704,
+      "learning_rate": 0.0008446001648804616,
+      "loss": 5.3067,
       "step": 1257
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.000253318219291014,
-      "loss": 4.5792,
+      "learning_rate": 0.00084439406430338,
+      "loss": 5.3502,
       "step": 1258
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025325638911788954,
-      "loss": 4.9731,
+      "learning_rate": 0.0008441879637262985,
+      "loss": 5.4845,
       "step": 1259
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.000253194558944765,
-      "loss": 5.2371,
+      "learning_rate": 0.0008439818631492169,
+      "loss": 5.8301,
       "step": 1260
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002531327287716405,
-      "loss": 4.8475,
+      "learning_rate": 0.0008437757625721352,
+      "loss": 5.5284,
       "step": 1261
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025307089859851606,
-      "loss": 4.9762,
+      "learning_rate": 0.0008435696619950536,
+      "loss": 5.519,
       "step": 1262
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025300906842539155,
-      "loss": 4.8224,
+      "learning_rate": 0.000843363561417972,
+      "loss": 5.3825,
       "step": 1263
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002529472382522671,
-      "loss": 4.7602,
+      "learning_rate": 0.0008431574608408904,
+      "loss": 5.3299,
       "step": 1264
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025288540807914264,
-      "loss": 5.4473,
+      "learning_rate": 0.0008429513602638088,
+      "loss": 7.3644,
       "step": 1265
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002528235779060181,
-      "loss": 4.9109,
+      "learning_rate": 0.0008427452596867271,
+      "loss": 6.1926,
       "step": 1266
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002527617477328936,
-      "loss": 5.342,
+      "learning_rate": 0.0008425391591096455,
+      "loss": 5.5954,
       "step": 1267
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025269991755976916,
-      "loss": 4.8607,
+      "learning_rate": 0.0008423330585325639,
+      "loss": 5.3388,
       "step": 1268
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025263808738664465,
-      "loss": 4.9263,
+      "learning_rate": 0.0008421269579554823,
+      "loss": 5.4468,
       "step": 1269
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025257625721352014,
-      "loss": 4.5743,
+      "learning_rate": 0.0008419208573784006,
+      "loss": 5.3038,
       "step": 1270
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002525144270403957,
-      "loss": 4.9887,
+      "learning_rate": 0.000841714756801319,
+      "loss": 5.9022,
       "step": 1271
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002524525968672712,
-      "loss": 4.8784,
+      "learning_rate": 0.0008415086562242374,
+      "loss": 5.396,
       "step": 1272
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0002523907666941467,
-      "loss": 5.0411,
+      "learning_rate": 0.0008413025556471558,
+      "loss": 5.5037,
       "step": 1273
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025232893652102226,
-      "loss": 4.6067,
+      "learning_rate": 0.0008410964550700742,
+      "loss": 5.4653,
       "step": 1274
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00025226710634789775,
-      "loss": 4.711,
+      "learning_rate": 0.0008408903544929925,
+      "loss": 5.4283,
       "step": 1275
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00025220527617477324,
-      "loss": 5.4213,
+      "learning_rate": 0.0008406842539159109,
+      "loss": 5.8997,
       "step": 1276
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002521434460016488,
-      "loss": 4.9152,
+      "learning_rate": 0.0008404781533388293,
+      "loss": 5.6983,
       "step": 1277
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002520816158285243,
-      "loss": 4.9507,
+      "learning_rate": 0.0008402720527617478,
+      "loss": 5.5442,
       "step": 1278
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002520197856553998,
-      "loss": 5.0109,
+      "learning_rate": 0.000840065952184666,
+      "loss": 5.775,
       "step": 1279
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002519579554822753,
-      "loss": 4.8066,
+      "learning_rate": 0.0008398598516075845,
+      "loss": 5.9126,
       "step": 1280
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00025189612530915085,
-      "loss": 4.658,
+      "learning_rate": 0.0008396537510305029,
+      "loss": 5.3171,
       "step": 1281
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002518342951360264,
-      "loss": 4.8252,
+      "learning_rate": 0.0008394476504534213,
+      "loss": 5.3271,
       "step": 1282
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002517724649629019,
-      "loss": 5.1631,
+      "learning_rate": 0.0008392415498763397,
+      "loss": 6.0536,
       "step": 1283
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00025171063478977737,
-      "loss": 4.4933,
+      "learning_rate": 0.000839035449299258,
+      "loss": 5.394,
       "step": 1284
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002516488046166529,
-      "loss": 4.5925,
+      "learning_rate": 0.0008388293487221764,
+      "loss": 5.3376,
       "step": 1285
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002515869744435284,
-      "loss": 4.7392,
+      "learning_rate": 0.0008386232481450948,
+      "loss": 5.5803,
       "step": 1286
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00025152514427040395,
-      "loss": 4.9272,
+      "learning_rate": 0.0008384171475680132,
+      "loss": 5.706,
       "step": 1287
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002514633140972795,
-      "loss": 4.6686,
+      "learning_rate": 0.0008382110469909316,
+      "loss": 5.331,
       "step": 1288
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.000251401483924155,
-      "loss": 4.7057,
+      "learning_rate": 0.0008380049464138499,
+      "loss": 5.4349,
       "step": 1289
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.00025133965375103047,
-      "loss": 4.6957,
+      "learning_rate": 0.0008377988458367683,
+      "loss": 5.5164,
       "step": 1290
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.000251277823577906,
-      "loss": 4.9638,
+      "learning_rate": 0.0008375927452596867,
+      "loss": 5.5294,
       "step": 1291
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0002512159934047815,
-      "loss": 5.4854,
+      "learning_rate": 0.0008373866446826051,
+      "loss": 7.0537,
       "step": 1292
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.000251154163231657,
-      "loss": 6.5952,
+      "learning_rate": 0.0008371805441055234,
+      "loss": 7.4283,
       "step": 1293
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025109233305853254,
-      "loss": 4.6575,
+      "learning_rate": 0.0008369744435284418,
+      "loss": 5.5964,
       "step": 1294
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002510305028854081,
-      "loss": 4.4765,
+      "learning_rate": 0.0008367683429513602,
+      "loss": 5.1711,
       "step": 1295
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025096867271228357,
-      "loss": 4.4329,
+      "learning_rate": 0.0008365622423742787,
+      "loss": 5.1941,
       "step": 1296
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002509068425391591,
-      "loss": 4.3552,
+      "learning_rate": 0.0008363561417971972,
+      "loss": 5.1003,
       "step": 1297
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002508450123660346,
-      "loss": 4.1729,
+      "learning_rate": 0.0008361500412201155,
+      "loss": 4.8134,
       "step": 1298
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002507831821929101,
-      "loss": 6.439,
+      "learning_rate": 0.0008359439406430339,
+      "loss": 7.1477,
       "step": 1299
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025072135201978564,
-      "loss": 4.3167,
+      "learning_rate": 0.0008357378400659523,
+      "loss": 4.9195,
       "step": 1300
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002506595218466612,
-      "loss": 5.7774,
+      "learning_rate": 0.0008355317394888707,
+      "loss": 6.711,
       "step": 1301
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025059769167353667,
-      "loss": 5.3302,
+      "learning_rate": 0.000835325638911789,
+      "loss": 6.8379,
       "step": 1302
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025053586150041216,
-      "loss": 4.9756,
+      "learning_rate": 0.0008351195383347074,
+      "loss": 5.6529,
       "step": 1303
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002504740313272877,
-      "loss": 5.2556,
+      "learning_rate": 0.0008349134377576258,
+      "loss": 5.798,
       "step": 1304
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002504122011541632,
-      "loss": 4.7331,
+      "learning_rate": 0.0008347073371805442,
+      "loss": 5.51,
       "step": 1305
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025035037098103874,
-      "loss": 4.8549,
+      "learning_rate": 0.0008345012366034626,
+      "loss": 5.6212,
       "step": 1306
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002502885408079142,
-      "loss": 5.0353,
+      "learning_rate": 0.0008342951360263809,
+      "loss": 6.1351,
       "step": 1307
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025022671063478977,
-      "loss": 4.5423,
+      "learning_rate": 0.0008340890354492993,
+      "loss": 5.3197,
       "step": 1308
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.00025016488046166526,
-      "loss": 5.1334,
+      "learning_rate": 0.0008338829348722177,
+      "loss": 7.2568,
       "step": 1309
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002501030502885408,
-      "loss": 4.5877,
+      "learning_rate": 0.0008336768342951361,
+      "loss": 5.3516,
       "step": 1310
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0002500412201154163,
-      "loss": 4.6783,
+      "learning_rate": 0.0008334707337180544,
+      "loss": 5.2981,
       "step": 1311
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0002499793899422918,
-      "loss": 4.9442,
+      "learning_rate": 0.0008332646331409728,
+      "loss": 5.6216,
       "step": 1312
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0002499175597691673,
-      "loss": 4.588,
+      "learning_rate": 0.0008330585325638912,
+      "loss": 5.3826,
       "step": 1313
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024985572959604287,
-      "loss": 4.7092,
+      "learning_rate": 0.0008328524319868096,
+      "loss": 5.1565,
       "step": 1314
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024979389942291836,
-      "loss": 5.0367,
+      "learning_rate": 0.000832646331409728,
+      "loss": 5.8199,
       "step": 1315
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024973206924979385,
-      "loss": 5.2309,
+      "learning_rate": 0.0008324402308326463,
+      "loss": 5.6888,
       "step": 1316
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0002496702390766694,
-      "loss": 4.5255,
+      "learning_rate": 0.0008322341302555647,
+      "loss": 5.1787,
       "step": 1317
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024960840890354493,
-      "loss": 5.7602,
+      "learning_rate": 0.0008320280296784832,
+      "loss": 5.7905,
       "step": 1318
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0002495465787304204,
-      "loss": 4.6042,
+      "learning_rate": 0.0008318219291014016,
+      "loss": 5.3534,
       "step": 1319
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024948474855729597,
-      "loss": 4.3831,
+      "learning_rate": 0.00083161582852432,
+      "loss": 5.2817,
       "step": 1320
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024942291838417146,
-      "loss": 4.3564,
+      "learning_rate": 0.0008314097279472383,
+      "loss": 5.063,
       "step": 1321
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024936108821104695,
-      "loss": 4.568,
+      "learning_rate": 0.0008312036273701567,
+      "loss": 5.2679,
       "step": 1322
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0002492992580379225,
-      "loss": 4.7763,
+      "learning_rate": 0.0008309975267930751,
+      "loss": 5.5354,
       "step": 1323
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024923742786479803,
-      "loss": 4.6934,
+      "learning_rate": 0.0008307914262159935,
+      "loss": 5.7153,
       "step": 1324
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0002491755976916735,
-      "loss": 5.0292,
+      "learning_rate": 0.0008305853256389118,
+      "loss": 5.6569,
       "step": 1325
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.000249113767518549,
-      "loss": 4.6495,
+      "learning_rate": 0.0008303792250618302,
+      "loss": 5.5469,
       "step": 1326
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024905193734542456,
-      "loss": 4.5958,
+      "learning_rate": 0.0008301731244847486,
+      "loss": 5.3349,
       "step": 1327
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00024899010717230005,
-      "loss": 5.0425,
+      "learning_rate": 0.000829967023907667,
+      "loss": 5.8347,
       "step": 1328
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0002489282769991756,
-      "loss": 4.97,
+      "learning_rate": 0.0008297609233305854,
+      "loss": 5.6732,
       "step": 1329
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002488664468260511,
-      "loss": 5.1066,
+      "learning_rate": 0.0008295548227535037,
+      "loss": 5.7834,
       "step": 1330
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002488046166529266,
-      "loss": 4.5578,
+      "learning_rate": 0.0008293487221764221,
+      "loss": 5.1808,
       "step": 1331
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002487427864798021,
-      "loss": 5.1355,
+      "learning_rate": 0.0008291426215993405,
+      "loss": 5.8839,
       "step": 1332
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00024868095630667766,
-      "loss": 4.7361,
+      "learning_rate": 0.0008289365210222589,
+      "loss": 5.8086,
       "step": 1333
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00024861912613355315,
-      "loss": 5.0657,
+      "learning_rate": 0.0008287304204451772,
+      "loss": 5.5766,
       "step": 1334
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00024855729596042864,
-      "loss": 4.5594,
+      "learning_rate": 0.0008285243198680956,
+      "loss": 5.2663,
       "step": 1335
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002484954657873042,
-      "loss": 4.841,
+      "learning_rate": 0.000828318219291014,
+      "loss": 5.3895,
       "step": 1336
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002484336356141797,
-      "loss": 4.9102,
+      "learning_rate": 0.0008281121187139324,
+      "loss": 5.5717,
       "step": 1337
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002483718054410552,
-      "loss": 4.8341,
+      "learning_rate": 0.0008279060181368509,
+      "loss": 5.3612,
       "step": 1338
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002483099752679307,
-      "loss": 5.0388,
+      "learning_rate": 0.0008276999175597691,
+      "loss": 5.6131,
       "step": 1339
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00024824814509480625,
-      "loss": 6.5249,
+      "learning_rate": 0.0008274938169826876,
+      "loss": 6.4087,
       "step": 1340
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00024818631492168173,
-      "loss": 4.9602,
+      "learning_rate": 0.000827287716405606,
+      "loss": 5.7496,
       "step": 1341
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002481244847485573,
-      "loss": 4.9535,
+      "learning_rate": 0.0008270816158285244,
+      "loss": 5.8622,
       "step": 1342
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00024806265457543277,
-      "loss": 4.9383,
+      "learning_rate": 0.0008268755152514427,
+      "loss": 5.7602,
       "step": 1343
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002480008244023083,
-      "loss": 5.512,
+      "learning_rate": 0.0008266694146743611,
+      "loss": 5.5818,
       "step": 1344
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0002479389942291838,
-      "loss": 5.0725,
+      "learning_rate": 0.0008264633140972795,
+      "loss": 6.1063,
       "step": 1345
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.00024787716405605934,
-      "loss": 4.8216,
+      "learning_rate": 0.0008262572135201979,
+      "loss": 5.5154,
       "step": 1346
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002478153338829349,
-      "loss": 5.8929,
+      "learning_rate": 0.0008260511129431163,
+      "loss": 6.4484,
       "step": 1347
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002477535037098104,
-      "loss": 4.5771,
+      "learning_rate": 0.0008258450123660346,
+      "loss": 5.1804,
       "step": 1348
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024769167353668587,
-      "loss": 4.8688,
+      "learning_rate": 0.000825638911788953,
+      "loss": 6.3442,
       "step": 1349
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002476298433635614,
-      "loss": 4.6368,
+      "learning_rate": 0.0008254328112118714,
+      "loss": 5.1424,
       "step": 1350
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002475680131904369,
-      "loss": 4.931,
+      "learning_rate": 0.0008252267106347898,
+      "loss": 5.8901,
       "step": 1351
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024750618301731244,
-      "loss": 5.0968,
+      "learning_rate": 0.0008250206100577082,
+      "loss": 5.666,
       "step": 1352
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024744435284418793,
-      "loss": 5.2642,
+      "learning_rate": 0.0008248145094806265,
+      "loss": 5.6649,
       "step": 1353
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002473825226710635,
-      "loss": 5.1734,
+      "learning_rate": 0.0008246084089035449,
+      "loss": 5.9874,
       "step": 1354
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024732069249793897,
-      "loss": 4.575,
+      "learning_rate": 0.0008244023083264633,
+      "loss": 5.3731,
       "step": 1355
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002472588623248145,
-      "loss": 5.3882,
+      "learning_rate": 0.0008241962077493817,
+      "loss": 5.7902,
       "step": 1356
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024719703215169,
-      "loss": 4.589,
+      "learning_rate": 0.0008239901071723,
+      "loss": 5.1884,
       "step": 1357
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002471352019785655,
-      "loss": 5.1052,
+      "learning_rate": 0.0008237840065952184,
+      "loss": 5.7171,
       "step": 1358
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024707337180544103,
-      "loss": 5.0728,
+      "learning_rate": 0.0008235779060181368,
+      "loss": 5.5956,
       "step": 1359
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002470115416323166,
-      "loss": 5.3642,
+      "learning_rate": 0.0008233718054410553,
+      "loss": 5.8232,
       "step": 1360
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024694971145919207,
-      "loss": 4.737,
+      "learning_rate": 0.0008231657048639737,
+      "loss": 5.4079,
       "step": 1361
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.00024688788128606756,
-      "loss": 4.709,
+      "learning_rate": 0.000822959604286892,
+      "loss": 5.303,
       "step": 1362
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002468260511129431,
-      "loss": 4.5362,
+      "learning_rate": 0.0008227535037098104,
+      "loss": 5.3391,
       "step": 1363
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0002467642209398186,
-      "loss": 5.5568,
+      "learning_rate": 0.0008225474031327288,
+      "loss": 5.9572,
       "step": 1364
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024670239076669413,
-      "loss": 4.7183,
+      "learning_rate": 0.0008223413025556472,
+      "loss": 6.1778,
       "step": 1365
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002466405605935696,
-      "loss": 4.6366,
+      "learning_rate": 0.0008221352019785655,
+      "loss": 5.4847,
       "step": 1366
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024657873042044517,
-      "loss": 4.9566,
+      "learning_rate": 0.0008219291014014839,
+      "loss": 5.6996,
       "step": 1367
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024651690024732066,
-      "loss": 4.5414,
+      "learning_rate": 0.0008217230008244023,
+      "loss": 5.3126,
       "step": 1368
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002464550700741962,
-      "loss": 4.8148,
+      "learning_rate": 0.0008215169002473207,
+      "loss": 5.4157,
       "step": 1369
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002463932399010717,
-      "loss": 4.8132,
+      "learning_rate": 0.0008213107996702391,
+      "loss": 5.516,
       "step": 1370
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002463314097279472,
-      "loss": 4.9339,
+      "learning_rate": 0.0008211046990931574,
+      "loss": 5.6983,
       "step": 1371
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002462695795548227,
-      "loss": 4.4505,
+      "learning_rate": 0.0008208985985160758,
+      "loss": 5.0904,
       "step": 1372
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024620774938169827,
-      "loss": 5.1289,
+      "learning_rate": 0.0008206924979389942,
+      "loss": 5.8467,
       "step": 1373
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024614591920857375,
-      "loss": 5.0176,
+      "learning_rate": 0.0008204863973619126,
+      "loss": 5.8424,
       "step": 1374
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024608408903544924,
-      "loss": 4.573,
+      "learning_rate": 0.0008202802967848309,
+      "loss": 5.6727,
       "step": 1375
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002460222588623248,
-      "loss": 4.4707,
+      "learning_rate": 0.0008200741962077493,
+      "loss": 5.2052,
       "step": 1376
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024596042868920033,
-      "loss": 4.8226,
+      "learning_rate": 0.0008198680956306677,
+      "loss": 6.0616,
       "step": 1377
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002458985985160758,
-      "loss": 4.7205,
+      "learning_rate": 0.0008196619950535861,
+      "loss": 5.6108,
       "step": 1378
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024583676834295137,
-      "loss": 4.764,
+      "learning_rate": 0.0008194558944765046,
+      "loss": 5.5427,
       "step": 1379
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024577493816982685,
-      "loss": 4.7416,
+      "learning_rate": 0.0008192497938994228,
+      "loss": 5.3388,
       "step": 1380
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.00024571310799670234,
-      "loss": 4.3174,
+      "learning_rate": 0.0008190436933223413,
+      "loss": 5.0211,
       "step": 1381
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0002456512778235779,
-      "loss": 4.7368,
+      "learning_rate": 0.0008188375927452597,
+      "loss": 5.4836,
       "step": 1382
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024558944765045343,
-      "loss": 4.6322,
+      "learning_rate": 0.0008186314921681781,
+      "loss": 5.3669,
       "step": 1383
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002455276174773289,
-      "loss": 5.3506,
+      "learning_rate": 0.0008184253915910965,
+      "loss": 5.8945,
       "step": 1384
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002454657873042044,
-      "loss": 5.0479,
+      "learning_rate": 0.0008182192910140148,
+      "loss": 5.9555,
       "step": 1385
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024540395713107995,
-      "loss": 4.8292,
+      "learning_rate": 0.0008180131904369332,
+      "loss": 5.4759,
       "step": 1386
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024534212695795544,
-      "loss": 5.2494,
+      "learning_rate": 0.0008178070898598516,
+      "loss": 5.5654,
       "step": 1387
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.000245280296784831,
-      "loss": 4.9984,
+      "learning_rate": 0.00081760098928277,
+      "loss": 5.5622,
       "step": 1388
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002452184666117065,
-      "loss": 4.5695,
+      "learning_rate": 0.0008173948887056883,
+      "loss": 5.2607,
       "step": 1389
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.000245156636438582,
-      "loss": 5.0589,
+      "learning_rate": 0.0008171887881286068,
+      "loss": 5.6258,
       "step": 1390
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002450948062654575,
-      "loss": 4.5789,
+      "learning_rate": 0.0008169826875515252,
+      "loss": 5.1475,
       "step": 1391
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024503297609233305,
-      "loss": 4.8829,
+      "learning_rate": 0.0008167765869744436,
+      "loss": 5.625,
       "step": 1392
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024497114591920854,
-      "loss": 4.5952,
+      "learning_rate": 0.000816570486397362,
+      "loss": 5.2315,
       "step": 1393
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024490931574608403,
-      "loss": 4.81,
+      "learning_rate": 0.0008163643858202803,
+      "loss": 5.6026,
       "step": 1394
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002448474855729596,
-      "loss": 5.7901,
+      "learning_rate": 0.0008161582852431987,
+      "loss": 7.0047,
       "step": 1395
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002447856553998351,
-      "loss": 4.6923,
+      "learning_rate": 0.0008159521846661171,
+      "loss": 5.2876,
       "step": 1396
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002447238252267106,
-      "loss": 4.9657,
+      "learning_rate": 0.0008157460840890355,
+      "loss": 5.3983,
       "step": 1397
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0002446619950535861,
-      "loss": 4.5236,
+      "learning_rate": 0.0008155399835119538,
+      "loss": 5.1521,
       "step": 1398
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024460016488046164,
-      "loss": 4.6527,
+      "learning_rate": 0.0008153338829348723,
+      "loss": 5.3206,
       "step": 1399
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00024453833470733713,
-      "loss": 4.7851,
+      "learning_rate": 0.0008151277823577907,
+      "loss": 5.5063,
       "step": 1400
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0002444765045342127,
-      "loss": 4.6802,
+      "learning_rate": 0.0008149216817807091,
+      "loss": 5.3094,
       "step": 1401
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0002444146743610882,
-      "loss": 4.9679,
+      "learning_rate": 0.0008147155812036275,
+      "loss": 5.8986,
       "step": 1402
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0002443528441879637,
-      "loss": 4.6069,
+      "learning_rate": 0.0008145094806265458,
+      "loss": 5.4165,
       "step": 1403
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0002442910140148392,
-      "loss": 4.8304,
+      "learning_rate": 0.0008143033800494642,
+      "loss": 6.0671,
       "step": 1404
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024422918384171474,
-      "loss": 4.8665,
+      "learning_rate": 0.0008140972794723826,
+      "loss": 5.5996,
       "step": 1405
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0002441673536685903,
-      "loss": 5.8431,
+      "learning_rate": 0.000813891178895301,
+      "loss": 6.4383,
       "step": 1406
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024410552349546578,
-      "loss": 4.9247,
+      "learning_rate": 0.0008136850783182194,
+      "loss": 5.5775,
       "step": 1407
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024404369332234126,
-      "loss": 4.9205,
+      "learning_rate": 0.0008134789777411377,
+      "loss": 6.0582,
       "step": 1408
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024398186314921678,
-      "loss": 5.0879,
+      "learning_rate": 0.0008132728771640561,
+      "loss": 5.7177,
       "step": 1409
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024392003297609232,
-      "loss": 4.8419,
+      "learning_rate": 0.0008130667765869745,
+      "loss": 5.3655,
       "step": 1410
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024385820280296784,
-      "loss": 4.4759,
+      "learning_rate": 0.0008128606760098929,
+      "loss": 5.3407,
       "step": 1411
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024379637262984333,
-      "loss": 4.5185,
+      "learning_rate": 0.0008126545754328112,
+      "loss": 5.442,
       "step": 1412
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024373454245671885,
-      "loss": 4.9616,
+      "learning_rate": 0.0008124484748557296,
+      "loss": 5.6227,
       "step": 1413
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024367271228359436,
-      "loss": 5.11,
+      "learning_rate": 0.000812242374278648,
+      "loss": 5.6585,
       "step": 1414
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0002436108821104699,
-      "loss": 4.8119,
+      "learning_rate": 0.0008120362737015664,
+      "loss": 5.3374,
       "step": 1415
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024354905193734542,
-      "loss": 4.9236,
+      "learning_rate": 0.0008118301731244848,
+      "loss": 5.7209,
       "step": 1416
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024348722176422091,
-      "loss": 5.1093,
+      "learning_rate": 0.0008116240725474031,
+      "loss": 5.796,
       "step": 1417
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.00024342539159109643,
-      "loss": 5.31,
+      "learning_rate": 0.0008114179719703215,
+      "loss": 6.1732,
       "step": 1418
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024336356141797195,
-      "loss": 5.4835,
+      "learning_rate": 0.00081121187139324,
+      "loss": 5.9102,
       "step": 1419
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024330173124484746,
-      "loss": 4.9429,
+      "learning_rate": 0.0008110057708161584,
+      "loss": 5.3661,
       "step": 1420
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024323990107172295,
-      "loss": 4.8542,
+      "learning_rate": 0.0008107996702390767,
+      "loss": 5.8076,
       "step": 1421
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0002431780708985985,
-      "loss": 4.9146,
+      "learning_rate": 0.0008105935696619951,
+      "loss": 6.0575,
       "step": 1422
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.000243116240725474,
-      "loss": 4.6364,
+      "learning_rate": 0.0008103874690849135,
+      "loss": 5.3093,
       "step": 1423
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024305441055234953,
-      "loss": 4.8322,
+      "learning_rate": 0.0008101813685078319,
+      "loss": 5.4761,
       "step": 1424
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024299258037922505,
-      "loss": 4.9857,
+      "learning_rate": 0.0008099752679307503,
+      "loss": 6.57,
       "step": 1425
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024293075020610054,
-      "loss": 5.0727,
+      "learning_rate": 0.0008097691673536686,
+      "loss": 5.6883,
       "step": 1426
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024286892003297608,
-      "loss": 4.5371,
+      "learning_rate": 0.000809563066776587,
+      "loss": 5.2033,
       "step": 1427
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0002428070898598516,
-      "loss": 4.6024,
+      "learning_rate": 0.0008093569661995054,
+      "loss": 5.4502,
       "step": 1428
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0002427452596867271,
-      "loss": 4.9698,
+      "learning_rate": 0.0008091508656224238,
+      "loss": 6.3243,
       "step": 1429
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0002426834295136026,
-      "loss": 6.0323,
+      "learning_rate": 0.0008089447650453421,
+      "loss": 6.9092,
       "step": 1430
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024262159934047812,
-      "loss": 5.7732,
+      "learning_rate": 0.0008087386644682605,
+      "loss": 6.8114,
       "step": 1431
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024255976916735364,
-      "loss": 4.8002,
+      "learning_rate": 0.0008085325638911789,
+      "loss": 5.8481,
       "step": 1432
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024249793899422918,
-      "loss": 5.589,
+      "learning_rate": 0.0008083264633140973,
+      "loss": 6.0423,
       "step": 1433
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0002424361088211047,
-      "loss": 4.5119,
+      "learning_rate": 0.0008081203627370157,
+      "loss": 5.5656,
       "step": 1434
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00024237427864798019,
-      "loss": 4.7496,
+      "learning_rate": 0.000807914262159934,
+      "loss": 5.8115,
       "step": 1435
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0002423124484748557,
-      "loss": 4.5913,
+      "learning_rate": 0.0008077081615828524,
+      "loss": 5.156,
       "step": 1436
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024225061830173122,
-      "loss": 4.8618,
+      "learning_rate": 0.0008075020610057708,
+      "loss": 5.7272,
       "step": 1437
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024218878812860674,
-      "loss": 4.7835,
+      "learning_rate": 0.0008072959604286892,
+      "loss": 5.2911,
       "step": 1438
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024212695795548228,
-      "loss": 4.6899,
+      "learning_rate": 0.0008070898598516077,
+      "loss": 5.3085,
       "step": 1439
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024206512778235777,
-      "loss": 5.0207,
+      "learning_rate": 0.000806883759274526,
+      "loss": 5.7923,
       "step": 1440
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024200329760923328,
-      "loss": 5.7498,
+      "learning_rate": 0.0008066776586974444,
+      "loss": 6.611,
       "step": 1441
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0002419414674361088,
-      "loss": 4.6002,
+      "learning_rate": 0.0008064715581203628,
+      "loss": 5.3446,
       "step": 1442
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024187963726298432,
-      "loss": 4.4766,
+      "learning_rate": 0.0008062654575432812,
+      "loss": 5.2329,
       "step": 1443
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0002418178070898598,
-      "loss": 4.837,
+      "learning_rate": 0.0008060593569661995,
+      "loss": 5.231,
       "step": 1444
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024175597691673535,
-      "loss": 4.6812,
+      "learning_rate": 0.0008058532563891179,
+      "loss": 5.3781,
       "step": 1445
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024169414674361087,
-      "loss": 5.0555,
+      "learning_rate": 0.0008056471558120363,
+      "loss": 5.4795,
       "step": 1446
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024163231657048638,
-      "loss": 4.7179,
+      "learning_rate": 0.0008054410552349547,
+      "loss": 5.35,
       "step": 1447
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0002415704863973619,
-      "loss": 4.7445,
+      "learning_rate": 0.0008052349546578731,
+      "loss": 5.2862,
       "step": 1448
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0002415086562242374,
-      "loss": 4.9463,
+      "learning_rate": 0.0008050288540807914,
+      "loss": 5.5892,
       "step": 1449
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0002414468260511129,
-      "loss": 4.9569,
+      "learning_rate": 0.0008048227535037098,
+      "loss": 5.3857,
       "step": 1450
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024138499587798845,
-      "loss": 4.7043,
+      "learning_rate": 0.0008046166529266282,
+      "loss": 5.6694,
       "step": 1451
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024132316570486397,
-      "loss": 4.8128,
+      "learning_rate": 0.0008044105523495466,
+      "loss": 5.9201,
       "step": 1452
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.00024126133553173946,
-      "loss": 4.6958,
+      "learning_rate": 0.0008042044517724649,
+      "loss": 5.3134,
       "step": 1453
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024119950535861497,
-      "loss": 4.5202,
+      "learning_rate": 0.0008039983511953833,
+      "loss": 5.3417,
       "step": 1454
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0002411376751854905,
-      "loss": 5.1629,
+      "learning_rate": 0.0008037922506183017,
+      "loss": 6.4734,
       "step": 1455
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.000241075845012366,
-      "loss": 4.6623,
+      "learning_rate": 0.0008035861500412201,
+      "loss": 5.8254,
       "step": 1456
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024101401483924155,
-      "loss": 5.1408,
+      "learning_rate": 0.0008033800494641385,
+      "loss": 5.8066,
       "step": 1457
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024095218466611704,
-      "loss": 4.9028,
+      "learning_rate": 0.0008031739488870568,
+      "loss": 6.0904,
       "step": 1458
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024089035449299256,
-      "loss": 4.9198,
+      "learning_rate": 0.0008029678483099752,
+      "loss": 5.6002,
       "step": 1459
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024082852431986807,
-      "loss": 5.0641,
+      "learning_rate": 0.0008027617477328936,
+      "loss": 6.4152,
       "step": 1460
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0002407666941467436,
-      "loss": 4.8818,
+      "learning_rate": 0.0008025556471558121,
+      "loss": 5.9509,
       "step": 1461
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024070486397361908,
-      "loss": 4.9657,
+      "learning_rate": 0.0008023495465787304,
+      "loss": 5.572,
       "step": 1462
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024064303380049462,
-      "loss": 4.799,
+      "learning_rate": 0.0008021434460016488,
+      "loss": 5.3032,
       "step": 1463
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024058120362737014,
-      "loss": 5.048,
+      "learning_rate": 0.0008019373454245672,
+      "loss": 5.6387,
       "step": 1464
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024051937345424566,
-      "loss": 5.3118,
+      "learning_rate": 0.0008017312448474856,
+      "loss": 6.2012,
       "step": 1465
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024045754328112117,
-      "loss": 4.5314,
+      "learning_rate": 0.000801525144270404,
+      "loss": 5.7121,
       "step": 1466
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024039571310799666,
-      "loss": 4.8923,
+      "learning_rate": 0.0008013190436933223,
+      "loss": 5.4505,
       "step": 1467
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024033388293487218,
-      "loss": 4.7976,
+      "learning_rate": 0.0008011129431162407,
+      "loss": 5.7396,
       "step": 1468
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024027205276174772,
-      "loss": 4.4189,
+      "learning_rate": 0.0008009068425391591,
+      "loss": 5.7421,
       "step": 1469
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024021022258862324,
-      "loss": 4.4584,
+      "learning_rate": 0.0008007007419620775,
+      "loss": 5.261,
       "step": 1470
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.00024014839241549876,
-      "loss": 4.8762,
+      "learning_rate": 0.0008004946413849959,
+      "loss": 5.6178,
       "step": 1471
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00024008656224237424,
-      "loss": 4.3889,
+      "learning_rate": 0.0008002885408079142,
+      "loss": 5.2159,
       "step": 1472
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00024002473206924976,
-      "loss": 4.7034,
+      "learning_rate": 0.0008000824402308326,
+      "loss": 5.3448,
       "step": 1473
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0002399629018961253,
-      "loss": 4.6724,
+      "learning_rate": 0.000799876339653751,
+      "loss": 5.3446,
       "step": 1474
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023990107172300082,
-      "loss": 4.7877,
+      "learning_rate": 0.0007996702390766694,
+      "loss": 5.7379,
       "step": 1475
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0002398392415498763,
-      "loss": 5.2143,
+      "learning_rate": 0.0007994641384995877,
+      "loss": 5.9696,
       "step": 1476
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023977741137675183,
-      "loss": 4.9502,
+      "learning_rate": 0.0007992580379225061,
+      "loss": 5.6844,
       "step": 1477
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023971558120362734,
-      "loss": 4.534,
+      "learning_rate": 0.0007990519373454245,
+      "loss": 5.2966,
       "step": 1478
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023965375103050286,
-      "loss": 4.6996,
+      "learning_rate": 0.0007988458367683429,
+      "loss": 5.544,
       "step": 1479
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0002395919208573784,
-      "loss": 4.4636,
+      "learning_rate": 0.0007986397361912613,
+      "loss": 5.6466,
       "step": 1480
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0002395300906842539,
-      "loss": 4.7006,
+      "learning_rate": 0.0007984336356141796,
+      "loss": 5.4104,
       "step": 1481
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0002394682605111294,
-      "loss": 5.3915,
+      "learning_rate": 0.000798227535037098,
+      "loss": 6.1775,
       "step": 1482
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023940643033800493,
-      "loss": 4.8269,
+      "learning_rate": 0.0007980214344600166,
+      "loss": 5.6859,
       "step": 1483
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023934460016488044,
-      "loss": 4.4892,
+      "learning_rate": 0.000797815333882935,
+      "loss": 5.0078,
       "step": 1484
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023928276999175593,
-      "loss": 4.656,
+      "learning_rate": 0.0007976092333058533,
+      "loss": 5.4872,
       "step": 1485
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023922093981863145,
-      "loss": 4.8823,
+      "learning_rate": 0.0007974031327287717,
+      "loss": 5.7126,
       "step": 1486
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.000239159109645507,
-      "loss": 5.0535,
+      "learning_rate": 0.0007971970321516901,
+      "loss": 5.5299,
       "step": 1487
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0002390972794723825,
-      "loss": 4.7276,
+      "learning_rate": 0.0007969909315746085,
+      "loss": 5.5282,
       "step": 1488
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.00023903544929925803,
-      "loss": 5.2805,
+      "learning_rate": 0.0007967848309975269,
+      "loss": 5.8606,
       "step": 1489
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023897361912613352,
-      "loss": 4.4985,
+      "learning_rate": 0.0007965787304204452,
+      "loss": 5.1812,
       "step": 1490
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023891178895300903,
-      "loss": 5.1172,
+      "learning_rate": 0.0007963726298433636,
+      "loss": 5.7487,
       "step": 1491
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023884995877988458,
-      "loss": 4.9215,
+      "learning_rate": 0.000796166529266282,
+      "loss": 5.6274,
       "step": 1492
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0002387881286067601,
-      "loss": 4.1764,
+      "learning_rate": 0.0007959604286892004,
+      "loss": 4.9282,
       "step": 1493
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023872629843363558,
-      "loss": 5.1492,
+      "learning_rate": 0.0007957543281121187,
+      "loss": 5.6627,
       "step": 1494
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0002386644682605111,
-      "loss": 4.4746,
+      "learning_rate": 0.0007955482275350371,
+      "loss": 5.3832,
       "step": 1495
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023860263808738662,
-      "loss": 5.911,
+      "learning_rate": 0.0007953421269579555,
+      "loss": 6.0631,
       "step": 1496
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023854080791426213,
-      "loss": 4.8999,
+      "learning_rate": 0.0007951360263808739,
+      "loss": 5.4315,
       "step": 1497
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023847897774113768,
-      "loss": 4.853,
+      "learning_rate": 0.0007949299258037923,
+      "loss": 5.7649,
       "step": 1498
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023841714756801317,
-      "loss": 4.5484,
+      "learning_rate": 0.0007947238252267106,
+      "loss": 4.9583,
       "step": 1499
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023835531739488868,
-      "loss": 4.6933,
+      "learning_rate": 0.000794517724649629,
+      "loss": 5.5664,
       "step": 1500
     },
     {
       "epoch": 0.84,
-      "eval_loss": 5.082033634185791,
-      "eval_runtime": 1247.1343,
-      "eval_samples_per_second": 2.118,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.9541848472828243,
+      "eval_loss": 5.693754196166992,
+      "eval_runtime": 1243.8508,
+      "eval_samples_per_second": 2.124,
+      "eval_steps_per_second": 0.266,
+      "eval_wer": 2.006644188813963,
       "step": 1500
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0002382934872217642,
-      "loss": 4.9285,
+      "learning_rate": 0.0007943116240725475,
+      "loss": 5.2739,
       "step": 1501
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023823165704863972,
-      "loss": 4.9595,
+      "learning_rate": 0.0007941055234954659,
+      "loss": 5.5035,
       "step": 1502
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023816982687551526,
-      "loss": 4.9483,
+      "learning_rate": 0.0007938994229183843,
+      "loss": 5.8292,
       "step": 1503
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023810799670239075,
-      "loss": 4.8377,
+      "learning_rate": 0.0007936933223413026,
+      "loss": 5.799,
       "step": 1504
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023804616652926627,
-      "loss": 5.0874,
+      "learning_rate": 0.000793487221764221,
+      "loss": 5.5344,
       "step": 1505
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.00023798433635614178,
-      "loss": 4.9229,
+      "learning_rate": 0.0007932811211871394,
+      "loss": 5.7577,
       "step": 1506
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0002379225061830173,
-      "loss": 4.5712,
+      "learning_rate": 0.0007930750206100578,
+      "loss": 5.273,
       "step": 1507
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0002378606760098928,
-      "loss": 4.6552,
+      "learning_rate": 0.0007928689200329761,
+      "loss": 5.3651,
       "step": 1508
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0002377988458367683,
-      "loss": 4.7745,
+      "learning_rate": 0.0007926628194558945,
+      "loss": 5.6284,
       "step": 1509
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023773701566364385,
-      "loss": 4.7217,
+      "learning_rate": 0.0007924567188788129,
+      "loss": 5.3762,
       "step": 1510
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023767518549051936,
-      "loss": 4.6303,
+      "learning_rate": 0.0007922506183017313,
+      "loss": 5.2086,
       "step": 1511
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023761335531739488,
-      "loss": 4.6386,
+      "learning_rate": 0.0007920445177246497,
+      "loss": 5.4511,
       "step": 1512
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023755152514427037,
-      "loss": 4.3382,
+      "learning_rate": 0.000791838417147568,
+      "loss": 5.1656,
       "step": 1513
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0002374896949711459,
-      "loss": 4.6542,
+      "learning_rate": 0.0007916323165704864,
+      "loss": 5.3216,
       "step": 1514
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0002374278647980214,
-      "loss": 4.6631,
+      "learning_rate": 0.0007914262159934048,
+      "loss": 5.236,
       "step": 1515
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023736603462489695,
-      "loss": 5.2341,
+      "learning_rate": 0.0007912201154163232,
+      "loss": 6.1776,
       "step": 1516
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023730420445177244,
-      "loss": 4.5276,
+      "learning_rate": 0.0007910140148392415,
+      "loss": 5.2409,
       "step": 1517
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023724237427864795,
-      "loss": 4.9346,
+      "learning_rate": 0.0007908079142621599,
+      "loss": 5.4533,
       "step": 1518
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023718054410552347,
-      "loss": 4.8863,
+      "learning_rate": 0.0007906018136850783,
+      "loss": 6.0416,
       "step": 1519
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.000237118713932399,
-      "loss": 4.8335,
+      "learning_rate": 0.0007903957131079967,
+      "loss": 5.7621,
       "step": 1520
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023705688375927453,
-      "loss": 4.7333,
+      "learning_rate": 0.0007901896125309152,
+      "loss": 5.3385,
       "step": 1521
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023699505358615002,
-      "loss": 4.7295,
+      "learning_rate": 0.0007899835119538335,
+      "loss": 5.5288,
       "step": 1522
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023693322341302554,
-      "loss": 5.2005,
+      "learning_rate": 0.0007897774113767519,
+      "loss": 5.8208,
       "step": 1523
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023687139323990105,
-      "loss": 4.5409,
+      "learning_rate": 0.0007895713107996703,
+      "loss": 5.2498,
       "step": 1524
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.00023680956306677657,
-      "loss": 4.6543,
+      "learning_rate": 0.0007893652102225887,
+      "loss": 5.1569,
       "step": 1525
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023674773289365209,
-      "loss": 4.5279,
+      "learning_rate": 0.0007891591096455071,
+      "loss": 5.0932,
       "step": 1526
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023668590272052758,
-      "loss": 4.6857,
+      "learning_rate": 0.0007889530090684254,
+      "loss": 5.3862,
       "step": 1527
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023662407254740312,
-      "loss": 4.5479,
+      "learning_rate": 0.0007887469084913438,
+      "loss": 5.2179,
       "step": 1528
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023656224237427864,
-      "loss": 4.6299,
+      "learning_rate": 0.0007885408079142622,
+      "loss": 5.2247,
       "step": 1529
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023650041220115415,
-      "loss": 4.5424,
+      "learning_rate": 0.0007883347073371806,
+      "loss": 5.2149,
       "step": 1530
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023643858202802964,
-      "loss": 4.6152,
+      "learning_rate": 0.0007881286067600989,
+      "loss": 5.2661,
       "step": 1531
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023637675185490516,
-      "loss": 4.8817,
+      "learning_rate": 0.0007879225061830173,
+      "loss": 6.1815,
       "step": 1532
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0002363149216817807,
-      "loss": 5.4905,
+      "learning_rate": 0.0007877164056059357,
+      "loss": 5.6746,
       "step": 1533
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023625309150865622,
-      "loss": 4.7001,
+      "learning_rate": 0.0007875103050288541,
+      "loss": 5.2847,
       "step": 1534
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023619126133553174,
-      "loss": 4.8376,
+      "learning_rate": 0.0007873042044517725,
+      "loss": 6.6166,
       "step": 1535
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023612943116240723,
-      "loss": 4.3855,
+      "learning_rate": 0.0007870981038746908,
+      "loss": 5.2027,
       "step": 1536
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023606760098928274,
-      "loss": 5.657,
+      "learning_rate": 0.0007868920032976092,
+      "loss": 6.5368,
       "step": 1537
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023600577081615826,
-      "loss": 4.9685,
+      "learning_rate": 0.0007866859027205276,
+      "loss": 5.7218,
       "step": 1538
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0002359439406430338,
-      "loss": 5.1409,
+      "learning_rate": 0.000786479802143446,
+      "loss": 5.4336,
       "step": 1539
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0002358821104699093,
-      "loss": 4.903,
+      "learning_rate": 0.0007862737015663643,
+      "loss": 5.4167,
       "step": 1540
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0002358202802967848,
-      "loss": 5.1782,
+      "learning_rate": 0.0007860676009892827,
+      "loss": 5.5607,
       "step": 1541
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023575845012366032,
-      "loss": 4.5818,
+      "learning_rate": 0.0007858615004122012,
+      "loss": 5.5747,
       "step": 1542
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.00023569661995053584,
-      "loss": 4.8142,
+      "learning_rate": 0.0007856553998351196,
+      "loss": 5.5386,
       "step": 1543
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023563478977741136,
-      "loss": 5.0285,
+      "learning_rate": 0.000785449299258038,
+      "loss": 5.7116,
       "step": 1544
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023557295960428685,
-      "loss": 5.7733,
+      "learning_rate": 0.0007852431986809563,
+      "loss": 6.2746,
       "step": 1545
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0002355111294311624,
-      "loss": 4.4133,
+      "learning_rate": 0.0007850370981038747,
+      "loss": 5.2856,
       "step": 1546
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0002354492992580379,
-      "loss": 4.8472,
+      "learning_rate": 0.0007848309975267931,
+      "loss": 5.4119,
       "step": 1547
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023538746908491342,
-      "loss": 4.703,
+      "learning_rate": 0.0007846248969497115,
+      "loss": 5.4726,
       "step": 1548
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023532563891178891,
-      "loss": 5.423,
+      "learning_rate": 0.0007844187963726298,
+      "loss": 5.7192,
       "step": 1549
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023526380873866443,
-      "loss": 4.2883,
+      "learning_rate": 0.0007842126957955482,
+      "loss": 5.7007,
       "step": 1550
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023520197856553997,
-      "loss": 4.983,
+      "learning_rate": 0.0007840065952184666,
+      "loss": 5.7659,
       "step": 1551
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0002351401483924155,
-      "loss": 5.005,
+      "learning_rate": 0.000783800494641385,
+      "loss": 5.7072,
       "step": 1552
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.000235078318219291,
-      "loss": 4.5468,
+      "learning_rate": 0.0007835943940643034,
+      "loss": 5.3353,
       "step": 1553
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0002350164880461665,
-      "loss": 4.7262,
+      "learning_rate": 0.0007833882934872217,
+      "loss": 5.3389,
       "step": 1554
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.000234954657873042,
-      "loss": 5.222,
+      "learning_rate": 0.0007831821929101401,
+      "loss": 5.7256,
       "step": 1555
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023489282769991753,
-      "loss": 5.2159,
+      "learning_rate": 0.0007829760923330585,
+      "loss": 6.2199,
       "step": 1556
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023483099752679307,
-      "loss": 4.7524,
+      "learning_rate": 0.0007827699917559769,
+      "loss": 5.7906,
       "step": 1557
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0002347691673536686,
-      "loss": 5.0152,
+      "learning_rate": 0.0007825638911788953,
+      "loss": 5.4976,
       "step": 1558
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.00023470733718054408,
-      "loss": 4.7309,
+      "learning_rate": 0.0007823577906018136,
+      "loss": 5.2849,
       "step": 1559
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0002346455070074196,
-      "loss": 4.8942,
+      "learning_rate": 0.000782151690024732,
+      "loss": 5.7787,
       "step": 1560
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0002345836768342951,
-      "loss": 4.9745,
+      "learning_rate": 0.0007819455894476504,
+      "loss": 5.7005,
       "step": 1561
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023452184666117063,
-      "loss": 4.7331,
+      "learning_rate": 0.0007817394888705689,
+      "loss": 5.5949,
       "step": 1562
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023446001648804612,
-      "loss": 4.8417,
+      "learning_rate": 0.0007815333882934872,
+      "loss": 5.8771,
       "step": 1563
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023439818631492166,
-      "loss": 4.848,
+      "learning_rate": 0.0007813272877164056,
+      "loss": 5.3363,
       "step": 1564
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023433635614179718,
-      "loss": 4.7451,
+      "learning_rate": 0.000781121187139324,
+      "loss": 5.3779,
       "step": 1565
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0002342745259686727,
-      "loss": 4.5532,
+      "learning_rate": 0.0007809150865622424,
+      "loss": 5.6413,
       "step": 1566
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0002342126957955482,
-      "loss": 4.6129,
+      "learning_rate": 0.0007807089859851608,
+      "loss": 5.0905,
       "step": 1567
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0002341508656224237,
-      "loss": 4.6276,
+      "learning_rate": 0.0007805028854080791,
+      "loss": 5.4187,
       "step": 1568
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023408903544929925,
-      "loss": 4.6892,
+      "learning_rate": 0.0007802967848309975,
+      "loss": 5.4871,
       "step": 1569
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023402720527617476,
-      "loss": 5.0961,
+      "learning_rate": 0.0007800906842539159,
+      "loss": 5.5974,
       "step": 1570
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023396537510305028,
-      "loss": 4.6512,
+      "learning_rate": 0.0007798845836768343,
+      "loss": 5.3391,
       "step": 1571
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023390354492992577,
-      "loss": 5.1904,
+      "learning_rate": 0.0007796784830997526,
+      "loss": 5.5727,
       "step": 1572
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023384171475680128,
-      "loss": 4.8098,
+      "learning_rate": 0.000779472382522671,
+      "loss": 5.5227,
       "step": 1573
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0002337798845836768,
-      "loss": 4.4232,
+      "learning_rate": 0.0007792662819455894,
+      "loss": 5.0716,
       "step": 1574
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023371805441055234,
-      "loss": 4.7384,
+      "learning_rate": 0.0007790601813685078,
+      "loss": 5.1044,
       "step": 1575
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023365622423742786,
-      "loss": 5.3975,
+      "learning_rate": 0.0007788540807914263,
+      "loss": 5.5813,
       "step": 1576
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023359439406430335,
-      "loss": 5.2509,
+      "learning_rate": 0.0007786479802143446,
+      "loss": 5.9337,
       "step": 1577
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.00023353256389117887,
-      "loss": 5.2051,
+      "learning_rate": 0.000778441879637263,
+      "loss": 5.5009,
       "step": 1578
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023347073371805438,
-      "loss": 4.8475,
+      "learning_rate": 0.0007782357790601814,
+      "loss": 5.4651,
       "step": 1579
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023340890354492993,
-      "loss": 5.052,
+      "learning_rate": 0.0007780296784830998,
+      "loss": 6.2407,
       "step": 1580
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023334707337180542,
-      "loss": 5.0007,
+      "learning_rate": 0.0007778235779060181,
+      "loss": 5.5154,
       "step": 1581
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023328524319868093,
-      "loss": 4.7532,
+      "learning_rate": 0.0007776174773289366,
+      "loss": 5.5077,
       "step": 1582
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023322341302555645,
-      "loss": 4.6573,
+      "learning_rate": 0.000777411376751855,
+      "loss": 5.4512,
       "step": 1583
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023316158285243197,
-      "loss": 4.4891,
+      "learning_rate": 0.0007772052761747734,
+      "loss": 5.2939,
       "step": 1584
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023309975267930748,
-      "loss": 4.8499,
+      "learning_rate": 0.0007769991755976918,
+      "loss": 5.384,
       "step": 1585
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023303792250618297,
-      "loss": 4.5333,
+      "learning_rate": 0.0007767930750206101,
+      "loss": 5.2142,
       "step": 1586
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023297609233305852,
-      "loss": 4.4806,
+      "learning_rate": 0.0007765869744435285,
+      "loss": 5.6352,
       "step": 1587
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023291426215993403,
-      "loss": 4.6918,
+      "learning_rate": 0.0007763808738664469,
+      "loss": 5.3955,
       "step": 1588
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023285243198680955,
-      "loss": 5.2144,
+      "learning_rate": 0.0007761747732893653,
+      "loss": 5.6869,
       "step": 1589
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023279060181368507,
-      "loss": 4.4978,
+      "learning_rate": 0.0007759686727122837,
+      "loss": 5.3159,
       "step": 1590
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023272877164056056,
-      "loss": 4.5555,
+      "learning_rate": 0.000775762572135202,
+      "loss": 5.1971,
       "step": 1591
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023266694146743607,
-      "loss": 4.6758,
+      "learning_rate": 0.0007755564715581204,
+      "loss": 5.1773,
       "step": 1592
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023260511129431162,
-      "loss": 4.4474,
+      "learning_rate": 0.0007753503709810388,
+      "loss": 5.1063,
       "step": 1593
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023254328112118713,
-      "loss": 4.5061,
+      "learning_rate": 0.0007751442704039572,
+      "loss": 5.1922,
       "step": 1594
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023248145094806262,
-      "loss": 4.4002,
+      "learning_rate": 0.0007749381698268755,
+      "loss": 5.1557,
       "step": 1595
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.00023241962077493814,
-      "loss": 4.564,
+      "learning_rate": 0.0007747320692497939,
+      "loss": 5.214,
       "step": 1596
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023235779060181366,
-      "loss": 4.8266,
+      "learning_rate": 0.0007745259686727123,
+      "loss": 5.5801,
       "step": 1597
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002322959604286892,
-      "loss": 5.0188,
+      "learning_rate": 0.0007743198680956307,
+      "loss": 5.5975,
       "step": 1598
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023223413025556472,
-      "loss": 5.4635,
+      "learning_rate": 0.0007741137675185491,
+      "loss": 6.3652,
       "step": 1599
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002321723000824402,
-      "loss": 4.7864,
+      "learning_rate": 0.0007739076669414674,
+      "loss": 5.4861,
       "step": 1600
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023211046990931572,
-      "loss": 4.7678,
+      "learning_rate": 0.0007737015663643858,
+      "loss": 5.6266,
       "step": 1601
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023204863973619124,
-      "loss": 4.5117,
+      "learning_rate": 0.0007734954657873043,
+      "loss": 5.2021,
       "step": 1602
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023198680956306676,
-      "loss": 4.6449,
+      "learning_rate": 0.0007732893652102227,
+      "loss": 5.7848,
       "step": 1603
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023192497938994224,
-      "loss": 4.7125,
+      "learning_rate": 0.000773083264633141,
+      "loss": 5.6969,
       "step": 1604
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002318631492168178,
-      "loss": 4.3693,
+      "learning_rate": 0.0007728771640560594,
+      "loss": 5.2357,
       "step": 1605
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002318013190436933,
-      "loss": 4.895,
+      "learning_rate": 0.0007726710634789778,
+      "loss": 5.402,
       "step": 1606
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023173948887056882,
-      "loss": 5.1108,
+      "learning_rate": 0.0007724649629018962,
+      "loss": 5.5232,
       "step": 1607
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023167765869744434,
-      "loss": 4.6662,
+      "learning_rate": 0.0007722588623248146,
+      "loss": 5.867,
       "step": 1608
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023161582852431983,
-      "loss": 4.7588,
+      "learning_rate": 0.0007720527617477329,
+      "loss": 5.7561,
       "step": 1609
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023155399835119537,
-      "loss": 4.636,
+      "learning_rate": 0.0007718466611706513,
+      "loss": 5.5532,
       "step": 1610
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002314921681780709,
-      "loss": 4.8012,
+      "learning_rate": 0.0007716405605935697,
+      "loss": 5.456,
       "step": 1611
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002314303380049464,
-      "loss": 4.6576,
+      "learning_rate": 0.0007714344600164881,
+      "loss": 5.6274,
       "step": 1612
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.00023136850783182192,
-      "loss": 5.3625,
+      "learning_rate": 0.0007712283594394065,
+      "loss": 5.9085,
       "step": 1613
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0002313066776586974,
-      "loss": 4.7287,
+      "learning_rate": 0.0007710222588623248,
+      "loss": 5.1706,
       "step": 1614
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023124484748557293,
-      "loss": 4.8719,
+      "learning_rate": 0.0007708161582852432,
+      "loss": 5.3254,
       "step": 1615
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023118301731244847,
-      "loss": 4.9757,
+      "learning_rate": 0.0007706100577081616,
+      "loss": 5.4561,
       "step": 1616
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.000231121187139324,
-      "loss": 4.8135,
+      "learning_rate": 0.00077040395713108,
+      "loss": 5.5489,
       "step": 1617
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023105935696619948,
-      "loss": 5.0357,
+      "learning_rate": 0.0007701978565539983,
+      "loss": 5.6313,
       "step": 1618
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.000230997526793075,
-      "loss": 4.5884,
+      "learning_rate": 0.0007699917559769167,
+      "loss": 5.306,
       "step": 1619
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0002309356966199505,
-      "loss": 4.7085,
+      "learning_rate": 0.0007697856553998351,
+      "loss": 5.2442,
       "step": 1620
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023087386644682603,
-      "loss": 4.9736,
+      "learning_rate": 0.0007695795548227535,
+      "loss": 6.3999,
       "step": 1621
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023081203627370157,
-      "loss": 4.4068,
+      "learning_rate": 0.000769373454245672,
+      "loss": 5.1559,
       "step": 1622
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023075020610057706,
-      "loss": 4.6702,
+      "learning_rate": 0.0007691673536685903,
+      "loss": 5.4703,
       "step": 1623
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023068837592745258,
-      "loss": 4.4935,
+      "learning_rate": 0.0007689612530915087,
+      "loss": 5.5496,
       "step": 1624
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0002306265457543281,
-      "loss": 4.6464,
+      "learning_rate": 0.0007687551525144271,
+      "loss": 5.4868,
       "step": 1625
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0002305647155812036,
-      "loss": 5.1718,
+      "learning_rate": 0.0007685490519373455,
+      "loss": 5.6641,
       "step": 1626
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0002305028854080791,
-      "loss": 4.7043,
+      "learning_rate": 0.0007683429513602638,
+      "loss": 5.5336,
       "step": 1627
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023044105523495464,
-      "loss": 4.4086,
+      "learning_rate": 0.0007681368507831822,
+      "loss": 5.2135,
       "step": 1628
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023037922506183016,
-      "loss": 4.6977,
+      "learning_rate": 0.0007679307502061006,
+      "loss": 5.4933,
       "step": 1629
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023031739488870568,
-      "loss": 4.4776,
+      "learning_rate": 0.000767724649629019,
+      "loss": 5.1853,
       "step": 1630
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0002302555647155812,
-      "loss": 4.648,
+      "learning_rate": 0.0007675185490519374,
+      "loss": 5.1833,
       "step": 1631
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00023019373454245668,
-      "loss": 4.7393,
+      "learning_rate": 0.0007673124484748557,
+      "loss": 5.5151,
       "step": 1632
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0002301319043693322,
-      "loss": 5.4424,
+      "learning_rate": 0.0007671063478977741,
+      "loss": 6.6036,
       "step": 1633
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00023007007419620774,
-      "loss": 4.5618,
+      "learning_rate": 0.0007669002473206925,
+      "loss": 5.2396,
       "step": 1634
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00023000824402308326,
-      "loss": 4.5858,
+      "learning_rate": 0.0007666941467436109,
+      "loss": 5.3314,
       "step": 1635
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022994641384995875,
-      "loss": 4.6626,
+      "learning_rate": 0.0007664880461665292,
+      "loss": 6.0022,
       "step": 1636
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022988458367683426,
-      "loss": 4.4701,
+      "learning_rate": 0.0007662819455894476,
+      "loss": 5.1715,
       "step": 1637
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022982275350370978,
-      "loss": 5.9422,
+      "learning_rate": 0.000766075845012366,
+      "loss": 6.3325,
       "step": 1638
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0002297609233305853,
-      "loss": 4.3606,
+      "learning_rate": 0.0007658697444352844,
+      "loss": 5.1207,
       "step": 1639
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022969909315746084,
-      "loss": 5.0088,
+      "learning_rate": 0.0007656636438582028,
+      "loss": 5.6787,
       "step": 1640
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022963726298433633,
-      "loss": 4.845,
+      "learning_rate": 0.0007654575432811211,
+      "loss": 5.6104,
       "step": 1641
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022957543281121185,
-      "loss": 5.0437,
+      "learning_rate": 0.0007652514427040395,
+      "loss": 5.3238,
       "step": 1642
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022951360263808736,
-      "loss": 4.6495,
+      "learning_rate": 0.000765045342126958,
+      "loss": 5.3629,
       "step": 1643
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022945177246496288,
-      "loss": 4.5278,
+      "learning_rate": 0.0007648392415498764,
+      "loss": 5.4534,
       "step": 1644
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022938994229183842,
-      "loss": 5.2099,
+      "learning_rate": 0.0007646331409727948,
+      "loss": 5.6934,
       "step": 1645
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022932811211871391,
-      "loss": 4.9844,
+      "learning_rate": 0.0007644270403957131,
+      "loss": 5.7245,
       "step": 1646
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022926628194558943,
-      "loss": 4.6259,
+      "learning_rate": 0.0007642209398186315,
+      "loss": 5.4302,
       "step": 1647
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022920445177246495,
-      "loss": 5.4259,
+      "learning_rate": 0.0007640148392415499,
+      "loss": 5.5526,
       "step": 1648
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022914262159934046,
-      "loss": 4.8576,
+      "learning_rate": 0.0007638087386644683,
+      "loss": 5.6798,
       "step": 1649
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.00022908079142621595,
-      "loss": 4.4109,
+      "learning_rate": 0.0007636026380873866,
+      "loss": 5.3074,
       "step": 1650
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022901896125309147,
-      "loss": 4.7714,
+      "learning_rate": 0.000763396537510305,
+      "loss": 6.9425,
       "step": 1651
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022895713107996701,
-      "loss": 4.6198,
+      "learning_rate": 0.0007631904369332234,
+      "loss": 5.4832,
       "step": 1652
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022889530090684253,
-      "loss": 4.718,
+      "learning_rate": 0.0007629843363561418,
+      "loss": 5.6279,
       "step": 1653
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022883347073371805,
-      "loss": 4.6786,
+      "learning_rate": 0.0007627782357790602,
+      "loss": 5.4551,
       "step": 1654
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022877164056059354,
-      "loss": 4.4223,
+      "learning_rate": 0.0007625721352019785,
+      "loss": 5.146,
       "step": 1655
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022870981038746905,
-      "loss": 4.7581,
+      "learning_rate": 0.0007623660346248969,
+      "loss": 5.4204,
       "step": 1656
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0002286479802143446,
-      "loss": 4.9705,
+      "learning_rate": 0.0007621599340478153,
+      "loss": 6.1318,
       "step": 1657
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0002285861500412201,
-      "loss": 4.8156,
+      "learning_rate": 0.0007619538334707337,
+      "loss": 5.4465,
       "step": 1658
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0002285243198680956,
-      "loss": 4.6403,
+      "learning_rate": 0.000761747732893652,
+      "loss": 5.6286,
       "step": 1659
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022846248969497112,
-      "loss": 4.5337,
+      "learning_rate": 0.0007615416323165704,
+      "loss": 5.2668,
       "step": 1660
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022840065952184664,
-      "loss": 4.69,
+      "learning_rate": 0.0007613355317394888,
+      "loss": 5.377,
       "step": 1661
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022833882934872215,
-      "loss": 4.5631,
+      "learning_rate": 0.0007611294311624072,
+      "loss": 5.299,
       "step": 1662
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0002282769991755977,
-      "loss": 4.3062,
+      "learning_rate": 0.0007609233305853257,
+      "loss": 5.1236,
       "step": 1663
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022821516900247319,
-      "loss": 4.7193,
+      "learning_rate": 0.000760717230008244,
+      "loss": 5.4572,
       "step": 1664
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0002281533388293487,
-      "loss": 4.3753,
+      "learning_rate": 0.0007605111294311624,
+      "loss": 5.1316,
       "step": 1665
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022809150865622422,
-      "loss": 4.7836,
+      "learning_rate": 0.0007603050288540808,
+      "loss": 5.3337,
       "step": 1666
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022802967848309974,
-      "loss": 4.8642,
+      "learning_rate": 0.0007600989282769992,
+      "loss": 5.2521,
       "step": 1667
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.00022796784830997522,
-      "loss": 4.3909,
+      "learning_rate": 0.0007598928276999175,
+      "loss": 5.8702,
       "step": 1668
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022790601813685074,
-      "loss": 4.4484,
+      "learning_rate": 0.0007596867271228359,
+      "loss": 5.0664,
       "step": 1669
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022784418796372629,
-      "loss": 4.7756,
+      "learning_rate": 0.0007594806265457544,
+      "loss": 5.5366,
       "step": 1670
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0002277823577906018,
-      "loss": 5.1696,
+      "learning_rate": 0.0007592745259686728,
+      "loss": 5.7692,
       "step": 1671
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022772052761747732,
-      "loss": 5.0716,
+      "learning_rate": 0.0007590684253915912,
+      "loss": 5.7895,
       "step": 1672
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0002276586974443528,
-      "loss": 4.6068,
+      "learning_rate": 0.0007588623248145095,
+      "loss": 5.4638,
       "step": 1673
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022759686727122832,
-      "loss": 4.6949,
+      "learning_rate": 0.0007586562242374279,
+      "loss": 5.4016,
       "step": 1674
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022753503709810387,
-      "loss": 4.974,
+      "learning_rate": 0.0007584501236603463,
+      "loss": 5.6734,
       "step": 1675
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022747320692497938,
-      "loss": 4.6888,
+      "learning_rate": 0.0007582440230832647,
+      "loss": 5.2043,
       "step": 1676
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0002274113767518549,
-      "loss": 4.3438,
+      "learning_rate": 0.0007580379225061831,
+      "loss": 5.0709,
       "step": 1677
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0002273495465787304,
-      "loss": 4.5404,
+      "learning_rate": 0.0007578318219291014,
+      "loss": 5.1183,
       "step": 1678
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0002272877164056059,
-      "loss": 5.1475,
+      "learning_rate": 0.0007576257213520198,
+      "loss": 5.7026,
       "step": 1679
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022722588623248142,
-      "loss": 5.1513,
+      "learning_rate": 0.0007574196207749382,
+      "loss": 5.6478,
       "step": 1680
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022716405605935697,
-      "loss": 4.267,
+      "learning_rate": 0.0007572135201978566,
+      "loss": 4.9254,
       "step": 1681
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022710222588623246,
-      "loss": 4.463,
+      "learning_rate": 0.000757007419620775,
+      "loss": 5.1116,
       "step": 1682
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.00022704039571310797,
-      "loss": 4.8971,
+      "learning_rate": 0.0007568013190436934,
+      "loss": 5.3446,
       "step": 1683
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0002269785655399835,
-      "loss": 4.56,
+      "learning_rate": 0.0007565952184666118,
+      "loss": 5.5533,
       "step": 1684
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.000226916735366859,
-      "loss": 4.6839,
+      "learning_rate": 0.0007563891178895302,
+      "loss": 5.3844,
       "step": 1685
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022685490519373455,
-      "loss": 4.5981,
+      "learning_rate": 0.0007561830173124486,
+      "loss": 5.2712,
       "step": 1686
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022679307502061004,
-      "loss": 4.9082,
+      "learning_rate": 0.0007559769167353669,
+      "loss": 5.4178,
       "step": 1687
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022673124484748556,
-      "loss": 4.9814,
+      "learning_rate": 0.0007557708161582853,
+      "loss": 5.7833,
       "step": 1688
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022666941467436107,
-      "loss": 4.6975,
+      "learning_rate": 0.0007555647155812037,
+      "loss": 5.5157,
       "step": 1689
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0002266075845012366,
-      "loss": 4.356,
+      "learning_rate": 0.0007553586150041221,
+      "loss": 5.03,
       "step": 1690
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022654575432811208,
-      "loss": 4.9605,
+      "learning_rate": 0.0007551525144270404,
+      "loss": 5.6973,
       "step": 1691
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0002264839241549876,
-      "loss": 4.7913,
+      "learning_rate": 0.0007549464138499588,
+      "loss": 5.3835,
       "step": 1692
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022642209398186314,
-      "loss": 4.7187,
+      "learning_rate": 0.0007547403132728772,
+      "loss": 5.5059,
       "step": 1693
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022636026380873866,
-      "loss": 4.8995,
+      "learning_rate": 0.0007545342126957956,
+      "loss": 5.5242,
       "step": 1694
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022629843363561417,
-      "loss": 4.8143,
+      "learning_rate": 0.000754328112118714,
+      "loss": 5.5533,
       "step": 1695
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022623660346248966,
-      "loss": 6.8823,
+      "learning_rate": 0.0007541220115416323,
+      "loss": 7.746,
       "step": 1696
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022617477328936518,
-      "loss": 4.5134,
+      "learning_rate": 0.0007539159109645507,
+      "loss": 5.1658,
       "step": 1697
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0002261129431162407,
-      "loss": 4.9737,
+      "learning_rate": 0.0007537098103874691,
+      "loss": 5.7214,
       "step": 1698
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022605111294311624,
-      "loss": 4.8491,
+      "learning_rate": 0.0007535037098103875,
+      "loss": 5.5684,
       "step": 1699
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022598928276999173,
-      "loss": 4.7937,
+      "learning_rate": 0.0007532976092333058,
+      "loss": 5.3929,
       "step": 1700
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022592745259686725,
-      "loss": 5.4718,
+      "learning_rate": 0.0007530915086562242,
+      "loss": 6.357,
       "step": 1701
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022586562242374276,
-      "loss": 4.8439,
+      "learning_rate": 0.0007528854080791426,
+      "loss": 5.4245,
       "step": 1702
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.00022580379225061828,
-      "loss": 4.7129,
+      "learning_rate": 0.000752679307502061,
+      "loss": 5.6498,
       "step": 1703
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022574196207749382,
-      "loss": 4.8185,
+      "learning_rate": 0.0007524732069249795,
+      "loss": 5.4691,
       "step": 1704
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0002256801319043693,
-      "loss": 4.5881,
+      "learning_rate": 0.0007522671063478978,
+      "loss": 5.4461,
       "step": 1705
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022561830173124483,
-      "loss": 4.6521,
+      "learning_rate": 0.0007520610057708162,
+      "loss": 5.2842,
       "step": 1706
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022555647155812034,
-      "loss": 4.4691,
+      "learning_rate": 0.0007518549051937346,
+      "loss": 5.3931,
       "step": 1707
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022549464138499586,
-      "loss": 4.9391,
+      "learning_rate": 0.000751648804616653,
+      "loss": 5.5527,
       "step": 1708
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022543281121187138,
-      "loss": 5.3869,
+      "learning_rate": 0.0007514427040395714,
+      "loss": 5.9373,
       "step": 1709
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022537098103874687,
-      "loss": 5.2916,
+      "learning_rate": 0.0007512366034624897,
+      "loss": 5.7114,
       "step": 1710
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0002253091508656224,
-      "loss": 4.4308,
+      "learning_rate": 0.0007510305028854081,
+      "loss": 5.0905,
       "step": 1711
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022524732069249793,
-      "loss": 4.7349,
+      "learning_rate": 0.0007508244023083265,
+      "loss": 5.7914,
       "step": 1712
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022518549051937344,
-      "loss": 5.1511,
+      "learning_rate": 0.0007506183017312449,
+      "loss": 5.8535,
       "step": 1713
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022512366034624893,
-      "loss": 4.5357,
+      "learning_rate": 0.0007504122011541632,
+      "loss": 5.4343,
       "step": 1714
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022506183017312445,
-      "loss": 4.7004,
+      "learning_rate": 0.0007502061005770816,
+      "loss": 5.2519,
       "step": 1715
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.000225,
-      "loss": 4.9619,
+      "learning_rate": 0.00075,
+      "loss": 5.8742,
       "step": 1716
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0002249381698268755,
-      "loss": 5.3511,
+      "learning_rate": 0.0007497938994229184,
+      "loss": 6.315,
       "step": 1717
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022487633965375103,
-      "loss": 5.4614,
+      "learning_rate": 0.0007495877988458368,
+      "loss": 6.4051,
       "step": 1718
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022481450948062652,
-      "loss": 4.8409,
+      "learning_rate": 0.0007493816982687551,
+      "loss": 5.6608,
       "step": 1719
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022475267930750203,
-      "loss": 4.7349,
+      "learning_rate": 0.0007491755976916735,
+      "loss": 5.3271,
       "step": 1720
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00022469084913437755,
-      "loss": 4.7739,
+      "learning_rate": 0.0007489694971145919,
+      "loss": 5.3743,
       "step": 1721
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0002246290189612531,
-      "loss": 5.5148,
+      "learning_rate": 0.0007487633965375103,
+      "loss": 5.8151,
       "step": 1722
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022456718878812858,
-      "loss": 5.6121,
+      "learning_rate": 0.0007485572959604286,
+      "loss": 5.7325,
       "step": 1723
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0002245053586150041,
-      "loss": 4.6997,
+      "learning_rate": 0.000748351195383347,
+      "loss": 5.299,
       "step": 1724
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022444352844187962,
-      "loss": 5.1495,
+      "learning_rate": 0.0007481450948062655,
+      "loss": 5.4698,
       "step": 1725
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022438169826875513,
-      "loss": 4.7757,
+      "learning_rate": 0.0007479389942291839,
+      "loss": 5.2479,
       "step": 1726
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022431986809563065,
-      "loss": 4.8949,
+      "learning_rate": 0.0007477328936521023,
+      "loss": 5.463,
       "step": 1727
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022425803792250614,
-      "loss": 4.6909,
+      "learning_rate": 0.0007475267930750206,
+      "loss": 5.4187,
       "step": 1728
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022419620774938168,
-      "loss": 5.1149,
+      "learning_rate": 0.000747320692497939,
+      "loss": 5.8931,
       "step": 1729
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0002241343775762572,
-      "loss": 5.2097,
+      "learning_rate": 0.0007471145919208574,
+      "loss": 5.362,
       "step": 1730
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022407254740313272,
-      "loss": 4.2226,
+      "learning_rate": 0.0007469084913437758,
+      "loss": 4.8648,
       "step": 1731
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022401071723000823,
-      "loss": 4.4025,
+      "learning_rate": 0.0007467023907666942,
+      "loss": 5.1222,
       "step": 1732
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022394888705688372,
-      "loss": 4.6662,
+      "learning_rate": 0.0007464962901896125,
+      "loss": 5.3404,
       "step": 1733
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022388705688375927,
-      "loss": 4.6814,
+      "learning_rate": 0.0007462901896125309,
+      "loss": 5.6978,
       "step": 1734
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022382522671063478,
-      "loss": 5.3791,
+      "learning_rate": 0.0007460840890354493,
+      "loss": 6.3196,
       "step": 1735
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0002237633965375103,
-      "loss": 5.4108,
+      "learning_rate": 0.0007458779884583677,
+      "loss": 6.147,
       "step": 1736
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0002237015663643858,
-      "loss": 4.4935,
+      "learning_rate": 0.000745671887881286,
+      "loss": 5.0887,
       "step": 1737
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0002236397361912613,
-      "loss": 5.0918,
+      "learning_rate": 0.0007454657873042044,
+      "loss": 5.7822,
       "step": 1738
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.00022357790601813682,
-      "loss": 4.7295,
+      "learning_rate": 0.0007452596867271228,
+      "loss": 5.5801,
       "step": 1739
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022351607584501236,
-      "loss": 4.8135,
+      "learning_rate": 0.0007450535861500412,
+      "loss": 5.4264,
       "step": 1740
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022345424567188788,
-      "loss": 5.7369,
+      "learning_rate": 0.0007448474855729596,
+      "loss": 5.5353,
       "step": 1741
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022339241549876337,
-      "loss": 4.6621,
+      "learning_rate": 0.0007446413849958779,
+      "loss": 5.3254,
       "step": 1742
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0002233305853256389,
-      "loss": 4.5457,
+      "learning_rate": 0.0007444352844187963,
+      "loss": 5.1066,
       "step": 1743
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0002232687551525144,
-      "loss": 4.7555,
+      "learning_rate": 0.0007442291838417148,
+      "loss": 5.354,
       "step": 1744
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022320692497938992,
-      "loss": 4.5796,
+      "learning_rate": 0.0007440230832646332,
+      "loss": 5.5587,
       "step": 1745
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0002231450948062654,
-      "loss": 4.6419,
+      "learning_rate": 0.0007438169826875515,
+      "loss": 5.246,
       "step": 1746
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022308326463314095,
-      "loss": 4.7337,
+      "learning_rate": 0.0007436108821104699,
+      "loss": 5.4677,
       "step": 1747
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022302143446001647,
-      "loss": 4.3085,
+      "learning_rate": 0.0007434047815333883,
+      "loss": 5.0217,
       "step": 1748
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.000222959604286892,
-      "loss": 4.5737,
+      "learning_rate": 0.0007431986809563067,
+      "loss": 5.1716,
       "step": 1749
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0002228977741137675,
-      "loss": 4.9717,
+      "learning_rate": 0.0007429925803792251,
+      "loss": 5.1908,
       "step": 1750
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.000222835943940643,
-      "loss": 4.3358,
+      "learning_rate": 0.0007427864798021434,
+      "loss": 5.039,
       "step": 1751
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022277411376751854,
-      "loss": 4.4861,
+      "learning_rate": 0.0007425803792250618,
+      "loss": 5.1832,
       "step": 1752
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022271228359439405,
-      "loss": 4.7721,
+      "learning_rate": 0.0007423742786479802,
+      "loss": 5.5627,
       "step": 1753
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022265045342126957,
-      "loss": 5.7925,
+      "learning_rate": 0.0007421681780708986,
+      "loss": 5.8069,
       "step": 1754
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022258862324814506,
-      "loss": 4.9722,
+      "learning_rate": 0.0007419620774938169,
+      "loss": 5.5625,
       "step": 1755
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.00022252679307502058,
-      "loss": 4.9091,
+      "learning_rate": 0.0007417559769167353,
+      "loss": 5.3716,
       "step": 1756
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0002224649629018961,
-      "loss": 4.6,
+      "learning_rate": 0.0007415498763396537,
+      "loss": 5.1685,
       "step": 1757
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022240313272877164,
-      "loss": 4.3669,
+      "learning_rate": 0.0007413437757625721,
+      "loss": 5.0729,
       "step": 1758
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022234130255564715,
-      "loss": 5.1315,
+      "learning_rate": 0.0007411376751854905,
+      "loss": 5.6822,
       "step": 1759
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022227947238252264,
-      "loss": 4.8058,
+      "learning_rate": 0.0007409315746084088,
+      "loss": 5.5346,
       "step": 1760
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022221764220939816,
-      "loss": 5.5005,
+      "learning_rate": 0.0007407254740313272,
+      "loss": 5.764,
       "step": 1761
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022215581203627368,
-      "loss": 4.3977,
+      "learning_rate": 0.0007405193734542456,
+      "loss": 5.0768,
       "step": 1762
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022209398186314922,
-      "loss": 4.6684,
+      "learning_rate": 0.0007403132728771642,
+      "loss": 5.4564,
       "step": 1763
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022203215169002474,
-      "loss": 5.2128,
+      "learning_rate": 0.0007401071723000826,
+      "loss": 5.6846,
       "step": 1764
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022197032151690023,
-      "loss": 4.7541,
+      "learning_rate": 0.0007399010717230009,
+      "loss": 5.3839,
       "step": 1765
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022190849134377574,
-      "loss": 4.7662,
+      "learning_rate": 0.0007396949711459193,
+      "loss": 5.5391,
       "step": 1766
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022184666117065126,
-      "loss": 4.6003,
+      "learning_rate": 0.0007394888705688377,
+      "loss": 5.1325,
       "step": 1767
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022178483099752678,
-      "loss": 4.5575,
+      "learning_rate": 0.0007392827699917561,
+      "loss": 5.3311,
       "step": 1768
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022172300082440226,
-      "loss": 4.5706,
+      "learning_rate": 0.0007390766694146744,
+      "loss": 5.3466,
       "step": 1769
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0002216611706512778,
-      "loss": 4.7971,
+      "learning_rate": 0.0007388705688375928,
+      "loss": 5.223,
       "step": 1770
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022159934047815332,
-      "loss": 4.956,
+      "learning_rate": 0.0007386644682605112,
+      "loss": 5.2403,
       "step": 1771
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022153751030502884,
-      "loss": 4.576,
+      "learning_rate": 0.0007384583676834296,
+      "loss": 5.1288,
       "step": 1772
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022147568013190436,
-      "loss": 4.5982,
+      "learning_rate": 0.000738252267106348,
+      "loss": 5.2811,
       "step": 1773
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022141384995877985,
-      "loss": 4.7445,
+      "learning_rate": 0.0007380461665292663,
+      "loss": 5.3473,
       "step": 1774
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.00022135201978565536,
-      "loss": 4.4654,
+      "learning_rate": 0.0007378400659521847,
+      "loss": 5.0818,
       "step": 1775
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002212901896125309,
-      "loss": 4.6078,
+      "learning_rate": 0.0007376339653751031,
+      "loss": 5.2832,
       "step": 1776
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022122835943940642,
-      "loss": 5.2675,
+      "learning_rate": 0.0007374278647980215,
+      "loss": 5.9785,
       "step": 1777
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022116652926628191,
-      "loss": 4.551,
+      "learning_rate": 0.0007372217642209398,
+      "loss": 5.2032,
       "step": 1778
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022110469909315743,
-      "loss": 5.7148,
+      "learning_rate": 0.0007370156636438582,
+      "loss": 6.1154,
       "step": 1779
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022104286892003295,
-      "loss": 4.5429,
+      "learning_rate": 0.0007368095630667766,
+      "loss": 5.2006,
       "step": 1780
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002209810387469085,
-      "loss": 5.5278,
+      "learning_rate": 0.000736603462489695,
+      "loss": 6.3009,
       "step": 1781
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.000220919208573784,
-      "loss": 4.5851,
+      "learning_rate": 0.0007363973619126134,
+      "loss": 5.3086,
       "step": 1782
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002208573784006595,
-      "loss": 5.3418,
+      "learning_rate": 0.0007361912613355317,
+      "loss": 5.3209,
       "step": 1783
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.000220795548227535,
-      "loss": 4.1075,
+      "learning_rate": 0.0007359851607584502,
+      "loss": 4.4866,
       "step": 1784
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022073371805441053,
-      "loss": 4.7177,
+      "learning_rate": 0.0007357790601813686,
+      "loss": 6.7161,
       "step": 1785
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022067188788128605,
-      "loss": 4.6688,
+      "learning_rate": 0.000735572959604287,
+      "loss": 6.2056,
       "step": 1786
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022061005770816154,
-      "loss": 4.1265,
+      "learning_rate": 0.0007353668590272053,
+      "loss": 5.1341,
       "step": 1787
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022054822753503708,
-      "loss": 4.1144,
+      "learning_rate": 0.0007351607584501237,
+      "loss": 5.0733,
       "step": 1788
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002204863973619126,
-      "loss": 4.3359,
+      "learning_rate": 0.0007349546578730421,
+      "loss": 5.3742,
       "step": 1789
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0002204245671887881,
-      "loss": 5.0456,
+      "learning_rate": 0.0007347485572959605,
+      "loss": 5.6744,
       "step": 1790
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022036273701566363,
-      "loss": 4.1368,
+      "learning_rate": 0.0007345424567188789,
+      "loss": 5.0179,
       "step": 1791
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.00022030090684253912,
-      "loss": 4.1232,
+      "learning_rate": 0.0007343363561417972,
+      "loss": 4.9713,
       "step": 1792
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00022023907666941466,
-      "loss": 4.841,
+      "learning_rate": 0.0007341302555647156,
+      "loss": 5.4606,
       "step": 1793
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00022017724649629018,
-      "loss": 4.1225,
+      "learning_rate": 0.000733924154987634,
+      "loss": 4.9745,
       "step": 1794
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0002201154163231657,
-      "loss": 4.3076,
+      "learning_rate": 0.0007337180544105524,
+      "loss": 5.1346,
       "step": 1795
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0002200535861500412,
-      "loss": 4.135,
+      "learning_rate": 0.0007335119538334708,
+      "loss": 5.2538,
       "step": 1796
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0002199917559769167,
-      "loss": 3.8696,
+      "learning_rate": 0.0007333058532563891,
+      "loss": 4.751,
       "step": 1797
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021992992580379222,
-      "loss": 4.3157,
+      "learning_rate": 0.0007330997526793075,
+      "loss": 5.1878,
       "step": 1798
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021986809563066776,
-      "loss": 4.3958,
+      "learning_rate": 0.0007328936521022259,
+      "loss": 4.9865,
       "step": 1799
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021980626545754328,
-      "loss": 4.2767,
+      "learning_rate": 0.0007326875515251443,
+      "loss": 5.3912,
       "step": 1800
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021974443528441877,
-      "loss": 4.2781,
+      "learning_rate": 0.0007324814509480626,
+      "loss": 5.1692,
       "step": 1801
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021968260511129428,
-      "loss": 4.158,
+      "learning_rate": 0.000732275350370981,
+      "loss": 4.8574,
       "step": 1802
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0002196207749381698,
-      "loss": 4.2403,
+      "learning_rate": 0.0007320692497938994,
+      "loss": 5.0755,
       "step": 1803
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021955894476504532,
-      "loss": 4.4887,
+      "learning_rate": 0.0007318631492168179,
+      "loss": 5.7155,
       "step": 1804
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021949711459192086,
-      "loss": 4.0098,
+      "learning_rate": 0.0007316570486397363,
+      "loss": 4.696,
       "step": 1805
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021943528441879635,
-      "loss": 4.1686,
+      "learning_rate": 0.0007314509480626546,
+      "loss": 5.0782,
       "step": 1806
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021937345424567187,
-      "loss": 3.9112,
+      "learning_rate": 0.000731244847485573,
+      "loss": 4.7626,
       "step": 1807
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.00021931162407254738,
-      "loss": 4.134,
+      "learning_rate": 0.0007310387469084914,
+      "loss": 5.4785,
       "step": 1808
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0002192497938994229,
-      "loss": 3.962,
+      "learning_rate": 0.0007308326463314098,
+      "loss": 5.4731,
       "step": 1809
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0002191879637262984,
-      "loss": 4.2424,
+      "learning_rate": 0.0007306265457543281,
+      "loss": 4.8763,
       "step": 1810
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021912613355317393,
-      "loss": 4.2122,
+      "learning_rate": 0.0007304204451772465,
+      "loss": 5.1314,
       "step": 1811
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021906430338004945,
-      "loss": 4.3248,
+      "learning_rate": 0.0007302143446001649,
+      "loss": 6.0777,
       "step": 1812
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021900247320692497,
-      "loss": 4.1252,
+      "learning_rate": 0.0007300082440230833,
+      "loss": 4.9886,
       "step": 1813
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021894064303380048,
-      "loss": 3.6728,
+      "learning_rate": 0.0007298021434460017,
+      "loss": 4.5431,
       "step": 1814
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021887881286067597,
-      "loss": 4.0904,
+      "learning_rate": 0.00072959604286892,
+      "loss": 5.2027,
       "step": 1815
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0002188169826875515,
-      "loss": 4.6337,
+      "learning_rate": 0.0007293899422918384,
+      "loss": 5.7955,
       "step": 1816
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021875515251442703,
-      "loss": 4.1198,
+      "learning_rate": 0.0007291838417147568,
+      "loss": 5.5344,
       "step": 1817
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021869332234130255,
-      "loss": 3.9191,
+      "learning_rate": 0.0007289777411376752,
+      "loss": 5.1074,
       "step": 1818
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021863149216817804,
-      "loss": 4.3493,
+      "learning_rate": 0.0007287716405605935,
+      "loss": 5.263,
       "step": 1819
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021856966199505356,
-      "loss": 4.1351,
+      "learning_rate": 0.0007285655399835119,
+      "loss": 5.099,
       "step": 1820
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021850783182192907,
-      "loss": 4.2987,
+      "learning_rate": 0.0007283594394064303,
+      "loss": 5.2997,
       "step": 1821
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0002184460016488046,
-      "loss": 3.9335,
+      "learning_rate": 0.0007281533388293487,
+      "loss": 4.9187,
       "step": 1822
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021838417147568013,
-      "loss": 4.269,
+      "learning_rate": 0.0007279472382522671,
+      "loss": 4.8737,
       "step": 1823
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021832234130255562,
-      "loss": 4.7815,
+      "learning_rate": 0.0007277411376751854,
+      "loss": 5.8766,
       "step": 1824
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021826051112943114,
-      "loss": 4.0027,
+      "learning_rate": 0.0007275350370981039,
+      "loss": 4.8703,
       "step": 1825
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021819868095630666,
-      "loss": 4.3558,
+      "learning_rate": 0.0007273289365210223,
+      "loss": 5.8658,
       "step": 1826
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021813685078318217,
-      "loss": 4.0046,
+      "learning_rate": 0.0007271228359439407,
+      "loss": 4.8,
       "step": 1827
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00021807502061005772,
-      "loss": 4.4823,
+      "learning_rate": 0.0007269167353668591,
+      "loss": 5.0083,
       "step": 1828
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0002180131904369332,
-      "loss": 4.2326,
+      "learning_rate": 0.0007267106347897774,
+      "loss": 5.0207,
       "step": 1829
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021795136026380872,
-      "loss": 4.308,
+      "learning_rate": 0.0007265045342126958,
+      "loss": 5.5562,
       "step": 1830
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021788953009068424,
-      "loss": 3.3861,
+      "learning_rate": 0.0007262984336356142,
+      "loss": 4.4522,
       "step": 1831
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021782769991755976,
-      "loss": 4.0877,
+      "learning_rate": 0.0007260923330585326,
+      "loss": 4.9096,
       "step": 1832
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021776586974443524,
-      "loss": 7.8572,
+      "learning_rate": 0.0007258862324814509,
+      "loss": 8.2927,
       "step": 1833
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021770403957131076,
-      "loss": 3.9283,
+      "learning_rate": 0.0007256801319043693,
+      "loss": 4.8999,
       "step": 1834
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0002176422093981863,
-      "loss": 4.4063,
+      "learning_rate": 0.0007254740313272877,
+      "loss": 5.2236,
       "step": 1835
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021758037922506182,
-      "loss": 4.5952,
+      "learning_rate": 0.0007252679307502061,
+      "loss": 5.3805,
       "step": 1836
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021751854905193734,
-      "loss": 4.3643,
+      "learning_rate": 0.0007250618301731245,
+      "loss": 5.0225,
       "step": 1837
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021745671887881283,
-      "loss": 4.2647,
+      "learning_rate": 0.0007248557295960428,
+      "loss": 5.0855,
       "step": 1838
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021739488870568834,
-      "loss": 4.045,
+      "learning_rate": 0.0007246496290189612,
+      "loss": 4.7451,
       "step": 1839
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0002173330585325639,
-      "loss": 4.2011,
+      "learning_rate": 0.0007244435284418796,
+      "loss": 5.0579,
       "step": 1840
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0002172712283594394,
-      "loss": 4.2884,
+      "learning_rate": 0.000724237427864798,
+      "loss": 5.1009,
       "step": 1841
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0002172093981863149,
-      "loss": 4.322,
+      "learning_rate": 0.0007240313272877163,
+      "loss": 5.0356,
       "step": 1842
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0002171475680131904,
-      "loss": 4.4743,
+      "learning_rate": 0.0007238252267106347,
+      "loss": 5.2618,
       "step": 1843
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021708573784006593,
-      "loss": 5.0363,
+      "learning_rate": 0.0007236191261335531,
+      "loss": 5.8649,
       "step": 1844
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00021702390766694144,
-      "loss": 4.3998,
+      "learning_rate": 0.0007234130255564716,
+      "loss": 5.5624,
       "step": 1845
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.000216962077493817,
-      "loss": 3.962,
+      "learning_rate": 0.00072320692497939,
+      "loss": 4.8541,
       "step": 1846
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021690024732069248,
-      "loss": 4.7073,
+      "learning_rate": 0.0007230008244023083,
+      "loss": 5.504,
       "step": 1847
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.000216838417147568,
-      "loss": 4.2439,
+      "learning_rate": 0.0007227947238252267,
+      "loss": 4.9489,
       "step": 1848
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0002167765869744435,
-      "loss": 4.3758,
+      "learning_rate": 0.0007225886232481451,
+      "loss": 5.0252,
       "step": 1849
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021671475680131903,
-      "loss": 4.3838,
+      "learning_rate": 0.0007223825226710635,
+      "loss": 5.3229,
       "step": 1850
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021665292662819454,
-      "loss": 3.844,
+      "learning_rate": 0.0007221764220939819,
+      "loss": 4.7599,
       "step": 1851
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021659109645507003,
-      "loss": 4.6697,
+      "learning_rate": 0.0007219703215169002,
+      "loss": 5.318,
       "step": 1852
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021652926628194558,
-      "loss": 3.9667,
+      "learning_rate": 0.0007217642209398186,
+      "loss": 4.7916,
       "step": 1853
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0002164674361088211,
-      "loss": 4.0571,
+      "learning_rate": 0.000721558120362737,
+      "loss": 4.9588,
       "step": 1854
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0002164056059356966,
-      "loss": 4.0903,
+      "learning_rate": 0.0007213520197856554,
+      "loss": 5.0499,
       "step": 1855
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0002163437757625721,
-      "loss": 4.1325,
+      "learning_rate": 0.0007211459192085737,
+      "loss": 4.9607,
       "step": 1856
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021628194558944762,
-      "loss": 4.1481,
+      "learning_rate": 0.0007209398186314922,
+      "loss": 4.879,
       "step": 1857
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021622011541632316,
-      "loss": 4.1682,
+      "learning_rate": 0.0007207337180544106,
+      "loss": 4.9978,
       "step": 1858
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021615828524319868,
-      "loss": 4.0307,
+      "learning_rate": 0.000720527617477329,
+      "loss": 4.8292,
       "step": 1859
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0002160964550700742,
-      "loss": 4.036,
+      "learning_rate": 0.0007203215169002474,
+      "loss": 4.8893,
       "step": 1860
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021603462489694968,
-      "loss": 3.8698,
+      "learning_rate": 0.0007201154163231657,
+      "loss": 4.6447,
       "step": 1861
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0002159727947238252,
-      "loss": 4.1114,
+      "learning_rate": 0.0007199093157460841,
+      "loss": 5.1184,
       "step": 1862
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021591096455070072,
-      "loss": 4.1051,
+      "learning_rate": 0.0007197032151690025,
+      "loss": 4.8806,
       "step": 1863
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.00021584913437757626,
-      "loss": 4.244,
+      "learning_rate": 0.000719497114591921,
+      "loss": 5.1212,
       "step": 1864
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021578730420445175,
-      "loss": 4.3088,
+      "learning_rate": 0.0007192910140148393,
+      "loss": 5.2627,
       "step": 1865
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021572547403132726,
-      "loss": 3.6972,
+      "learning_rate": 0.0007190849134377577,
+      "loss": 4.5504,
       "step": 1866
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021566364385820278,
-      "loss": 4.0327,
+      "learning_rate": 0.0007188788128606761,
+      "loss": 5.1228,
       "step": 1867
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0002156018136850783,
-      "loss": 4.0826,
+      "learning_rate": 0.0007186727122835945,
+      "loss": 4.8928,
       "step": 1868
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021553998351195384,
-      "loss": 4.0307,
+      "learning_rate": 0.0007184666117065129,
+      "loss": 4.8402,
       "step": 1869
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021547815333882933,
-      "loss": 4.5499,
+      "learning_rate": 0.0007182605111294312,
+      "loss": 5.755,
       "step": 1870
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021541632316570485,
-      "loss": 4.1259,
+      "learning_rate": 0.0007180544105523496,
+      "loss": 5.0116,
       "step": 1871
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021535449299258036,
-      "loss": 4.3174,
+      "learning_rate": 0.000717848309975268,
+      "loss": 5.0736,
       "step": 1872
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021529266281945588,
-      "loss": 3.7573,
+      "learning_rate": 0.0007176422093981864,
+      "loss": 4.6837,
       "step": 1873
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021523083264633137,
-      "loss": 3.9238,
+      "learning_rate": 0.0007174361088211047,
+      "loss": 4.7889,
       "step": 1874
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0002151690024732069,
-      "loss": 4.066,
+      "learning_rate": 0.0007172300082440231,
+      "loss": 5.0305,
       "step": 1875
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021510717230008243,
-      "loss": 3.9913,
+      "learning_rate": 0.0007170239076669415,
+      "loss": 4.941,
       "step": 1876
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021504534212695795,
-      "loss": 4.1772,
+      "learning_rate": 0.0007168178070898599,
+      "loss": 5.0131,
       "step": 1877
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021498351195383346,
-      "loss": 3.6977,
+      "learning_rate": 0.0007166117065127783,
+      "loss": 5.0236,
       "step": 1878
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021492168178070895,
-      "loss": 4.1588,
+      "learning_rate": 0.0007164056059356966,
+      "loss": 5.0443,
       "step": 1879
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021485985160758447,
-      "loss": 3.5456,
+      "learning_rate": 0.000716199505358615,
+      "loss": 4.3573,
       "step": 1880
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021479802143446,
-      "loss": 3.6997,
+      "learning_rate": 0.0007159934047815334,
+      "loss": 4.7753,
       "step": 1881
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.00021473619126133553,
-      "loss": 4.262,
+      "learning_rate": 0.0007157873042044518,
+      "loss": 5.1045,
       "step": 1882
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021467436108821105,
-      "loss": 3.4603,
+      "learning_rate": 0.0007155812036273702,
+      "loss": 4.3955,
       "step": 1883
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021461253091508654,
-      "loss": 3.8694,
+      "learning_rate": 0.0007153751030502885,
+      "loss": 4.8774,
       "step": 1884
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021455070074196205,
-      "loss": 4.0525,
+      "learning_rate": 0.000715169002473207,
+      "loss": 4.8209,
       "step": 1885
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021448887056883757,
-      "loss": 4.092,
+      "learning_rate": 0.0007149629018961254,
+      "loss": 4.8963,
       "step": 1886
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0002144270403957131,
-      "loss": 4.4829,
+      "learning_rate": 0.0007147568013190438,
+      "loss": 5.3112,
       "step": 1887
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0002143652102225886,
-      "loss": 4.1764,
+      "learning_rate": 0.0007145507007419621,
+      "loss": 5.2001,
       "step": 1888
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021430338004946412,
-      "loss": 3.9978,
+      "learning_rate": 0.0007143446001648805,
+      "loss": 4.7737,
       "step": 1889
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021424154987633964,
-      "loss": 4.4018,
+      "learning_rate": 0.0007141384995877989,
+      "loss": 5.4572,
       "step": 1890
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021417971970321515,
-      "loss": 3.9842,
+      "learning_rate": 0.0007139323990107173,
+      "loss": 4.9596,
       "step": 1891
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021411788953009067,
-      "loss": 4.233,
+      "learning_rate": 0.0007137262984336357,
+      "loss": 5.1131,
       "step": 1892
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021405605935696616,
-      "loss": 4.3937,
+      "learning_rate": 0.000713520197856554,
+      "loss": 5.2216,
       "step": 1893
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0002139942291838417,
-      "loss": 4.1085,
+      "learning_rate": 0.0007133140972794724,
+      "loss": 4.8685,
       "step": 1894
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021393239901071722,
-      "loss": 4.1898,
+      "learning_rate": 0.0007131079967023908,
+      "loss": 5.2497,
       "step": 1895
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021387056883759274,
-      "loss": 4.1343,
+      "learning_rate": 0.0007129018961253092,
+      "loss": 4.9546,
       "step": 1896
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021380873866446822,
-      "loss": 4.0235,
+      "learning_rate": 0.0007126957955482275,
+      "loss": 4.9165,
       "step": 1897
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021374690849134374,
-      "loss": 4.3047,
+      "learning_rate": 0.0007124896949711459,
+      "loss": 5.1838,
       "step": 1898
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.00021368507831821926,
-      "loss": 4.016,
+      "learning_rate": 0.0007122835943940643,
+      "loss": 4.8545,
       "step": 1899
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0002136232481450948,
-      "loss": 4.1397,
+      "learning_rate": 0.0007120774938169827,
+      "loss": 4.8862,
       "step": 1900
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021356141797197032,
-      "loss": 4.7137,
+      "learning_rate": 0.0007118713932399011,
+      "loss": 5.5258,
       "step": 1901
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0002134995877988458,
-      "loss": 4.0906,
+      "learning_rate": 0.0007116652926628194,
+      "loss": 4.909,
       "step": 1902
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021343775762572132,
-      "loss": 4.208,
+      "learning_rate": 0.0007114591920857378,
+      "loss": 5.1176,
       "step": 1903
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021337592745259684,
-      "loss": 4.3633,
+      "learning_rate": 0.0007112530915086562,
+      "loss": 5.3091,
       "step": 1904
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021331409727947238,
-      "loss": 4.3346,
+      "learning_rate": 0.0007110469909315747,
+      "loss": 5.2653,
       "step": 1905
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021325226710634787,
-      "loss": 4.9483,
+      "learning_rate": 0.000710840890354493,
+      "loss": 5.7553,
       "step": 1906
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0002131904369332234,
-      "loss": 4.4207,
+      "learning_rate": 0.0007106347897774114,
+      "loss": 5.3612,
       "step": 1907
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0002131286067600989,
-      "loss": 4.3347,
+      "learning_rate": 0.0007104286892003298,
+      "loss": 5.1875,
       "step": 1908
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021306677658697442,
-      "loss": 4.164,
+      "learning_rate": 0.0007102225886232482,
+      "loss": 4.9192,
       "step": 1909
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021300494641384994,
-      "loss": 4.2876,
+      "learning_rate": 0.0007100164880461666,
+      "loss": 5.0839,
       "step": 1910
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021294311624072543,
-      "loss": 4.2924,
+      "learning_rate": 0.0007098103874690849,
+      "loss": 5.1741,
       "step": 1911
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021288128606760097,
-      "loss": 4.1621,
+      "learning_rate": 0.0007096042868920033,
+      "loss": 5.1112,
       "step": 1912
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0002128194558944765,
-      "loss": 4.1517,
+      "learning_rate": 0.0007093981863149217,
+      "loss": 4.9319,
       "step": 1913
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.000212757625721352,
-      "loss": 4.5683,
+      "learning_rate": 0.0007091920857378401,
+      "loss": 5.1723,
       "step": 1914
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021269579554822752,
-      "loss": 4.3971,
+      "learning_rate": 0.0007089859851607585,
+      "loss": 5.0431,
       "step": 1915
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.000212633965375103,
-      "loss": 4.0739,
+      "learning_rate": 0.0007087798845836768,
+      "loss": 5.134,
       "step": 1916
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.00021257213520197856,
-      "loss": 4.0623,
+      "learning_rate": 0.0007085737840065952,
+      "loss": 4.9249,
       "step": 1917
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021251030502885407,
-      "loss": 4.3568,
+      "learning_rate": 0.0007083676834295136,
+      "loss": 5.2112,
       "step": 1918
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0002124484748557296,
-      "loss": 3.8684,
+      "learning_rate": 0.000708161582852432,
+      "loss": 5.1078,
       "step": 1919
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021238664468260508,
-      "loss": 3.684,
+      "learning_rate": 0.0007079554822753503,
+      "loss": 4.7082,
       "step": 1920
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0002123248145094806,
-      "loss": 3.9499,
+      "learning_rate": 0.0007077493816982687,
+      "loss": 4.741,
       "step": 1921
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0002122629843363561,
-      "loss": 4.6862,
+      "learning_rate": 0.0007075432811211871,
+      "loss": 5.6072,
       "step": 1922
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021220115416323166,
-      "loss": 4.4907,
+      "learning_rate": 0.0007073371805441055,
+      "loss": 5.1473,
       "step": 1923
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021213932399010717,
-      "loss": 4.1284,
+      "learning_rate": 0.0007071310799670239,
+      "loss": 4.9798,
       "step": 1924
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021207749381698266,
-      "loss": 4.4415,
+      "learning_rate": 0.0007069249793899422,
+      "loss": 5.5453,
       "step": 1925
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021201566364385818,
-      "loss": 4.0495,
+      "learning_rate": 0.0007067188788128607,
+      "loss": 4.8624,
       "step": 1926
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0002119538334707337,
-      "loss": 3.8226,
+      "learning_rate": 0.0007065127782357791,
+      "loss": 4.7663,
       "step": 1927
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0002118920032976092,
-      "loss": 4.1506,
+      "learning_rate": 0.0007063066776586975,
+      "loss": 4.9708,
       "step": 1928
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0002118301731244847,
-      "loss": 3.6491,
+      "learning_rate": 0.0007061005770816158,
+      "loss": 4.6775,
       "step": 1929
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021176834295136025,
-      "loss": 3.4903,
+      "learning_rate": 0.0007058944765045342,
+      "loss": 4.3418,
       "step": 1930
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021170651277823576,
-      "loss": 3.3366,
+      "learning_rate": 0.0007056883759274526,
+      "loss": 4.2575,
       "step": 1931
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021164468260511128,
-      "loss": 4.5944,
+      "learning_rate": 0.000705482275350371,
+      "loss": 6.3119,
       "step": 1932
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0002115828524319868,
-      "loss": 3.5268,
+      "learning_rate": 0.0007052761747732894,
+      "loss": 4.4464,
       "step": 1933
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021152102225886228,
-      "loss": 4.4154,
+      "learning_rate": 0.0007050700741962077,
+      "loss": 6.336,
       "step": 1934
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.00021145919208573783,
-      "loss": 4.2783,
+      "learning_rate": 0.0007048639736191261,
+      "loss": 5.2734,
       "step": 1935
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021139736191261334,
-      "loss": 4.1529,
+      "learning_rate": 0.0007046578730420445,
+      "loss": 4.9526,
       "step": 1936
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021133553173948886,
-      "loss": 4.2383,
+      "learning_rate": 0.0007044517724649629,
+      "loss": 5.1503,
       "step": 1937
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021127370156636438,
-      "loss": 4.9958,
+      "learning_rate": 0.0007042456718878813,
+      "loss": 6.6474,
       "step": 1938
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021121187139323987,
-      "loss": 4.1661,
+      "learning_rate": 0.0007040395713107996,
+      "loss": 5.0745,
       "step": 1939
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021115004122011538,
-      "loss": 4.6984,
+      "learning_rate": 0.000703833470733718,
+      "loss": 5.6552,
       "step": 1940
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021108821104699093,
-      "loss": 3.9306,
+      "learning_rate": 0.0007036273701566364,
+      "loss": 4.8594,
       "step": 1941
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021102638087386644,
-      "loss": 4.6782,
+      "learning_rate": 0.0007034212695795548,
+      "loss": 5.5089,
       "step": 1942
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021096455070074193,
-      "loss": 4.5104,
+      "learning_rate": 0.0007032151690024731,
+      "loss": 5.8453,
       "step": 1943
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021090272052761745,
-      "loss": 4.3137,
+      "learning_rate": 0.0007030090684253915,
+      "loss": 5.1221,
       "step": 1944
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021084089035449297,
-      "loss": 4.122,
+      "learning_rate": 0.0007028029678483099,
+      "loss": 5.1654,
       "step": 1945
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0002107790601813685,
-      "loss": 4.649,
+      "learning_rate": 0.0007025968672712284,
+      "loss": 4.8595,
       "step": 1946
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021071723000824403,
-      "loss": 4.4324,
+      "learning_rate": 0.0007023907666941468,
+      "loss": 5.4633,
       "step": 1947
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021065539983511952,
-      "loss": 4.1716,
+      "learning_rate": 0.0007021846661170651,
+      "loss": 5.1242,
       "step": 1948
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021059356966199503,
-      "loss": 4.2012,
+      "learning_rate": 0.0007019785655399835,
+      "loss": 5.2278,
       "step": 1949
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021053173948887055,
-      "loss": 4.1668,
+      "learning_rate": 0.000701772464962902,
+      "loss": 4.8614,
       "step": 1950
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021046990931574607,
-      "loss": 4.4321,
+      "learning_rate": 0.0007015663643858204,
+      "loss": 5.3163,
       "step": 1951
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.00021040807914262156,
-      "loss": 3.7523,
+      "learning_rate": 0.0007013602638087387,
+      "loss": 4.778,
       "step": 1952
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0002103462489694971,
-      "loss": 4.3832,
+      "learning_rate": 0.0007011541632316571,
+      "loss": 5.4249,
       "step": 1953
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00021028441879637262,
-      "loss": 4.0424,
+      "learning_rate": 0.0007009480626545755,
+      "loss": 5.0991,
       "step": 1954
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00021022258862324813,
-      "loss": 4.1957,
+      "learning_rate": 0.0007007419620774939,
+      "loss": 5.1253,
       "step": 1955
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00021016075845012365,
-      "loss": 3.9432,
+      "learning_rate": 0.0007005358615004123,
+      "loss": 5.0108,
       "step": 1956
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00021009892827699914,
-      "loss": 4.1075,
+      "learning_rate": 0.0007003297609233306,
+      "loss": 5.2935,
       "step": 1957
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00021003709810387466,
-      "loss": 4.1551,
+      "learning_rate": 0.000700123660346249,
+      "loss": 5.1025,
       "step": 1958
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0002099752679307502,
-      "loss": 4.4526,
+      "learning_rate": 0.0006999175597691674,
+      "loss": 4.9619,
       "step": 1959
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020991343775762572,
-      "loss": 3.9584,
+      "learning_rate": 0.0006997114591920858,
+      "loss": 4.8111,
       "step": 1960
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0002098516075845012,
-      "loss": 3.7979,
+      "learning_rate": 0.0006995053586150041,
+      "loss": 4.9499,
       "step": 1961
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020978977741137672,
-      "loss": 3.8536,
+      "learning_rate": 0.0006992992580379225,
+      "loss": 4.6626,
       "step": 1962
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020972794723825224,
-      "loss": 4.1592,
+      "learning_rate": 0.0006990931574608409,
+      "loss": 4.9205,
       "step": 1963
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020966611706512778,
-      "loss": 4.3766,
+      "learning_rate": 0.0006988870568837593,
+      "loss": 5.0247,
       "step": 1964
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0002096042868920033,
-      "loss": 4.3715,
+      "learning_rate": 0.0006986809563066778,
+      "loss": 5.6156,
       "step": 1965
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0002095424567188788,
-      "loss": 3.777,
+      "learning_rate": 0.000698474855729596,
+      "loss": 4.6375,
       "step": 1966
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0002094806265457543,
-      "loss": 3.972,
+      "learning_rate": 0.0006982687551525145,
+      "loss": 4.965,
       "step": 1967
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020941879637262982,
-      "loss": 3.7467,
+      "learning_rate": 0.0006980626545754329,
+      "loss": 4.5958,
       "step": 1968
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020935696619950534,
-      "loss": 4.267,
+      "learning_rate": 0.0006978565539983513,
+      "loss": 5.2138,
       "step": 1969
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020929513602638088,
-      "loss": 3.9663,
+      "learning_rate": 0.0006976504534212697,
+      "loss": 4.931,
       "step": 1970
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.00020923330585325637,
-      "loss": 4.1087,
+      "learning_rate": 0.000697444352844188,
+      "loss": 5.1836,
       "step": 1971
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0002091714756801319,
-      "loss": 4.004,
+      "learning_rate": 0.0006972382522671064,
+      "loss": 4.9763,
       "step": 1972
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0002091096455070074,
-      "loss": 4.031,
+      "learning_rate": 0.0006970321516900248,
+      "loss": 4.8877,
       "step": 1973
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020904781533388292,
-      "loss": 3.8294,
+      "learning_rate": 0.0006968260511129432,
+      "loss": 4.745,
       "step": 1974
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0002089859851607584,
-      "loss": 4.3503,
+      "learning_rate": 0.0006966199505358615,
+      "loss": 5.3999,
       "step": 1975
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020892415498763395,
-      "loss": 3.7831,
+      "learning_rate": 0.0006964138499587799,
+      "loss": 4.8253,
       "step": 1976
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020886232481450947,
-      "loss": 4.0451,
+      "learning_rate": 0.0006962077493816983,
+      "loss": 4.8059,
       "step": 1977
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.000208800494641385,
-      "loss": 3.7531,
+      "learning_rate": 0.0006960016488046167,
+      "loss": 4.557,
       "step": 1978
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0002087386644682605,
-      "loss": 3.5045,
+      "learning_rate": 0.0006957955482275351,
+      "loss": 4.3461,
       "step": 1979
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.000208676834295136,
-      "loss": 3.8253,
+      "learning_rate": 0.0006955894476504534,
+      "loss": 5.3581,
       "step": 1980
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0002086150041220115,
-      "loss": 4.4512,
+      "learning_rate": 0.0006953833470733718,
+      "loss": 5.1943,
       "step": 1981
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020855317394888705,
-      "loss": 4.3466,
+      "learning_rate": 0.0006951772464962902,
+      "loss": 5.4875,
       "step": 1982
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020849134377576257,
-      "loss": 3.8365,
+      "learning_rate": 0.0006949711459192086,
+      "loss": 4.9124,
       "step": 1983
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020842951360263806,
-      "loss": 3.3622,
+      "learning_rate": 0.0006947650453421269,
+      "loss": 4.2013,
       "step": 1984
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020836768342951358,
-      "loss": 4.3552,
+      "learning_rate": 0.0006945589447650453,
+      "loss": 5.0194,
       "step": 1985
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0002083058532563891,
-      "loss": 4.0351,
+      "learning_rate": 0.0006943528441879638,
+      "loss": 5.304,
       "step": 1986
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0002082440230832646,
-      "loss": 4.1183,
+      "learning_rate": 0.0006941467436108822,
+      "loss": 5.1742,
       "step": 1987
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020818219291014015,
-      "loss": 4.2601,
+      "learning_rate": 0.0006939406430338006,
+      "loss": 5.8334,
       "step": 1988
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.00020812036273701564,
-      "loss": 4.3821,
+      "learning_rate": 0.0006937345424567189,
+      "loss": 5.0403,
       "step": 1989
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020805853256389116,
-      "loss": 4.2085,
+      "learning_rate": 0.0006935284418796373,
+      "loss": 5.1716,
       "step": 1990
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020799670239076668,
-      "loss": 4.2899,
+      "learning_rate": 0.0006933223413025557,
+      "loss": 5.2278,
       "step": 1991
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0002079348722176422,
-      "loss": 3.9023,
+      "learning_rate": 0.0006931162407254741,
+      "loss": 4.8494,
       "step": 1992
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020787304204451768,
-      "loss": 3.9813,
+      "learning_rate": 0.0006929101401483924,
+      "loss": 4.968,
       "step": 1993
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020781121187139323,
-      "loss": 4.5191,
+      "learning_rate": 0.0006927040395713108,
+      "loss": 5.0556,
       "step": 1994
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020774938169826874,
-      "loss": 4.1332,
+      "learning_rate": 0.0006924979389942292,
+      "loss": 5.0608,
       "step": 1995
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020768755152514426,
-      "loss": 4.3867,
+      "learning_rate": 0.0006922918384171476,
+      "loss": 5.3535,
       "step": 1996
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020762572135201978,
-      "loss": 4.2042,
+      "learning_rate": 0.000692085737840066,
+      "loss": 5.0817,
       "step": 1997
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020756389117889526,
-      "loss": 4.1924,
+      "learning_rate": 0.0006918796372629843,
+      "loss": 5.3395,
       "step": 1998
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020750206100577078,
-      "loss": 4.2637,
+      "learning_rate": 0.0006916735366859027,
+      "loss": 5.0693,
       "step": 1999
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020744023083264632,
-      "loss": 4.0588,
+      "learning_rate": 0.0006914674361088211,
+      "loss": 4.9162,
       "step": 2000
     },
     {
       "epoch": 1.12,
-      "eval_loss": 4.778836727142334,
-      "eval_runtime": 1247.5217,
-      "eval_samples_per_second": 2.118,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.8836969456564856,
+      "eval_loss": 5.641247272491455,
+      "eval_runtime": 1246.1483,
+      "eval_samples_per_second": 2.12,
+      "eval_steps_per_second": 0.266,
+      "eval_wer": 1.9732844109480365,
       "step": 2000
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020737840065952184,
-      "loss": 4.4025,
+      "learning_rate": 0.0006912613355317395,
+      "loss": 5.1554,
       "step": 2001
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020731657048639736,
-      "loss": 3.8561,
+      "learning_rate": 0.0006910552349546579,
+      "loss": 4.8891,
       "step": 2002
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020725474031327285,
-      "loss": 4.0224,
+      "learning_rate": 0.0006908491343775762,
+      "loss": 4.9826,
       "step": 2003
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020719291014014836,
-      "loss": 4.0755,
+      "learning_rate": 0.0006906430338004946,
+      "loss": 5.1483,
       "step": 2004
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020713107996702388,
-      "loss": 4.1765,
+      "learning_rate": 0.000690436933223413,
+      "loss": 4.7722,
       "step": 2005
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020706924979389942,
-      "loss": 3.7585,
+      "learning_rate": 0.0006902308326463315,
+      "loss": 4.915,
       "step": 2006
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.00020700741962077491,
-      "loss": 3.9571,
+      "learning_rate": 0.0006900247320692497,
+      "loss": 4.7808,
       "step": 2007
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020694558944765043,
-      "loss": 4.094,
+      "learning_rate": 0.0006898186314921682,
+      "loss": 5.1847,
       "step": 2008
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020688375927452595,
-      "loss": 4.0756,
+      "learning_rate": 0.0006896125309150866,
+      "loss": 4.9068,
       "step": 2009
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020682192910140146,
-      "loss": 3.9382,
+      "learning_rate": 0.000689406430338005,
+      "loss": 4.7811,
       "step": 2010
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.000206760098928277,
-      "loss": 3.7819,
+      "learning_rate": 0.0006892003297609234,
+      "loss": 4.8156,
       "step": 2011
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0002066982687551525,
-      "loss": 3.7945,
+      "learning_rate": 0.0006889942291838417,
+      "loss": 5.0743,
       "step": 2012
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.000206636438582028,
-      "loss": 4.9615,
+      "learning_rate": 0.0006887881286067601,
+      "loss": 5.2864,
       "step": 2013
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020657460840890353,
-      "loss": 4.9477,
+      "learning_rate": 0.0006885820280296785,
+      "loss": 5.6686,
       "step": 2014
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020651277823577905,
-      "loss": 4.0562,
+      "learning_rate": 0.0006883759274525969,
+      "loss": 5.014,
       "step": 2015
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020645094806265454,
-      "loss": 4.2115,
+      "learning_rate": 0.0006881698268755152,
+      "loss": 4.8173,
       "step": 2016
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020638911788953005,
-      "loss": 4.1687,
+      "learning_rate": 0.0006879637262984336,
+      "loss": 5.0305,
       "step": 2017
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0002063272877164056,
-      "loss": 4.1252,
+      "learning_rate": 0.000687757625721352,
+      "loss": 5.1696,
       "step": 2018
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0002062654575432811,
-      "loss": 3.9894,
+      "learning_rate": 0.0006875515251442704,
+      "loss": 4.9071,
       "step": 2019
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020620362737015663,
-      "loss": 4.0056,
+      "learning_rate": 0.0006873454245671888,
+      "loss": 5.0102,
       "step": 2020
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020614179719703212,
-      "loss": 4.7586,
+      "learning_rate": 0.0006871393239901071,
+      "loss": 5.1051,
       "step": 2021
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020607996702390764,
-      "loss": 3.8911,
+      "learning_rate": 0.0006869332234130255,
+      "loss": 4.6792,
       "step": 2022
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.00020601813685078318,
-      "loss": 3.8066,
+      "learning_rate": 0.0006867271228359439,
+      "loss": 4.7521,
       "step": 2023
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0002059563066776587,
-      "loss": 4.1624,
+      "learning_rate": 0.0006865210222588623,
+      "loss": 5.682,
       "step": 2024
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020589447650453419,
-      "loss": 4.5219,
+      "learning_rate": 0.0006863149216817806,
+      "loss": 5.6318,
       "step": 2025
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0002058326463314097,
-      "loss": 3.8806,
+      "learning_rate": 0.000686108821104699,
+      "loss": 5.7699,
       "step": 2026
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020577081615828522,
-      "loss": 4.3836,
+      "learning_rate": 0.0006859027205276174,
+      "loss": 5.8706,
       "step": 2027
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020570898598516074,
-      "loss": 3.7542,
+      "learning_rate": 0.0006856966199505359,
+      "loss": 4.7844,
       "step": 2028
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020564715581203628,
-      "loss": 4.4337,
+      "learning_rate": 0.0006854905193734543,
+      "loss": 6.2101,
       "step": 2029
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020558532563891177,
-      "loss": 4.1687,
+      "learning_rate": 0.0006852844187963726,
+      "loss": 5.4375,
       "step": 2030
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020552349546578728,
-      "loss": 4.9655,
+      "learning_rate": 0.000685078318219291,
+      "loss": 6.7288,
       "step": 2031
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0002054616652926628,
-      "loss": 3.4128,
+      "learning_rate": 0.0006848722176422094,
+      "loss": 4.4745,
       "step": 2032
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020539983511953832,
-      "loss": 3.8979,
+      "learning_rate": 0.0006846661170651278,
+      "loss": 5.2756,
       "step": 2033
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020533800494641383,
-      "loss": 4.8252,
+      "learning_rate": 0.0006844600164880462,
+      "loss": 4.9741,
       "step": 2034
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020527617477328932,
-      "loss": 4.2699,
+      "learning_rate": 0.0006842539159109645,
+      "loss": 5.3267,
       "step": 2035
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020521434460016487,
-      "loss": 4.1037,
+      "learning_rate": 0.0006840478153338829,
+      "loss": 5.0628,
       "step": 2036
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020515251442704038,
-      "loss": 4.5303,
+      "learning_rate": 0.0006838417147568013,
+      "loss": 5.4387,
       "step": 2037
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0002050906842539159,
-      "loss": 4.1389,
+      "learning_rate": 0.0006836356141797197,
+      "loss": 4.8829,
       "step": 2038
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0002050288540807914,
-      "loss": 4.2016,
+      "learning_rate": 0.000683429513602638,
+      "loss": 5.2361,
       "step": 2039
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0002049670239076669,
-      "loss": 4.6772,
+      "learning_rate": 0.0006832234130255564,
+      "loss": 5.438,
       "step": 2040
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020490519373454245,
-      "loss": 3.8612,
+      "learning_rate": 0.0006830173124484748,
+      "loss": 4.7598,
       "step": 2041
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.00020484336356141797,
-      "loss": 4.3163,
+      "learning_rate": 0.0006828112118713932,
+      "loss": 5.3378,
       "step": 2042
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020478153338829348,
-      "loss": 4.0103,
+      "learning_rate": 0.0006826051112943117,
+      "loss": 4.9224,
       "step": 2043
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020471970321516897,
-      "loss": 3.8704,
+      "learning_rate": 0.00068239901071723,
+      "loss": 4.8368,
       "step": 2044
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0002046578730420445,
-      "loss": 4.6009,
+      "learning_rate": 0.0006821929101401484,
+      "loss": 5.5691,
       "step": 2045
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020459604286892,
-      "loss": 4.262,
+      "learning_rate": 0.0006819868095630669,
+      "loss": 5.0673,
       "step": 2046
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020453421269579555,
-      "loss": 3.8992,
+      "learning_rate": 0.0006817807089859853,
+      "loss": 4.8679,
       "step": 2047
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020447238252267104,
-      "loss": 4.8274,
+      "learning_rate": 0.0006815746084089036,
+      "loss": 5.7343,
       "step": 2048
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020441055234954656,
-      "loss": 4.3125,
+      "learning_rate": 0.000681368507831822,
+      "loss": 5.1696,
       "step": 2049
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020434872217642207,
-      "loss": 4.3951,
+      "learning_rate": 0.0006811624072547404,
+      "loss": 5.115,
       "step": 2050
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0002042868920032976,
-      "loss": 3.9244,
+      "learning_rate": 0.0006809563066776588,
+      "loss": 4.8737,
       "step": 2051
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020422506183017313,
-      "loss": 3.9818,
+      "learning_rate": 0.0006807502061005772,
+      "loss": 4.8922,
       "step": 2052
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020416323165704862,
-      "loss": 4.2277,
+      "learning_rate": 0.0006805441055234955,
+      "loss": 5.2827,
       "step": 2053
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020410140148392414,
-      "loss": 3.8742,
+      "learning_rate": 0.0006803380049464139,
+      "loss": 4.832,
       "step": 2054
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020403957131079966,
-      "loss": 4.1545,
+      "learning_rate": 0.0006801319043693323,
+      "loss": 5.1816,
       "step": 2055
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020397774113767517,
-      "loss": 3.9582,
+      "learning_rate": 0.0006799258037922507,
+      "loss": 4.8719,
       "step": 2056
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0002039159109645507,
-      "loss": 3.9679,
+      "learning_rate": 0.0006797197032151691,
+      "loss": 4.848,
       "step": 2057
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020385408079142618,
-      "loss": 3.941,
+      "learning_rate": 0.0006795136026380874,
+      "loss": 4.8054,
       "step": 2058
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020379225061830172,
-      "loss": 4.4527,
+      "learning_rate": 0.0006793075020610058,
+      "loss": 5.3394,
       "step": 2059
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00020373042044517724,
-      "loss": 4.0415,
+      "learning_rate": 0.0006791014014839242,
+      "loss": 4.8862,
       "step": 2060
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020366859027205276,
-      "loss": 5.157,
+      "learning_rate": 0.0006788953009068426,
+      "loss": 5.4533,
       "step": 2061
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020360676009892824,
-      "loss": 4.0606,
+      "learning_rate": 0.0006786892003297609,
+      "loss": 5.2488,
       "step": 2062
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020354492992580376,
-      "loss": 3.9466,
+      "learning_rate": 0.0006784830997526793,
+      "loss": 4.9514,
       "step": 2063
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020348309975267928,
-      "loss": 3.9105,
+      "learning_rate": 0.0006782769991755977,
+      "loss": 4.6766,
       "step": 2064
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020342126957955482,
-      "loss": 4.1702,
+      "learning_rate": 0.0006780708985985161,
+      "loss": 5.0576,
       "step": 2065
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020335943940643034,
-      "loss": 4.2173,
+      "learning_rate": 0.0006778647980214346,
+      "loss": 5.5898,
       "step": 2066
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020329760923330583,
-      "loss": 3.8298,
+      "learning_rate": 0.0006776586974443528,
+      "loss": 4.7287,
       "step": 2067
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020323577906018134,
-      "loss": 4.318,
+      "learning_rate": 0.0006774525968672713,
+      "loss": 5.0363,
       "step": 2068
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020317394888705686,
-      "loss": 4.1233,
+      "learning_rate": 0.0006772464962901897,
+      "loss": 5.0711,
       "step": 2069
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0002031121187139324,
-      "loss": 4.1238,
+      "learning_rate": 0.0006770403957131081,
+      "loss": 4.9878,
       "step": 2070
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0002030502885408079,
-      "loss": 3.9864,
+      "learning_rate": 0.0006768342951360264,
+      "loss": 4.9903,
       "step": 2071
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0002029884583676834,
-      "loss": 4.6548,
+      "learning_rate": 0.0006766281945589448,
+      "loss": 5.771,
       "step": 2072
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020292662819455893,
-      "loss": 3.8653,
+      "learning_rate": 0.0006764220939818632,
+      "loss": 4.6962,
       "step": 2073
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020286479802143444,
-      "loss": 3.8062,
+      "learning_rate": 0.0006762159934047816,
+      "loss": 4.6339,
       "step": 2074
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020280296784830996,
-      "loss": 3.9825,
+      "learning_rate": 0.0006760098928277,
+      "loss": 4.9984,
       "step": 2075
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.00020274113767518545,
-      "loss": 4.8792,
+      "learning_rate": 0.0006758037922506183,
+      "loss": 5.1957,
       "step": 2076
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.000202679307502061,
-      "loss": 4.3456,
+      "learning_rate": 0.0006755976916735367,
+      "loss": 5.4708,
       "step": 2077
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0002026174773289365,
-      "loss": 4.2937,
+      "learning_rate": 0.0006753915910964551,
+      "loss": 5.0499,
       "step": 2078
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020255564715581203,
-      "loss": 3.5597,
+      "learning_rate": 0.0006751854905193735,
+      "loss": 4.6799,
       "step": 2079
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020249381698268752,
-      "loss": 4.9338,
+      "learning_rate": 0.0006749793899422918,
+      "loss": 6.6565,
       "step": 2080
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020243198680956303,
-      "loss": 4.1119,
+      "learning_rate": 0.0006747732893652102,
+      "loss": 5.3778,
       "step": 2081
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020237015663643855,
-      "loss": 3.6265,
+      "learning_rate": 0.0006745671887881286,
+      "loss": 4.9116,
       "step": 2082
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0002023083264633141,
-      "loss": 3.8201,
+      "learning_rate": 0.000674361088211047,
+      "loss": 5.4526,
       "step": 2083
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0002022464962901896,
-      "loss": 3.8632,
+      "learning_rate": 0.0006741549876339654,
+      "loss": 7.7358,
       "step": 2084
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0002021846661170651,
-      "loss": 4.3041,
+      "learning_rate": 0.0006739488870568837,
+      "loss": 5.1698,
       "step": 2085
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020212283594394062,
-      "loss": 3.921,
+      "learning_rate": 0.0006737427864798021,
+      "loss": 4.6365,
       "step": 2086
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020206100577081613,
-      "loss": 4.2238,
+      "learning_rate": 0.0006735366859027205,
+      "loss": 5.1635,
       "step": 2087
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020199917559769168,
-      "loss": 4.1807,
+      "learning_rate": 0.000673330585325639,
+      "loss": 5.0961,
       "step": 2088
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0002019373454245672,
-      "loss": 3.9639,
+      "learning_rate": 0.0006731244847485574,
+      "loss": 4.9722,
       "step": 2089
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020187551525144268,
-      "loss": 4.4557,
+      "learning_rate": 0.0006729183841714757,
+      "loss": 5.4982,
       "step": 2090
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0002018136850783182,
-      "loss": 4.2668,
+      "learning_rate": 0.0006727122835943941,
+      "loss": 5.1925,
       "step": 2091
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020175185490519372,
-      "loss": 4.0918,
+      "learning_rate": 0.0006725061830173125,
+      "loss": 5.236,
       "step": 2092
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020169002473206923,
-      "loss": 4.3044,
+      "learning_rate": 0.0006723000824402309,
+      "loss": 5.0232,
       "step": 2093
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020162819455894472,
-      "loss": 4.0166,
+      "learning_rate": 0.0006720939818631492,
+      "loss": 5.0245,
       "step": 2094
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020156636438582027,
-      "loss": 4.2291,
+      "learning_rate": 0.0006718878812860676,
+      "loss": 5.1335,
       "step": 2095
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.00020150453421269578,
-      "loss": 4.2311,
+      "learning_rate": 0.000671681780708986,
+      "loss": 4.9398,
       "step": 2096
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0002014427040395713,
-      "loss": 4.3602,
+      "learning_rate": 0.0006714756801319044,
+      "loss": 5.2128,
       "step": 2097
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020138087386644681,
-      "loss": 4.1202,
+      "learning_rate": 0.0006712695795548228,
+      "loss": 5.4953,
       "step": 2098
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0002013190436933223,
-      "loss": 4.5616,
+      "learning_rate": 0.0006710634789777411,
+      "loss": 5.675,
       "step": 2099
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020125721352019785,
-      "loss": 3.8499,
+      "learning_rate": 0.0006708573784006595,
+      "loss": 4.7282,
       "step": 2100
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020119538334707336,
-      "loss": 4.1582,
+      "learning_rate": 0.0006706512778235779,
+      "loss": 5.0345,
       "step": 2101
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020113355317394888,
-      "loss": 3.9711,
+      "learning_rate": 0.0006704451772464963,
+      "loss": 4.8078,
       "step": 2102
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020107172300082437,
-      "loss": 3.838,
+      "learning_rate": 0.0006702390766694146,
+      "loss": 4.7974,
       "step": 2103
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0002010098928276999,
-      "loss": 4.233,
+      "learning_rate": 0.000670032976092333,
+      "loss": 5.9797,
       "step": 2104
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0002009480626545754,
-      "loss": 4.1528,
+      "learning_rate": 0.0006698268755152514,
+      "loss": 5.2685,
       "step": 2105
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020088623248145095,
-      "loss": 4.2584,
+      "learning_rate": 0.0006696207749381698,
+      "loss": 5.1761,
       "step": 2106
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020082440230832646,
-      "loss": 4.3581,
+      "learning_rate": 0.0006694146743610882,
+      "loss": 5.2984,
       "step": 2107
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020076257213520195,
-      "loss": 4.4746,
+      "learning_rate": 0.0006692085737840065,
+      "loss": 5.0798,
       "step": 2108
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020070074196207747,
-      "loss": 4.4105,
+      "learning_rate": 0.000669002473206925,
+      "loss": 5.2807,
       "step": 2109
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.000200638911788953,
-      "loss": 3.9595,
+      "learning_rate": 0.0006687963726298434,
+      "loss": 4.8416,
       "step": 2110
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0002005770816158285,
-      "loss": 4.2329,
+      "learning_rate": 0.0006685902720527618,
+      "loss": 5.1076,
       "step": 2111
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.000200515251442704,
-      "loss": 4.1554,
+      "learning_rate": 0.0006683841714756801,
+      "loss": 5.5306,
       "step": 2112
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020045342126957954,
-      "loss": 4.1002,
+      "learning_rate": 0.0006681780708985985,
+      "loss": 5.152,
       "step": 2113
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.00020039159109645505,
-      "loss": 3.923,
+      "learning_rate": 0.0006679719703215169,
+      "loss": 4.957,
       "step": 2114
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00020032976092333057,
-      "loss": 4.493,
+      "learning_rate": 0.0006677658697444353,
+      "loss": 5.2095,
       "step": 2115
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00020026793075020609,
-      "loss": 3.6723,
+      "learning_rate": 0.0006675597691673537,
+      "loss": 4.7728,
       "step": 2116
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00020020610057708158,
-      "loss": 3.6911,
+      "learning_rate": 0.000667353668590272,
+      "loss": 4.6956,
       "step": 2117
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00020014427040395712,
-      "loss": 3.9949,
+      "learning_rate": 0.0006671475680131904,
+      "loss": 5.2243,
       "step": 2118
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00020008244023083264,
-      "loss": 3.8763,
+      "learning_rate": 0.0006669414674361088,
+      "loss": 4.7681,
       "step": 2119
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00020002061005770815,
-      "loss": 4.3232,
+      "learning_rate": 0.0006667353668590272,
+      "loss": 5.1002,
       "step": 2120
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019995877988458367,
-      "loss": 4.2582,
+      "learning_rate": 0.0006665292662819456,
+      "loss": 5.2388,
       "step": 2121
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019989694971145916,
-      "loss": 4.0342,
+      "learning_rate": 0.0006663231657048639,
+      "loss": 4.9552,
       "step": 2122
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019983511953833468,
-      "loss": 4.1259,
+      "learning_rate": 0.0006661170651277823,
+      "loss": 4.9023,
       "step": 2123
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019977328936521022,
-      "loss": 3.8213,
+      "learning_rate": 0.0006659109645507007,
+      "loss": 4.7282,
       "step": 2124
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019971145919208574,
-      "loss": 4.4665,
+      "learning_rate": 0.0006657048639736191,
+      "loss": 5.5284,
       "step": 2125
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019964962901896123,
-      "loss": 4.4768,
+      "learning_rate": 0.0006654987633965374,
+      "loss": 5.4734,
       "step": 2126
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019958779884583674,
-      "loss": 3.7138,
+      "learning_rate": 0.0006652926628194558,
+      "loss": 4.7063,
       "step": 2127
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019952596867271226,
-      "loss": 4.1133,
+      "learning_rate": 0.0006650865622423742,
+      "loss": 5.0588,
       "step": 2128
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0001994641384995878,
-      "loss": 3.8248,
+      "learning_rate": 0.0006648804616652927,
+      "loss": 4.8522,
       "step": 2129
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.00019940230832646332,
-      "loss": 4.3557,
+      "learning_rate": 0.0006646743610882111,
+      "loss": 6.3917,
       "step": 2130
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0001993404781533388,
-      "loss": 3.6741,
+      "learning_rate": 0.0006644682605111294,
+      "loss": 4.5742,
       "step": 2131
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019927864798021432,
-      "loss": 4.0013,
+      "learning_rate": 0.0006642621599340478,
+      "loss": 5.0439,
       "step": 2132
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019921681780708984,
-      "loss": 4.9306,
+      "learning_rate": 0.0006640560593569662,
+      "loss": 5.0811,
       "step": 2133
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019915498763396536,
-      "loss": 3.3022,
+      "learning_rate": 0.0006638499587798846,
+      "loss": 4.8239,
       "step": 2134
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019909315746084085,
-      "loss": 4.2632,
+      "learning_rate": 0.0006636438582028029,
+      "loss": 5.4555,
       "step": 2135
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001990313272877164,
-      "loss": 4.2516,
+      "learning_rate": 0.0006634377576257213,
+      "loss": 5.2008,
       "step": 2136
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001989694971145919,
-      "loss": 4.0982,
+      "learning_rate": 0.0006632316570486398,
+      "loss": 4.9969,
       "step": 2137
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019890766694146742,
-      "loss": 3.8634,
+      "learning_rate": 0.0006630255564715582,
+      "loss": 4.7945,
       "step": 2138
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019884583676834294,
-      "loss": 3.8264,
+      "learning_rate": 0.0006628194558944766,
+      "loss": 4.7548,
       "step": 2139
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019878400659521843,
-      "loss": 4.7033,
+      "learning_rate": 0.0006626133553173949,
+      "loss": 5.4747,
       "step": 2140
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019872217642209395,
-      "loss": 4.574,
+      "learning_rate": 0.0006624072547403133,
+      "loss": 5.8072,
       "step": 2141
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001986603462489695,
-      "loss": 3.9794,
+      "learning_rate": 0.0006622011541632317,
+      "loss": 4.7993,
       "step": 2142
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.000198598516075845,
-      "loss": 4.0757,
+      "learning_rate": 0.0006619950535861501,
+      "loss": 5.079,
       "step": 2143
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001985366859027205,
-      "loss": 4.2983,
+      "learning_rate": 0.0006617889530090684,
+      "loss": 5.0109,
       "step": 2144
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.000198474855729596,
-      "loss": 4.3881,
+      "learning_rate": 0.0006615828524319868,
+      "loss": 5.3951,
       "step": 2145
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019841302555647153,
-      "loss": 4.2352,
+      "learning_rate": 0.0006613767518549052,
+      "loss": 5.3109,
       "step": 2146
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019835119538334707,
-      "loss": 4.0129,
+      "learning_rate": 0.0006611706512778237,
+      "loss": 4.9196,
       "step": 2147
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0001982893652102226,
-      "loss": 3.9713,
+      "learning_rate": 0.0006609645507007421,
+      "loss": 4.9524,
       "step": 2148
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.00019822753503709808,
-      "loss": 3.8948,
+      "learning_rate": 0.0006607584501236604,
+      "loss": 4.9983,
       "step": 2149
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001981657048639736,
-      "loss": 4.3957,
+      "learning_rate": 0.0006605523495465788,
+      "loss": 5.2353,
       "step": 2150
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001981038746908491,
-      "loss": 4.3111,
+      "learning_rate": 0.0006603462489694972,
+      "loss": 5.2597,
       "step": 2151
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019804204451772463,
-      "loss": 4.4165,
+      "learning_rate": 0.0006601401483924156,
+      "loss": 5.698,
       "step": 2152
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019798021434460017,
-      "loss": 4.0781,
+      "learning_rate": 0.000659934047815334,
+      "loss": 5.1041,
       "step": 2153
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019791838417147566,
-      "loss": 4.214,
+      "learning_rate": 0.0006597279472382523,
+      "loss": 5.1161,
       "step": 2154
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019785655399835118,
-      "loss": 4.3653,
+      "learning_rate": 0.0006595218466611707,
+      "loss": 5.1219,
       "step": 2155
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001977947238252267,
-      "loss": 3.9221,
+      "learning_rate": 0.0006593157460840891,
+      "loss": 4.8318,
       "step": 2156
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001977328936521022,
-      "loss": 4.3661,
+      "learning_rate": 0.0006591096455070075,
+      "loss": 5.456,
       "step": 2157
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001976710634789777,
-      "loss": 4.0103,
+      "learning_rate": 0.0006589035449299258,
+      "loss": 5.0469,
       "step": 2158
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019760923330585325,
-      "loss": 4.3102,
+      "learning_rate": 0.0006586974443528442,
+      "loss": 5.1517,
       "step": 2159
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019754740313272876,
-      "loss": 4.2616,
+      "learning_rate": 0.0006584913437757626,
+      "loss": 5.2521,
       "step": 2160
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019748557295960428,
-      "loss": 3.9891,
+      "learning_rate": 0.000658285243198681,
+      "loss": 4.861,
       "step": 2161
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001974237427864798,
-      "loss": 3.8054,
+      "learning_rate": 0.0006580791426215994,
+      "loss": 4.9641,
       "step": 2162
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019736191261335528,
-      "loss": 4.1011,
+      "learning_rate": 0.0006578730420445177,
+      "loss": 5.1267,
       "step": 2163
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0001973000824402308,
-      "loss": 3.9636,
+      "learning_rate": 0.0006576669414674361,
+      "loss": 4.8712,
       "step": 2164
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019723825226710634,
-      "loss": 4.146,
+      "learning_rate": 0.0006574608408903545,
+      "loss": 5.0347,
       "step": 2165
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019717642209398186,
-      "loss": 4.133,
+      "learning_rate": 0.0006572547403132729,
+      "loss": 5.0449,
       "step": 2166
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.00019711459192085735,
-      "loss": 4.4815,
+      "learning_rate": 0.0006570486397361912,
+      "loss": 5.3153,
       "step": 2167
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019705276174773287,
-      "loss": 3.8254,
+      "learning_rate": 0.0006568425391591096,
+      "loss": 4.8321,
       "step": 2168
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019699093157460838,
-      "loss": 4.0858,
+      "learning_rate": 0.0006566364385820281,
+      "loss": 5.093,
       "step": 2169
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0001969291014014839,
-      "loss": 4.1229,
+      "learning_rate": 0.0006564303380049465,
+      "loss": 5.1645,
       "step": 2170
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019686727122835944,
-      "loss": 3.8402,
+      "learning_rate": 0.0006562242374278649,
+      "loss": 4.7657,
       "step": 2171
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019680544105523493,
-      "loss": 4.7132,
+      "learning_rate": 0.0006560181368507832,
+      "loss": 5.6444,
       "step": 2172
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019674361088211045,
-      "loss": 3.8882,
+      "learning_rate": 0.0006558120362737016,
+      "loss": 4.9064,
       "step": 2173
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019668178070898597,
-      "loss": 3.8115,
+      "learning_rate": 0.00065560593569662,
+      "loss": 4.8151,
       "step": 2174
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019661995053586148,
-      "loss": 3.9414,
+      "learning_rate": 0.0006553998351195384,
+      "loss": 4.9157,
       "step": 2175
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019655812036273703,
-      "loss": 3.5885,
+      "learning_rate": 0.0006551937345424568,
+      "loss": 4.7286,
       "step": 2176
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019649629018961252,
-      "loss": 3.4554,
+      "learning_rate": 0.0006549876339653751,
+      "loss": 4.511,
       "step": 2177
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019643446001648803,
-      "loss": 4.5087,
+      "learning_rate": 0.0006547815333882935,
+      "loss": 5.6233,
       "step": 2178
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019637262984336355,
-      "loss": 3.4061,
+      "learning_rate": 0.0006545754328112119,
+      "loss": 4.5108,
       "step": 2179
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019631079967023907,
-      "loss": 3.9712,
+      "learning_rate": 0.0006543693322341303,
+      "loss": 5.164,
       "step": 2180
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019624896949711456,
-      "loss": 3.2876,
+      "learning_rate": 0.0006541632316570486,
+      "loss": 4.7171,
       "step": 2181
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019618713932399007,
-      "loss": 3.4065,
+      "learning_rate": 0.000653957131079967,
+      "loss": 4.5693,
       "step": 2182
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019612530915086562,
-      "loss": 4.8778,
+      "learning_rate": 0.0006537510305028854,
+      "loss": 6.1144,
       "step": 2183
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019606347897774113,
-      "loss": 3.1723,
+      "learning_rate": 0.0006535449299258038,
+      "loss": 4.3996,
       "step": 2184
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00019600164880461665,
-      "loss": 4.0891,
+      "learning_rate": 0.0006533388293487222,
+      "loss": 5.2977,
       "step": 2185
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019593981863149214,
-      "loss": 4.2706,
+      "learning_rate": 0.0006531327287716405,
+      "loss": 5.2022,
       "step": 2186
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019587798845836766,
-      "loss": 3.9619,
+      "learning_rate": 0.0006529266281945589,
+      "loss": 4.8768,
       "step": 2187
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019581615828524317,
-      "loss": 4.2374,
+      "learning_rate": 0.0006527205276174773,
+      "loss": 5.2843,
       "step": 2188
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019575432811211872,
-      "loss": 4.3053,
+      "learning_rate": 0.0006525144270403958,
+      "loss": 5.3472,
       "step": 2189
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001956924979389942,
-      "loss": 4.0289,
+      "learning_rate": 0.000652308326463314,
+      "loss": 5.0712,
       "step": 2190
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019563066776586972,
-      "loss": 4.087,
+      "learning_rate": 0.0006521022258862325,
+      "loss": 5.0313,
       "step": 2191
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019556883759274524,
-      "loss": 4.3771,
+      "learning_rate": 0.0006518961253091509,
+      "loss": 5.3047,
       "step": 2192
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019550700741962076,
-      "loss": 4.6316,
+      "learning_rate": 0.0006516900247320693,
+      "loss": 5.5777,
       "step": 2193
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001954451772464963,
-      "loss": 4.0698,
+      "learning_rate": 0.0006514839241549877,
+      "loss": 5.1147,
       "step": 2194
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001953833470733718,
-      "loss": 4.0881,
+      "learning_rate": 0.000651277823577906,
+      "loss": 5.0325,
       "step": 2195
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001953215169002473,
-      "loss": 4.1779,
+      "learning_rate": 0.0006510717230008244,
+      "loss": 5.1492,
       "step": 2196
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019525968672712282,
-      "loss": 4.0553,
+      "learning_rate": 0.0006508656224237428,
+      "loss": 5.1345,
       "step": 2197
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019519785655399834,
-      "loss": 4.4104,
+      "learning_rate": 0.0006506595218466612,
+      "loss": 5.3183,
       "step": 2198
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019513602638087383,
-      "loss": 3.9602,
+      "learning_rate": 0.0006504534212695795,
+      "loss": 4.9002,
       "step": 2199
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019507419620774934,
-      "loss": 4.5405,
+      "learning_rate": 0.0006502473206924979,
+      "loss": 5.5162,
       "step": 2200
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001950123660346249,
-      "loss": 4.1794,
+      "learning_rate": 0.0006500412201154163,
+      "loss": 5.4818,
       "step": 2201
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0001949505358615004,
-      "loss": 4.1668,
+      "learning_rate": 0.0006498351195383347,
+      "loss": 5.1126,
       "step": 2202
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.00019488870568837592,
-      "loss": 3.8171,
+      "learning_rate": 0.0006496290189612531,
+      "loss": 4.833,
       "step": 2203
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0001948268755152514,
-      "loss": 4.3467,
+      "learning_rate": 0.0006494229183841714,
+      "loss": 5.3174,
       "step": 2204
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019476504534212693,
-      "loss": 4.151,
+      "learning_rate": 0.0006492168178070898,
+      "loss": 5.2388,
       "step": 2205
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019470321516900247,
-      "loss": 3.9868,
+      "learning_rate": 0.0006490107172300082,
+      "loss": 4.9963,
       "step": 2206
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.000194641384995878,
-      "loss": 4.0134,
+      "learning_rate": 0.0006488046166529266,
+      "loss": 5.2855,
       "step": 2207
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0001945795548227535,
-      "loss": 3.8111,
+      "learning_rate": 0.000648598516075845,
+      "loss": 5.0522,
       "step": 2208
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.000194517724649629,
-      "loss": 3.8494,
+      "learning_rate": 0.0006483924154987633,
+      "loss": 4.8046,
       "step": 2209
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0001944558944765045,
-      "loss": 4.0535,
+      "learning_rate": 0.0006481863149216818,
+      "loss": 5.0021,
       "step": 2210
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019439406430338003,
-      "loss": 3.8847,
+      "learning_rate": 0.0006479802143446002,
+      "loss": 4.8862,
       "step": 2211
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019433223413025557,
-      "loss": 3.7672,
+      "learning_rate": 0.0006477741137675186,
+      "loss": 4.723,
       "step": 2212
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019427040395713106,
-      "loss": 4.5733,
+      "learning_rate": 0.0006475680131904369,
+      "loss": 6.2146,
       "step": 2213
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019420857378400658,
-      "loss": 3.9959,
+      "learning_rate": 0.0006473619126133553,
+      "loss": 5.2786,
       "step": 2214
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0001941467436108821,
-      "loss": 3.8117,
+      "learning_rate": 0.0006471558120362737,
+      "loss": 4.8828,
       "step": 2215
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0001940849134377576,
-      "loss": 3.889,
+      "learning_rate": 0.0006469497114591921,
+      "loss": 5.0016,
       "step": 2216
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019402308326463313,
-      "loss": 4.2598,
+      "learning_rate": 0.0006467436108821105,
+      "loss": 5.7356,
       "step": 2217
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019396125309150862,
-      "loss": 4.2156,
+      "learning_rate": 0.0006465375103050288,
+      "loss": 5.269,
       "step": 2218
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019389942291838416,
-      "loss": 4.1113,
+      "learning_rate": 0.0006463314097279472,
+      "loss": 5.1999,
       "step": 2219
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.00019383759274525968,
-      "loss": 4.0509,
+      "learning_rate": 0.0006461253091508656,
+      "loss": 5.0834,
       "step": 2220
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0001937757625721352,
-      "loss": 4.2341,
+      "learning_rate": 0.000645919208573784,
+      "loss": 5.1478,
       "step": 2221
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019371393239901068,
-      "loss": 3.8446,
+      "learning_rate": 0.0006457131079967023,
+      "loss": 4.9197,
       "step": 2222
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0001936521022258862,
-      "loss": 3.6567,
+      "learning_rate": 0.0006455070074196207,
+      "loss": 4.6722,
       "step": 2223
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019359027205276174,
-      "loss": 3.8297,
+      "learning_rate": 0.0006453009068425391,
+      "loss": 7.0552,
       "step": 2224
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019352844187963726,
-      "loss": 4.1757,
+      "learning_rate": 0.0006450948062654575,
+      "loss": 5.1267,
       "step": 2225
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019346661170651278,
-      "loss": 4.0292,
+      "learning_rate": 0.0006448887056883759,
+      "loss": 5.0003,
       "step": 2226
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019340478153338826,
-      "loss": 3.6909,
+      "learning_rate": 0.0006446826051112942,
+      "loss": 4.9738,
       "step": 2227
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019334295136026378,
-      "loss": 4.0493,
+      "learning_rate": 0.0006444765045342126,
+      "loss": 5.3063,
       "step": 2228
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0001932811211871393,
-      "loss": 4.0557,
+      "learning_rate": 0.000644270403957131,
+      "loss": 5.1288,
       "step": 2229
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019321929101401484,
-      "loss": 4.4261,
+      "learning_rate": 0.0006440643033800496,
+      "loss": 5.8976,
       "step": 2230
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019315746084089033,
-      "loss": 3.5046,
+      "learning_rate": 0.0006438582028029679,
+      "loss": 4.8955,
       "step": 2231
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019309563066776585,
-      "loss": 3.6623,
+      "learning_rate": 0.0006436521022258863,
+      "loss": 4.9153,
       "step": 2232
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019303380049464136,
-      "loss": 3.8932,
+      "learning_rate": 0.0006434460016488047,
+      "loss": 5.3267,
       "step": 2233
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019297197032151688,
-      "loss": 3.4198,
+      "learning_rate": 0.0006432399010717231,
+      "loss": 4.9063,
       "step": 2234
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019291014014839242,
-      "loss": 4.2549,
+      "learning_rate": 0.0006430338004946415,
+      "loss": 5.3899,
       "step": 2235
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019284830997526791,
-      "loss": 3.7857,
+      "learning_rate": 0.0006428276999175598,
+      "loss": 4.8636,
       "step": 2236
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019278647980214343,
-      "loss": 3.7994,
+      "learning_rate": 0.0006426215993404782,
+      "loss": 4.9403,
       "step": 2237
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.00019272464962901895,
-      "loss": 4.2012,
+      "learning_rate": 0.0006424154987633966,
+      "loss": 5.3336,
       "step": 2238
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019266281945589446,
-      "loss": 3.9931,
+      "learning_rate": 0.000642209398186315,
+      "loss": 5.1489,
       "step": 2239
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019260098928276998,
-      "loss": 4.6375,
+      "learning_rate": 0.0006420032976092334,
+      "loss": 6.1086,
       "step": 2240
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019253915910964547,
-      "loss": 4.08,
+      "learning_rate": 0.0006417971970321517,
+      "loss": 5.4432,
       "step": 2241
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019247732893652101,
-      "loss": 4.5831,
+      "learning_rate": 0.0006415910964550701,
+      "loss": 6.203,
       "step": 2242
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019241549876339653,
-      "loss": 3.8669,
+      "learning_rate": 0.0006413849958779885,
+      "loss": 4.9286,
       "step": 2243
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019235366859027205,
-      "loss": 3.9095,
+      "learning_rate": 0.0006411788953009069,
+      "loss": 4.9594,
       "step": 2244
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019229183841714754,
-      "loss": 3.884,
+      "learning_rate": 0.0006409727947238252,
+      "loss": 5.2662,
       "step": 2245
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019223000824402305,
-      "loss": 3.674,
+      "learning_rate": 0.0006407666941467436,
+      "loss": 4.9979,
       "step": 2246
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019216817807089857,
-      "loss": 3.7536,
+      "learning_rate": 0.000640560593569662,
+      "loss": 4.9016,
       "step": 2247
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0001921063478977741,
-      "loss": 4.3044,
+      "learning_rate": 0.0006403544929925804,
+      "loss": 4.9496,
       "step": 2248
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019204451772464963,
-      "loss": 3.8856,
+      "learning_rate": 0.0006401483924154989,
+      "loss": 4.8467,
       "step": 2249
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019198268755152512,
-      "loss": 3.7221,
+      "learning_rate": 0.0006399422918384172,
+      "loss": 4.7107,
       "step": 2250
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019192085737840064,
-      "loss": 4.1697,
+      "learning_rate": 0.0006397361912613356,
+      "loss": 5.3177,
       "step": 2251
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019185902720527615,
-      "loss": 3.9187,
+      "learning_rate": 0.000639530090684254,
+      "loss": 5.1966,
       "step": 2252
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0001917971970321517,
-      "loss": 4.1375,
+      "learning_rate": 0.0006393239901071724,
+      "loss": 5.0696,
       "step": 2253
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019173536685902719,
-      "loss": 3.9276,
+      "learning_rate": 0.0006391178895300907,
+      "loss": 5.2792,
       "step": 2254
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0001916735366859027,
-      "loss": 4.1381,
+      "learning_rate": 0.0006389117889530091,
+      "loss": 5.3309,
       "step": 2255
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.00019161170651277822,
-      "loss": 4.1914,
+      "learning_rate": 0.0006387056883759275,
+      "loss": 5.1494,
       "step": 2256
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019154987633965374,
-      "loss": 3.7526,
+      "learning_rate": 0.0006384995877988459,
+      "loss": 4.8451,
       "step": 2257
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019148804616652925,
-      "loss": 3.8629,
+      "learning_rate": 0.0006382934872217643,
+      "loss": 4.8265,
       "step": 2258
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019142621599340474,
-      "loss": 3.8445,
+      "learning_rate": 0.0006380873866446826,
+      "loss": 5.15,
       "step": 2259
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019136438582028029,
-      "loss": 4.2138,
+      "learning_rate": 0.000637881286067601,
+      "loss": 5.5345,
       "step": 2260
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0001913025556471558,
-      "loss": 3.845,
+      "learning_rate": 0.0006376751854905194,
+      "loss": 5.0338,
       "step": 2261
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019124072547403132,
-      "loss": 3.6153,
+      "learning_rate": 0.0006374690849134378,
+      "loss": 4.7857,
       "step": 2262
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0001911788953009068,
-      "loss": 3.8525,
+      "learning_rate": 0.0006372629843363561,
+      "loss": 5.3663,
       "step": 2263
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019111706512778232,
-      "loss": 3.8609,
+      "learning_rate": 0.0006370568837592745,
+      "loss": 5.3876,
       "step": 2264
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019105523495465784,
-      "loss": 4.7375,
+      "learning_rate": 0.0006368507831821929,
+      "loss": 5.333,
       "step": 2265
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019099340478153338,
-      "loss": 3.891,
+      "learning_rate": 0.0006366446826051113,
+      "loss": 5.4174,
       "step": 2266
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0001909315746084089,
-      "loss": 4.6697,
+      "learning_rate": 0.0006364385820280297,
+      "loss": 6.1951,
       "step": 2267
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0001908697444352844,
-      "loss": 3.6703,
+      "learning_rate": 0.000636232481450948,
+      "loss": 4.9606,
       "step": 2268
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0001908079142621599,
-      "loss": 3.8557,
+      "learning_rate": 0.0006360263808738664,
+      "loss": 5.1086,
       "step": 2269
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019074608408903542,
-      "loss": 3.6243,
+      "learning_rate": 0.0006358202802967849,
+      "loss": 4.8038,
       "step": 2270
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019068425391591097,
-      "loss": 3.8064,
+      "learning_rate": 0.0006356141797197033,
+      "loss": 5.4244,
       "step": 2271
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019062242374278648,
-      "loss": 3.92,
+      "learning_rate": 0.0006354080791426217,
+      "loss": 4.8888,
       "step": 2272
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00019056059356966197,
-      "loss": 3.9536,
+      "learning_rate": 0.00063520197856554,
+      "loss": 4.9327,
       "step": 2273
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0001904987633965375,
-      "loss": 3.4424,
+      "learning_rate": 0.0006349958779884584,
+      "loss": 4.4509,
       "step": 2274
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.000190436933223413,
-      "loss": 3.8912,
+      "learning_rate": 0.0006347897774113768,
+      "loss": 5.3272,
       "step": 2275
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00019037510305028852,
-      "loss": 3.7055,
+      "learning_rate": 0.0006345836768342952,
+      "loss": 4.8705,
       "step": 2276
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.000190313272877164,
-      "loss": 3.4063,
+      "learning_rate": 0.0006343775762572135,
+      "loss": 4.7159,
       "step": 2277
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00019025144270403956,
-      "loss": 3.3095,
+      "learning_rate": 0.0006341714756801319,
+      "loss": 4.6203,
       "step": 2278
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00019018961253091507,
-      "loss": 3.693,
+      "learning_rate": 0.0006339653751030503,
+      "loss": 4.8268,
       "step": 2279
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0001901277823577906,
-      "loss": 4.0112,
+      "learning_rate": 0.0006337592745259687,
+      "loss": 5.08,
       "step": 2280
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0001900659521846661,
-      "loss": 3.5939,
+      "learning_rate": 0.0006335531739488871,
+      "loss": 4.9203,
       "step": 2281
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0001900041220115416,
-      "loss": 3.5213,
+      "learning_rate": 0.0006333470733718054,
+      "loss": 4.8121,
       "step": 2282
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018994229183841714,
-      "loss": 3.6804,
+      "learning_rate": 0.0006331409727947238,
+      "loss": 5.0517,
       "step": 2283
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018988046166529266,
-      "loss": 3.8181,
+      "learning_rate": 0.0006329348722176422,
+      "loss": 5.0679,
       "step": 2284
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018981863149216817,
-      "loss": 4.5093,
+      "learning_rate": 0.0006327287716405606,
+      "loss": 5.8816,
       "step": 2285
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018975680131904366,
-      "loss": 4.1455,
+      "learning_rate": 0.0006325226710634789,
+      "loss": 5.5661,
       "step": 2286
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018969497114591918,
-      "loss": 4.0395,
+      "learning_rate": 0.0006323165704863973,
+      "loss": 5.0868,
       "step": 2287
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0001896331409727947,
-      "loss": 4.2807,
+      "learning_rate": 0.0006321104699093157,
+      "loss": 5.1378,
       "step": 2288
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018957131079967024,
-      "loss": 4.1334,
+      "learning_rate": 0.0006319043693322341,
+      "loss": 5.0546,
       "step": 2289
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018950948062654576,
-      "loss": 3.7305,
+      "learning_rate": 0.0006316982687551526,
+      "loss": 4.9491,
       "step": 2290
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018944765045342124,
-      "loss": 4.2967,
+      "learning_rate": 0.0006314921681780709,
+      "loss": 5.3176,
       "step": 2291
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.00018938582028029676,
-      "loss": 4.0125,
+      "learning_rate": 0.0006312860676009893,
+      "loss": 5.2094,
       "step": 2292
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018932399010717228,
-      "loss": 3.9504,
+      "learning_rate": 0.0006310799670239077,
+      "loss": 4.9723,
       "step": 2293
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0001892621599340478,
-      "loss": 4.0064,
+      "learning_rate": 0.0006308738664468261,
+      "loss": 4.9423,
       "step": 2294
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018920032976092334,
-      "loss": 4.0922,
+      "learning_rate": 0.0006306677658697445,
+      "loss": 5.0661,
       "step": 2295
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018913849958779883,
-      "loss": 4.0143,
+      "learning_rate": 0.0006304616652926628,
+      "loss": 4.937,
       "step": 2296
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018907666941467434,
-      "loss": 3.7485,
+      "learning_rate": 0.0006302555647155812,
+      "loss": 4.8061,
       "step": 2297
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018901483924154986,
-      "loss": 4.0403,
+      "learning_rate": 0.0006300494641384996,
+      "loss": 4.9952,
       "step": 2298
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018895300906842538,
-      "loss": 4.8745,
+      "learning_rate": 0.000629843363561418,
+      "loss": 5.7981,
       "step": 2299
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018889117889530087,
-      "loss": 3.7921,
+      "learning_rate": 0.0006296372629843363,
+      "loss": 4.8712,
       "step": 2300
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0001888293487221764,
-      "loss": 3.6781,
+      "learning_rate": 0.0006294311624072547,
+      "loss": 4.8221,
       "step": 2301
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018876751854905193,
-      "loss": 3.7197,
+      "learning_rate": 0.0006292250618301731,
+      "loss": 4.6996,
       "step": 2302
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018870568837592744,
-      "loss": 3.937,
+      "learning_rate": 0.0006290189612530915,
+      "loss": 5.3258,
       "step": 2303
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018864385820280296,
-      "loss": 3.7296,
+      "learning_rate": 0.0006288128606760099,
+      "loss": 4.818,
       "step": 2304
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018858202802967845,
-      "loss": 3.5919,
+      "learning_rate": 0.0006286067600989282,
+      "loss": 4.9076,
       "step": 2305
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018852019785655397,
-      "loss": 4.3512,
+      "learning_rate": 0.0006284006595218466,
+      "loss": 5.7065,
       "step": 2306
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0001884583676834295,
-      "loss": 4.1229,
+      "learning_rate": 0.000628194558944765,
+      "loss": 5.3776,
       "step": 2307
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018839653751030503,
-      "loss": 4.0557,
+      "learning_rate": 0.0006279884583676834,
+      "loss": 4.9641,
       "step": 2308
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018833470733718052,
-      "loss": 3.7899,
+      "learning_rate": 0.0006277823577906017,
+      "loss": 4.869,
       "step": 2309
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.00018827287716405603,
-      "loss": 3.7088,
+      "learning_rate": 0.0006275762572135201,
+      "loss": 4.9031,
       "step": 2310
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018821104699093155,
-      "loss": 4.0229,
+      "learning_rate": 0.0006273701566364386,
+      "loss": 5.7918,
       "step": 2311
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0001881492168178071,
-      "loss": 3.8619,
+      "learning_rate": 0.000627164056059357,
+      "loss": 4.7717,
       "step": 2312
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0001880873866446826,
-      "loss": 3.9004,
+      "learning_rate": 0.0006269579554822754,
+      "loss": 5.0949,
       "step": 2313
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0001880255564715581,
-      "loss": 4.0461,
+      "learning_rate": 0.0006267518549051937,
+      "loss": 5.2501,
       "step": 2314
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018796372629843362,
-      "loss": 3.7166,
+      "learning_rate": 0.0006265457543281121,
+      "loss": 5.0696,
       "step": 2315
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018790189612530913,
-      "loss": 4.0321,
+      "learning_rate": 0.0006263396537510305,
+      "loss": 5.3677,
       "step": 2316
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018784006595218465,
-      "loss": 3.6642,
+      "learning_rate": 0.0006261335531739489,
+      "loss": 5.3111,
       "step": 2317
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018777823577906014,
-      "loss": 4.1751,
+      "learning_rate": 0.0006259274525968672,
+      "loss": 5.3194,
       "step": 2318
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018771640560593568,
-      "loss": 3.8996,
+      "learning_rate": 0.0006257213520197856,
+      "loss": 5.0015,
       "step": 2319
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0001876545754328112,
-      "loss": 4.0512,
+      "learning_rate": 0.000625515251442704,
+      "loss": 5.27,
       "step": 2320
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018759274525968672,
-      "loss": 3.8226,
+      "learning_rate": 0.0006253091508656224,
+      "loss": 4.9523,
       "step": 2321
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018753091508656223,
-      "loss": 3.514,
+      "learning_rate": 0.0006251030502885409,
+      "loss": 4.7454,
       "step": 2322
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018746908491343772,
-      "loss": 3.757,
+      "learning_rate": 0.0006248969497114591,
+      "loss": 5.177,
       "step": 2323
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018740725474031324,
-      "loss": 4.3297,
+      "learning_rate": 0.0006246908491343776,
+      "loss": 5.2175,
       "step": 2324
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018734542456718878,
-      "loss": 3.7252,
+      "learning_rate": 0.000624484748557296,
+      "loss": 5.1673,
       "step": 2325
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0001872835943940643,
-      "loss": 3.3993,
+      "learning_rate": 0.0006242786479802144,
+      "loss": 4.8122,
       "step": 2326
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.00018722176422093981,
-      "loss": 3.3439,
+      "learning_rate": 0.0006240725474031328,
+      "loss": 4.4473,
       "step": 2327
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0001871599340478153,
-      "loss": 3.3136,
+      "learning_rate": 0.0006238664468260511,
+      "loss": 4.5248,
       "step": 2328
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018709810387469082,
-      "loss": 3.9435,
+      "learning_rate": 0.0006236603462489695,
+      "loss": 5.195,
       "step": 2329
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018703627370156636,
-      "loss": 3.4282,
+      "learning_rate": 0.000623454245671888,
+      "loss": 4.9204,
       "step": 2330
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018697444352844188,
-      "loss": 3.1239,
+      "learning_rate": 0.0006232481450948064,
+      "loss": 4.4068,
       "step": 2331
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018691261335531737,
-      "loss": 3.1063,
+      "learning_rate": 0.0006230420445177247,
+      "loss": 4.611,
       "step": 2332
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0001868507831821929,
-      "loss": 3.5468,
+      "learning_rate": 0.0006228359439406431,
+      "loss": 5.0655,
       "step": 2333
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0001867889530090684,
-      "loss": 4.673,
+      "learning_rate": 0.0006226298433635615,
+      "loss": 6.2401,
       "step": 2334
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018672712283594392,
-      "loss": 4.3262,
+      "learning_rate": 0.0006224237427864799,
+      "loss": 5.3271,
       "step": 2335
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018666529266281946,
-      "loss": 3.7734,
+      "learning_rate": 0.0006222176422093983,
+      "loss": 4.8161,
       "step": 2336
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018660346248969495,
-      "loss": 3.8483,
+      "learning_rate": 0.0006220115416323166,
+      "loss": 5.0242,
       "step": 2337
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018654163231657047,
-      "loss": 3.988,
+      "learning_rate": 0.000621805441055235,
+      "loss": 5.0262,
       "step": 2338
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.000186479802143446,
-      "loss": 3.8795,
+      "learning_rate": 0.0006215993404781534,
+      "loss": 4.9182,
       "step": 2339
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0001864179719703215,
-      "loss": 3.9498,
+      "learning_rate": 0.0006213932399010718,
+      "loss": 4.965,
       "step": 2340
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.000186356141797197,
-      "loss": 4.4934,
+      "learning_rate": 0.0006211871393239901,
+      "loss": 5.4312,
       "step": 2341
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0001862943116240725,
-      "loss": 3.9883,
+      "learning_rate": 0.0006209810387469085,
+      "loss": 5.0733,
       "step": 2342
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018623248145094805,
-      "loss": 3.9173,
+      "learning_rate": 0.0006207749381698269,
+      "loss": 5.4481,
       "step": 2343
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018617065127782357,
-      "loss": 3.6668,
+      "learning_rate": 0.0006205688375927453,
+      "loss": 4.9876,
       "step": 2344
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.00018610882110469909,
-      "loss": 3.8797,
+      "learning_rate": 0.0006203627370156637,
+      "loss": 4.9378,
       "step": 2345
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018604699093157458,
-      "loss": 4.0184,
+      "learning_rate": 0.000620156636438582,
+      "loss": 4.8935,
       "step": 2346
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0001859851607584501,
-      "loss": 3.9719,
+      "learning_rate": 0.0006199505358615004,
+      "loss": 5.1521,
       "step": 2347
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018592333058532564,
-      "loss": 3.6984,
+      "learning_rate": 0.0006197444352844188,
+      "loss": 4.9547,
       "step": 2348
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018586150041220115,
-      "loss": 3.7491,
+      "learning_rate": 0.0006195383347073372,
+      "loss": 5.0906,
       "step": 2349
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018579967023907664,
-      "loss": 4.319,
+      "learning_rate": 0.0006193322341302555,
+      "loss": 5.3036,
       "step": 2350
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018573784006595216,
-      "loss": 3.6742,
+      "learning_rate": 0.000619126133553174,
+      "loss": 4.9321,
       "step": 2351
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018567600989282768,
-      "loss": 3.4522,
+      "learning_rate": 0.0006189200329760924,
+      "loss": 4.6555,
       "step": 2352
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0001856141797197032,
-      "loss": 4.0596,
+      "learning_rate": 0.0006187139323990108,
+      "loss": 4.9486,
       "step": 2353
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018555234954657874,
-      "loss": 3.4933,
+      "learning_rate": 0.0006185078318219292,
+      "loss": 4.6891,
       "step": 2354
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018549051937345423,
-      "loss": 3.6909,
+      "learning_rate": 0.0006183017312448475,
+      "loss": 4.8177,
       "step": 2355
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018542868920032974,
-      "loss": 3.636,
+      "learning_rate": 0.0006180956306677659,
+      "loss": 4.6584,
       "step": 2356
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018536685902720526,
-      "loss": 4.2602,
+      "learning_rate": 0.0006178895300906843,
+      "loss": 5.3163,
       "step": 2357
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018530502885408077,
-      "loss": 3.7032,
+      "learning_rate": 0.0006176834295136027,
+      "loss": 4.9132,
       "step": 2358
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018524319868095632,
-      "loss": 3.7596,
+      "learning_rate": 0.0006174773289365211,
+      "loss": 5.0863,
       "step": 2359
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0001851813685078318,
-      "loss": 4.0096,
+      "learning_rate": 0.0006172712283594394,
+      "loss": 5.227,
       "step": 2360
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018511953833470732,
-      "loss": 3.9833,
+      "learning_rate": 0.0006170651277823578,
+      "loss": 5.248,
       "step": 2361
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018505770816158284,
-      "loss": 3.7023,
+      "learning_rate": 0.0006168590272052762,
+      "loss": 5.0642,
       "step": 2362
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.00018499587798845836,
-      "loss": 3.8812,
+      "learning_rate": 0.0006166529266281946,
+      "loss": 5.5344,
       "step": 2363
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018493404781533385,
-      "loss": 4.1775,
+      "learning_rate": 0.0006164468260511129,
+      "loss": 5.3228,
       "step": 2364
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018487221764220936,
-      "loss": 3.5757,
+      "learning_rate": 0.0006162407254740313,
+      "loss": 5.1179,
       "step": 2365
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0001848103874690849,
-      "loss": 3.6291,
+      "learning_rate": 0.0006160346248969497,
+      "loss": 4.8827,
       "step": 2366
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018474855729596042,
-      "loss": 3.8897,
+      "learning_rate": 0.0006158285243198681,
+      "loss": 5.3349,
       "step": 2367
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018468672712283594,
-      "loss": 3.7074,
+      "learning_rate": 0.0006156224237427865,
+      "loss": 5.5565,
       "step": 2368
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018462489694971143,
-      "loss": 3.8045,
+      "learning_rate": 0.0006154163231657048,
+      "loss": 5.3051,
       "step": 2369
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018456306677658695,
-      "loss": 3.718,
+      "learning_rate": 0.0006152102225886232,
+      "loss": 5.0246,
       "step": 2370
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018450123660346246,
-      "loss": 3.7278,
+      "learning_rate": 0.0006150041220115417,
+      "loss": 4.8395,
       "step": 2371
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.000184439406430338,
-      "loss": 4.2854,
+      "learning_rate": 0.0006147980214344601,
+      "loss": 5.2386,
       "step": 2372
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0001843775762572135,
-      "loss": 3.6508,
+      "learning_rate": 0.0006145919208573784,
+      "loss": 4.8263,
       "step": 2373
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.000184315746084089,
-      "loss": 3.6375,
+      "learning_rate": 0.0006143858202802968,
+      "loss": 5.0549,
       "step": 2374
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018425391591096453,
-      "loss": 3.8015,
+      "learning_rate": 0.0006141797197032152,
+      "loss": 5.4882,
       "step": 2375
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018419208573784005,
-      "loss": 3.4988,
+      "learning_rate": 0.0006139736191261336,
+      "loss": 5.1165,
       "step": 2376
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0001841302555647156,
-      "loss": 3.3281,
+      "learning_rate": 0.000613767518549052,
+      "loss": 4.8328,
       "step": 2377
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018406842539159108,
-      "loss": 3.4745,
+      "learning_rate": 0.0006135614179719703,
+      "loss": 4.9474,
       "step": 2378
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0001840065952184666,
-      "loss": 3.1499,
+      "learning_rate": 0.0006133553173948887,
+      "loss": 4.8292,
       "step": 2379
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0001839447650453421,
-      "loss": 3.5055,
+      "learning_rate": 0.0006131492168178071,
+      "loss": 4.9562,
       "step": 2380
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.00018388293487221763,
-      "loss": 2.9132,
+      "learning_rate": 0.0006129431162407255,
+      "loss": 4.9211,
       "step": 2381
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018382110469909315,
-      "loss": 3.1232,
+      "learning_rate": 0.0006127370156636439,
+      "loss": 4.9979,
       "step": 2382
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018375927452596864,
-      "loss": 2.9088,
+      "learning_rate": 0.0006125309150865622,
+      "loss": 4.5683,
       "step": 2383
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018369744435284418,
-      "loss": 3.1686,
+      "learning_rate": 0.0006123248145094806,
+      "loss": 4.862,
       "step": 2384
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0001836356141797197,
-      "loss": 4.2066,
+      "learning_rate": 0.000612118713932399,
+      "loss": 5.1538,
       "step": 2385
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0001835737840065952,
-      "loss": 3.7934,
+      "learning_rate": 0.0006119126133553174,
+      "loss": 5.3313,
       "step": 2386
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0001835119538334707,
-      "loss": 4.1148,
+      "learning_rate": 0.0006117065127782357,
+      "loss": 5.1489,
       "step": 2387
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018345012366034622,
-      "loss": 3.6934,
+      "learning_rate": 0.0006115004122011541,
+      "loss": 4.7654,
       "step": 2388
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018338829348722176,
-      "loss": 3.7935,
+      "learning_rate": 0.0006112943116240725,
+      "loss": 4.9831,
       "step": 2389
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018332646331409728,
-      "loss": 4.0387,
+      "learning_rate": 0.000611088211046991,
+      "loss": 5.0932,
       "step": 2390
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0001832646331409728,
-      "loss": 3.8157,
+      "learning_rate": 0.0006108821104699094,
+      "loss": 5.1229,
       "step": 2391
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018320280296784828,
-      "loss": 3.8716,
+      "learning_rate": 0.0006106760098928277,
+      "loss": 5.5162,
       "step": 2392
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0001831409727947238,
-      "loss": 4.0173,
+      "learning_rate": 0.0006104699093157461,
+      "loss": 5.3291,
       "step": 2393
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018307914262159932,
-      "loss": 3.8718,
+      "learning_rate": 0.0006102638087386645,
+      "loss": 5.0907,
       "step": 2394
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018301731244847486,
-      "loss": 3.3405,
+      "learning_rate": 0.0006100577081615829,
+      "loss": 4.4309,
       "step": 2395
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018295548227535035,
-      "loss": 3.466,
+      "learning_rate": 0.0006098516075845012,
+      "loss": 4.7308,
       "step": 2396
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018289365210222587,
-      "loss": 3.5958,
+      "learning_rate": 0.0006096455070074196,
+      "loss": 4.9618,
       "step": 2397
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.00018283182192910138,
-      "loss": 3.5852,
+      "learning_rate": 0.000609439406430338,
+      "loss": 4.7887,
       "step": 2398
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0001827699917559769,
-      "loss": 4.6443,
+      "learning_rate": 0.0006092333058532564,
+      "loss": 5.6544,
       "step": 2399
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018270816158285242,
-      "loss": 4.1172,
+      "learning_rate": 0.0006090272052761748,
+      "loss": 5.1565,
       "step": 2400
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0001826463314097279,
-      "loss": 4.2823,
+      "learning_rate": 0.0006088211046990931,
+      "loss": 5.6548,
       "step": 2401
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018258450123660345,
-      "loss": 3.5486,
+      "learning_rate": 0.0006086150041220115,
+      "loss": 4.8998,
       "step": 2402
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018252267106347897,
-      "loss": 3.8165,
+      "learning_rate": 0.0006084089035449299,
+      "loss": 5.5053,
       "step": 2403
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018246084089035448,
-      "loss": 4.0201,
+      "learning_rate": 0.0006082028029678483,
+      "loss": 5.9069,
       "step": 2404
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018239901071722997,
-      "loss": 3.7292,
+      "learning_rate": 0.0006079967023907666,
+      "loss": 5.4749,
       "step": 2405
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0001823371805441055,
-      "loss": 3.4639,
+      "learning_rate": 0.000607790601813685,
+      "loss": 4.9121,
       "step": 2406
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018227535037098103,
-      "loss": 3.7258,
+      "learning_rate": 0.0006075845012366034,
+      "loss": 4.8255,
       "step": 2407
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018221352019785655,
-      "loss": 3.5853,
+      "learning_rate": 0.0006073784006595218,
+      "loss": 4.996,
       "step": 2408
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018215169002473207,
-      "loss": 3.7823,
+      "learning_rate": 0.0006071723000824402,
+      "loss": 5.1797,
       "step": 2409
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018208985985160756,
-      "loss": 3.5241,
+      "learning_rate": 0.0006069661995053585,
+      "loss": 4.9398,
       "step": 2410
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018202802967848307,
-      "loss": 3.5509,
+      "learning_rate": 0.0006067600989282769,
+      "loss": 4.8418,
       "step": 2411
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0001819661995053586,
-      "loss": 3.4202,
+      "learning_rate": 0.0006065539983511954,
+      "loss": 4.8482,
       "step": 2412
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018190436933223413,
-      "loss": 3.9924,
+      "learning_rate": 0.0006063478977741138,
+      "loss": 5.5902,
       "step": 2413
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018184253915910965,
-      "loss": 3.2852,
+      "learning_rate": 0.0006061417971970322,
+      "loss": 4.8184,
       "step": 2414
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018178070898598514,
-      "loss": 3.5393,
+      "learning_rate": 0.0006059356966199505,
+      "loss": 4.8406,
       "step": 2415
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018171887881286066,
-      "loss": 4.4754,
+      "learning_rate": 0.000605729596042869,
+      "loss": 5.9371,
       "step": 2416
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.00018165704863973617,
-      "loss": 3.5795,
+      "learning_rate": 0.0006055234954657874,
+      "loss": 4.9705,
       "step": 2417
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018159521846661172,
-      "loss": 3.5219,
+      "learning_rate": 0.0006053173948887058,
+      "loss": 5.1964,
       "step": 2418
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0001815333882934872,
-      "loss": 3.9107,
+      "learning_rate": 0.0006051112943116241,
+      "loss": 5.2744,
       "step": 2419
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018147155812036272,
-      "loss": 3.3815,
+      "learning_rate": 0.0006049051937345425,
+      "loss": 4.9904,
       "step": 2420
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018140972794723824,
-      "loss": 3.5611,
+      "learning_rate": 0.0006046990931574609,
+      "loss": 5.239,
       "step": 2421
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018134789777411376,
-      "loss": 3.733,
+      "learning_rate": 0.0006044929925803793,
+      "loss": 5.1387,
       "step": 2422
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018128606760098927,
-      "loss": 3.4543,
+      "learning_rate": 0.0006042868920032977,
+      "loss": 4.7403,
       "step": 2423
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018122423742786476,
-      "loss": 3.4365,
+      "learning_rate": 0.000604080791426216,
+      "loss": 5.0144,
       "step": 2424
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0001811624072547403,
-      "loss": 3.4727,
+      "learning_rate": 0.0006038746908491344,
+      "loss": 4.9431,
       "step": 2425
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018110057708161582,
-      "loss": 3.408,
+      "learning_rate": 0.0006036685902720528,
+      "loss": 5.0025,
       "step": 2426
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018103874690849134,
-      "loss": 3.3554,
+      "learning_rate": 0.0006034624896949712,
+      "loss": 4.8716,
       "step": 2427
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018097691673536683,
-      "loss": 3.4479,
+      "learning_rate": 0.0006032563891178895,
+      "loss": 4.8287,
       "step": 2428
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018091508656224234,
-      "loss": 3.0285,
+      "learning_rate": 0.0006030502885408079,
+      "loss": 4.7636,
       "step": 2429
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018085325638911786,
-      "loss": 3.0091,
+      "learning_rate": 0.0006028441879637263,
+      "loss": 4.6974,
       "step": 2430
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0001807914262159934,
-      "loss": 3.3258,
+      "learning_rate": 0.0006026380873866448,
+      "loss": 4.9268,
       "step": 2431
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018072959604286892,
-      "loss": 2.5815,
+      "learning_rate": 0.0006024319868095632,
+      "loss": 4.4698,
       "step": 2432
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0001806677658697444,
-      "loss": 4.3003,
+      "learning_rate": 0.0006022258862324815,
+      "loss": 7.1452,
       "step": 2433
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018060593569661993,
-      "loss": 3.3961,
+      "learning_rate": 0.0006020197856553999,
+      "loss": 4.905,
       "step": 2434
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.00018054410552349544,
-      "loss": 3.8488,
+      "learning_rate": 0.0006018136850783183,
+      "loss": 5.0231,
       "step": 2435
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.000180482275350371,
-      "loss": 4.8399,
+      "learning_rate": 0.0006016075845012367,
+      "loss": 5.8013,
       "step": 2436
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00018042044517724648,
-      "loss": 4.2166,
+      "learning_rate": 0.000601401483924155,
+      "loss": 5.5135,
       "step": 2437
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.000180358615004122,
-      "loss": 4.373,
+      "learning_rate": 0.0006011953833470734,
+      "loss": 5.3022,
       "step": 2438
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0001802967848309975,
-      "loss": 4.2064,
+      "learning_rate": 0.0006009892827699918,
+      "loss": 5.0997,
       "step": 2439
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00018023495465787303,
-      "loss": 3.6286,
+      "learning_rate": 0.0006007831821929102,
+      "loss": 4.9978,
       "step": 2440
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00018017312448474854,
-      "loss": 3.8155,
+      "learning_rate": 0.0006005770816158286,
+      "loss": 5.2948,
       "step": 2441
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00018011129431162403,
-      "loss": 3.9064,
+      "learning_rate": 0.0006003709810387469,
+      "loss": 5.1099,
       "step": 2442
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00018004946413849958,
-      "loss": 3.6498,
+      "learning_rate": 0.0006001648804616653,
+      "loss": 4.9653,
       "step": 2443
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0001799876339653751,
-      "loss": 3.6088,
+      "learning_rate": 0.0005999587798845837,
+      "loss": 4.9423,
       "step": 2444
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0001799258037922506,
-      "loss": 3.6032,
+      "learning_rate": 0.0005997526793075021,
+      "loss": 5.2526,
       "step": 2445
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00017986397361912613,
-      "loss": 3.9861,
+      "learning_rate": 0.0005995465787304205,
+      "loss": 5.3132,
       "step": 2446
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00017980214344600162,
-      "loss": 3.9662,
+      "learning_rate": 0.0005993404781533388,
+      "loss": 5.2025,
       "step": 2447
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00017974031327287713,
-      "loss": 3.4003,
+      "learning_rate": 0.0005991343775762572,
+      "loss": 5.227,
       "step": 2448
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00017967848309975268,
-      "loss": 3.8318,
+      "learning_rate": 0.0005989282769991756,
+      "loss": 5.0464,
       "step": 2449
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0001796166529266282,
-      "loss": 3.4758,
+      "learning_rate": 0.000598722176422094,
+      "loss": 5.1404,
       "step": 2450
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.00017955482275350368,
-      "loss": 3.5433,
+      "learning_rate": 0.0005985160758450123,
+      "loss": 4.9662,
       "step": 2451
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0001794929925803792,
-      "loss": 3.8339,
+      "learning_rate": 0.0005983099752679308,
+      "loss": 5.353,
       "step": 2452
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017943116240725472,
-      "loss": 4.3693,
+      "learning_rate": 0.0005981038746908492,
+      "loss": 5.2103,
       "step": 2453
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017936933223413026,
-      "loss": 3.2955,
+      "learning_rate": 0.0005978977741137676,
+      "loss": 4.8355,
       "step": 2454
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017930750206100578,
-      "loss": 3.5874,
+      "learning_rate": 0.000597691673536686,
+      "loss": 4.9085,
       "step": 2455
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017924567188788126,
-      "loss": 3.4901,
+      "learning_rate": 0.0005974855729596043,
+      "loss": 4.9233,
       "step": 2456
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017918384171475678,
-      "loss": 3.9017,
+      "learning_rate": 0.0005972794723825227,
+      "loss": 5.3614,
       "step": 2457
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0001791220115416323,
-      "loss": 3.4531,
+      "learning_rate": 0.0005970733718054411,
+      "loss": 4.9241,
       "step": 2458
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017906018136850781,
-      "loss": 3.6075,
+      "learning_rate": 0.0005968672712283595,
+      "loss": 5.1216,
       "step": 2459
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0001789983511953833,
-      "loss": 3.1808,
+      "learning_rate": 0.0005966611706512778,
+      "loss": 4.6808,
       "step": 2460
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017893652102225885,
-      "loss": 3.3851,
+      "learning_rate": 0.0005964550700741962,
+      "loss": 4.9575,
       "step": 2461
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017887469084913436,
-      "loss": 3.4458,
+      "learning_rate": 0.0005962489694971146,
+      "loss": 4.9834,
       "step": 2462
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017881286067600988,
-      "loss": 3.4911,
+      "learning_rate": 0.000596042868920033,
+      "loss": 5.1103,
       "step": 2463
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0001787510305028854,
-      "loss": 3.3754,
+      "learning_rate": 0.0005958367683429514,
+      "loss": 4.9371,
       "step": 2464
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0001786892003297609,
-      "loss": 3.4734,
+      "learning_rate": 0.0005956306677658697,
+      "loss": 5.0491,
       "step": 2465
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017862737015663643,
-      "loss": 3.0238,
+      "learning_rate": 0.0005954245671887881,
+      "loss": 4.703,
       "step": 2466
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017856553998351195,
-      "loss": 3.2465,
+      "learning_rate": 0.0005952184666117065,
+      "loss": 5.0961,
       "step": 2467
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017850370981038746,
-      "loss": 3.4145,
+      "learning_rate": 0.0005950123660346249,
+      "loss": 5.564,
       "step": 2468
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017844187963726295,
-      "loss": 3.5526,
+      "learning_rate": 0.0005948062654575432,
+      "loss": 5.2638,
       "step": 2469
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.00017838004946413847,
-      "loss": 3.4434,
+      "learning_rate": 0.0005946001648804616,
+      "loss": 5.0757,
       "step": 2470
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.000178318219291014,
-      "loss": 3.0941,
+      "learning_rate": 0.00059439406430338,
+      "loss": 4.9441,
       "step": 2471
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017825638911788953,
-      "loss": 3.1583,
+      "learning_rate": 0.0005941879637262985,
+      "loss": 4.7709,
       "step": 2472
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017819455894476505,
-      "loss": 3.0768,
+      "learning_rate": 0.0005939818631492169,
+      "loss": 4.779,
       "step": 2473
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017813272877164054,
-      "loss": 3.0656,
+      "learning_rate": 0.0005937757625721352,
+      "loss": 4.6202,
       "step": 2474
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017807089859851605,
-      "loss": 3.3133,
+      "learning_rate": 0.0005935696619950536,
+      "loss": 5.1384,
       "step": 2475
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017800906842539157,
-      "loss": 3.2076,
+      "learning_rate": 0.000593363561417972,
+      "loss": 5.6382,
       "step": 2476
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017794723825226709,
-      "loss": 3.4598,
+      "learning_rate": 0.0005931574608408904,
+      "loss": 5.0036,
       "step": 2477
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017788540807914263,
-      "loss": 3.499,
+      "learning_rate": 0.0005929513602638088,
+      "loss": 5.2854,
       "step": 2478
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017782357790601812,
-      "loss": 2.6269,
+      "learning_rate": 0.0005927452596867271,
+      "loss": 4.5547,
       "step": 2479
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017776174773289364,
-      "loss": 3.7386,
+      "learning_rate": 0.0005925391591096455,
+      "loss": 5.3497,
       "step": 2480
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017769991755976915,
-      "loss": 2.435,
+      "learning_rate": 0.0005923330585325639,
+      "loss": 4.6184,
       "step": 2481
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017763808738664467,
-      "loss": 2.596,
+      "learning_rate": 0.0005921269579554823,
+      "loss": 4.6025,
       "step": 2482
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017757625721352016,
-      "loss": 2.6575,
+      "learning_rate": 0.0005919208573784006,
+      "loss": 4.6355,
       "step": 2483
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0001775144270403957,
-      "loss": 2.846,
+      "learning_rate": 0.000591714756801319,
+      "loss": 5.2645,
       "step": 2484
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017745259686727122,
-      "loss": 4.819,
+      "learning_rate": 0.0005915086562242374,
+      "loss": 6.3168,
       "step": 2485
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017739076669414674,
-      "loss": 4.3478,
+      "learning_rate": 0.0005913025556471558,
+      "loss": 4.9091,
       "step": 2486
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017732893652102225,
-      "loss": 3.9171,
+      "learning_rate": 0.0005910964550700742,
+      "loss": 4.9234,
       "step": 2487
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00017726710634789774,
-      "loss": 3.7595,
+      "learning_rate": 0.0005908903544929925,
+      "loss": 5.3417,
       "step": 2488
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017720527617477326,
-      "loss": 3.6527,
+      "learning_rate": 0.0005906842539159109,
+      "loss": 4.9838,
       "step": 2489
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0001771434460016488,
-      "loss": 3.6049,
+      "learning_rate": 0.0005904781533388293,
+      "loss": 5.2376,
       "step": 2490
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017708161582852432,
-      "loss": 3.5076,
+      "learning_rate": 0.0005902720527617477,
+      "loss": 4.967,
       "step": 2491
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0001770197856553998,
-      "loss": 3.8018,
+      "learning_rate": 0.000590065952184666,
+      "loss": 5.5947,
       "step": 2492
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017695795548227532,
-      "loss": 3.4853,
+      "learning_rate": 0.0005898598516075845,
+      "loss": 4.8256,
       "step": 2493
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017689612530915084,
-      "loss": 3.5966,
+      "learning_rate": 0.0005896537510305029,
+      "loss": 4.8316,
       "step": 2494
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017683429513602638,
-      "loss": 3.455,
+      "learning_rate": 0.0005894476504534213,
+      "loss": 4.8631,
       "step": 2495
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0001767724649629019,
-      "loss": 3.7383,
+      "learning_rate": 0.0005892415498763397,
+      "loss": 4.861,
       "step": 2496
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0001767106347897774,
-      "loss": 3.3022,
+      "learning_rate": 0.000589035449299258,
+      "loss": 5.2099,
       "step": 2497
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0001766488046166529,
-      "loss": 3.5611,
+      "learning_rate": 0.0005888293487221764,
+      "loss": 5.1522,
       "step": 2498
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017658697444352842,
-      "loss": 3.6399,
+      "learning_rate": 0.0005886232481450948,
+      "loss": 4.8832,
       "step": 2499
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017652514427040394,
-      "loss": 3.4627,
+      "learning_rate": 0.0005884171475680132,
+      "loss": 4.9413,
       "step": 2500
     },
     {
       "epoch": 1.4,
-      "eval_loss": 3.5971059799194336,
-      "eval_runtime": 1247.7548,
+      "eval_loss": 5.506562232971191,
+      "eval_runtime": 1247.7608,
       "eval_samples_per_second": 2.117,
       "eval_steps_per_second": 0.265,
-      "eval_wer": 1.7001388337961127,
+      "eval_wer": 1.9556326854422847,
       "step": 2500
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017646331409727948,
-      "loss": 3.2204,
+      "learning_rate": 0.0005882110469909316,
+      "loss": 4.7913,
       "step": 2501
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017640148392415497,
-      "loss": 3.6548,
+      "learning_rate": 0.0005880049464138499,
+      "loss": 4.9295,
       "step": 2502
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0001763396537510305,
-      "loss": 3.2662,
+      "learning_rate": 0.0005877988458367683,
+      "loss": 5.0334,
       "step": 2503
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.000176277823577906,
-      "loss": 3.0807,
+      "learning_rate": 0.0005875927452596867,
+      "loss": 4.5795,
       "step": 2504
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.00017621599340478152,
-      "loss": 2.9659,
+      "learning_rate": 0.0005873866446826051,
+      "loss": 4.7373,
       "step": 2505
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.000176154163231657,
-      "loss": 3.1876,
+      "learning_rate": 0.0005871805441055234,
+      "loss": 5.14,
       "step": 2506
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017609233305853253,
-      "loss": 2.9733,
+      "learning_rate": 0.0005869744435284418,
+      "loss": 4.7387,
       "step": 2507
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017603050288540807,
-      "loss": 3.152,
+      "learning_rate": 0.0005867683429513602,
+      "loss": 4.9077,
       "step": 2508
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001759686727122836,
-      "loss": 2.9788,
+      "learning_rate": 0.0005865622423742787,
+      "loss": 4.6913,
       "step": 2509
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001759068425391591,
-      "loss": 3.5379,
+      "learning_rate": 0.0005863561417971971,
+      "loss": 5.4268,
       "step": 2510
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001758450123660346,
-      "loss": 3.2736,
+      "learning_rate": 0.0005861500412201154,
+      "loss": 4.8315,
       "step": 2511
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001757831821929101,
-      "loss": 2.935,
+      "learning_rate": 0.0005859439406430339,
+      "loss": 5.0394,
       "step": 2512
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017572135201978566,
-      "loss": 3.7559,
+      "learning_rate": 0.0005857378400659523,
+      "loss": 4.6983,
       "step": 2513
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017565952184666117,
-      "loss": 3.0925,
+      "learning_rate": 0.0005855317394888707,
+      "loss": 4.8267,
       "step": 2514
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017559769167353666,
-      "loss": 3.7332,
+      "learning_rate": 0.000585325638911789,
+      "loss": 5.5017,
       "step": 2515
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017553586150041218,
-      "loss": 3.0546,
+      "learning_rate": 0.0005851195383347074,
+      "loss": 4.9261,
       "step": 2516
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001754740313272877,
-      "loss": 3.081,
+      "learning_rate": 0.0005849134377576258,
+      "loss": 4.8778,
       "step": 2517
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001754122011541632,
-      "loss": 3.8916,
+      "learning_rate": 0.0005847073371805442,
+      "loss": 5.4911,
       "step": 2518
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017535037098103876,
-      "loss": 3.2369,
+      "learning_rate": 0.0005845012366034626,
+      "loss": 5.2547,
       "step": 2519
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017528854080791425,
-      "loss": 3.3929,
+      "learning_rate": 0.0005842951360263809,
+      "loss": 4.9811,
       "step": 2520
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017522671063478976,
-      "loss": 3.0752,
+      "learning_rate": 0.0005840890354492993,
+      "loss": 4.8932,
       "step": 2521
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017516488046166528,
-      "loss": 2.7807,
+      "learning_rate": 0.0005838829348722177,
+      "loss": 4.6617,
       "step": 2522
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0001751030502885408,
-      "loss": 2.9335,
+      "learning_rate": 0.0005836768342951361,
+      "loss": 4.4748,
       "step": 2523
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.00017504122011541628,
-      "loss": 2.7527,
+      "learning_rate": 0.0005834707337180544,
+      "loss": 5.2021,
       "step": 2524
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0001749793899422918,
-      "loss": 2.6512,
+      "learning_rate": 0.0005832646331409728,
+      "loss": 4.5598,
       "step": 2525
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017491755976916734,
-      "loss": 3.1591,
+      "learning_rate": 0.0005830585325638912,
+      "loss": 5.0887,
       "step": 2526
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017485572959604286,
-      "loss": 3.2566,
+      "learning_rate": 0.0005828524319868096,
+      "loss": 5.33,
       "step": 2527
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017479389942291838,
-      "loss": 2.779,
+      "learning_rate": 0.000582646331409728,
+      "loss": 4.765,
       "step": 2528
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017473206924979387,
-      "loss": 2.5147,
+      "learning_rate": 0.0005824402308326463,
+      "loss": 4.6768,
       "step": 2529
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017467023907666938,
-      "loss": 3.1574,
+      "learning_rate": 0.0005822341302555647,
+      "loss": 5.7151,
       "step": 2530
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017460840890354493,
-      "loss": 2.6865,
+      "learning_rate": 0.0005820280296784831,
+      "loss": 4.93,
       "step": 2531
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017454657873042044,
-      "loss": 3.3352,
+      "learning_rate": 0.0005818219291014016,
+      "loss": 5.4786,
       "step": 2532
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017448474855729596,
-      "loss": 2.5529,
+      "learning_rate": 0.00058161582852432,
+      "loss": 5.3676,
       "step": 2533
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017442291838417145,
-      "loss": 3.3362,
+      "learning_rate": 0.0005814097279472383,
+      "loss": 5.1701,
       "step": 2534
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017436108821104697,
-      "loss": 3.8931,
+      "learning_rate": 0.0005812036273701567,
+      "loss": 4.8913,
       "step": 2535
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017429925803792248,
-      "loss": 4.3727,
+      "learning_rate": 0.0005809975267930751,
+      "loss": 5.4508,
       "step": 2536
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017423742786479803,
-      "loss": 3.964,
+      "learning_rate": 0.0005807914262159935,
+      "loss": 5.6667,
       "step": 2537
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017417559769167352,
-      "loss": 4.399,
+      "learning_rate": 0.0005805853256389118,
+      "loss": 6.0381,
       "step": 2538
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017411376751854903,
-      "loss": 3.3654,
+      "learning_rate": 0.0005803792250618302,
+      "loss": 5.2134,
       "step": 2539
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017405193734542455,
-      "loss": 3.4719,
+      "learning_rate": 0.0005801731244847486,
+      "loss": 5.1166,
       "step": 2540
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00017399010717230007,
-      "loss": 3.4178,
+      "learning_rate": 0.000579967023907667,
+      "loss": 5.0468,
       "step": 2541
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0001739282769991756,
-      "loss": 3.6312,
+      "learning_rate": 0.0005797609233305854,
+      "loss": 4.9933,
       "step": 2542
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0001738664468260511,
-      "loss": 3.3493,
+      "learning_rate": 0.0005795548227535037,
+      "loss": 4.8563,
       "step": 2543
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017380461665292662,
-      "loss": 3.1482,
+      "learning_rate": 0.0005793487221764221,
+      "loss": 4.8789,
       "step": 2544
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017374278647980213,
-      "loss": 3.4189,
+      "learning_rate": 0.0005791426215993405,
+      "loss": 4.8522,
       "step": 2545
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017368095630667765,
-      "loss": 3.2384,
+      "learning_rate": 0.0005789365210222589,
+      "loss": 4.8898,
       "step": 2546
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017361912613355314,
-      "loss": 3.4601,
+      "learning_rate": 0.0005787304204451772,
+      "loss": 5.0933,
       "step": 2547
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017355729596042866,
-      "loss": 3.2905,
+      "learning_rate": 0.0005785243198680956,
+      "loss": 4.9885,
       "step": 2548
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0001734954657873042,
-      "loss": 2.7918,
+      "learning_rate": 0.000578318219291014,
+      "loss": 4.612,
       "step": 2549
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017343363561417972,
-      "loss": 3.0333,
+      "learning_rate": 0.0005781121187139324,
+      "loss": 4.8738,
       "step": 2550
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017337180544105523,
-      "loss": 3.0676,
+      "learning_rate": 0.0005779060181368508,
+      "loss": 4.7944,
       "step": 2551
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017330997526793072,
-      "loss": 3.3548,
+      "learning_rate": 0.0005776999175597691,
+      "loss": 5.0589,
       "step": 2552
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017324814509480624,
-      "loss": 3.2526,
+      "learning_rate": 0.0005774938169826876,
+      "loss": 5.0891,
       "step": 2553
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017318631492168175,
-      "loss": 3.3803,
+      "learning_rate": 0.000577287716405606,
+      "loss": 5.512,
       "step": 2554
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0001731244847485573,
-      "loss": 3.3166,
+      "learning_rate": 0.0005770816158285244,
+      "loss": 5.0868,
       "step": 2555
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0001730626545754328,
-      "loss": 2.5653,
+      "learning_rate": 0.0005768755152514427,
+      "loss": 4.6583,
       "step": 2556
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0001730008244023083,
-      "loss": 3.1777,
+      "learning_rate": 0.0005766694146743611,
+      "loss": 5.1158,
       "step": 2557
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017293899422918382,
-      "loss": 2.8836,
+      "learning_rate": 0.0005764633140972795,
+      "loss": 4.7543,
       "step": 2558
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017287716405605934,
-      "loss": 3.42,
+      "learning_rate": 0.0005762572135201979,
+      "loss": 5.1819,
       "step": 2559
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.00017281533388293488,
-      "loss": 3.0938,
+      "learning_rate": 0.0005760511129431163,
+      "loss": 5.0667,
       "step": 2560
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017275350370981037,
-      "loss": 2.8936,
+      "learning_rate": 0.0005758450123660346,
+      "loss": 4.8594,
       "step": 2561
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001726916735366859,
-      "loss": 2.8118,
+      "learning_rate": 0.000575638911788953,
+      "loss": 4.8421,
       "step": 2562
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001726298433635614,
-      "loss": 3.1582,
+      "learning_rate": 0.0005754328112118714,
+      "loss": 4.8633,
       "step": 2563
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017256801319043692,
-      "loss": 3.0231,
+      "learning_rate": 0.0005752267106347898,
+      "loss": 4.9137,
       "step": 2564
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017250618301731244,
-      "loss": 2.8702,
+      "learning_rate": 0.0005750206100577082,
+      "loss": 4.946,
       "step": 2565
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017244435284418793,
-      "loss": 2.9489,
+      "learning_rate": 0.0005748145094806265,
+      "loss": 4.928,
       "step": 2566
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017238252267106347,
-      "loss": 3.0058,
+      "learning_rate": 0.0005746084089035449,
+      "loss": 5.1521,
       "step": 2567
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.000172320692497939,
-      "loss": 3.006,
+      "learning_rate": 0.0005744023083264633,
+      "loss": 4.8473,
       "step": 2568
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001722588623248145,
-      "loss": 3.2847,
+      "learning_rate": 0.0005741962077493817,
+      "loss": 5.0472,
       "step": 2569
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017219703215169,
-      "loss": 2.7456,
+      "learning_rate": 0.0005739901071723,
+      "loss": 5.2403,
       "step": 2570
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001721352019785655,
-      "loss": 2.7471,
+      "learning_rate": 0.0005737840065952184,
+      "loss": 5.1523,
       "step": 2571
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017207337180544105,
-      "loss": 2.8514,
+      "learning_rate": 0.0005735779060181368,
+      "loss": 5.2161,
       "step": 2572
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017201154163231657,
-      "loss": 3.0193,
+      "learning_rate": 0.0005733718054410553,
+      "loss": 5.8123,
       "step": 2573
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001719497114591921,
-      "loss": 3.2426,
+      "learning_rate": 0.0005731657048639737,
+      "loss": 5.4737,
       "step": 2574
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.00017188788128606758,
-      "loss": 2.3354,
+      "learning_rate": 0.000572959604286892,
+      "loss": 4.6615,
       "step": 2575
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001718260511129431,
-      "loss": 2.8669,
+      "learning_rate": 0.0005727535037098104,
+      "loss": 4.8602,
       "step": 2576
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0001717642209398186,
-      "loss": 3.7589,
+      "learning_rate": 0.0005725474031327288,
+      "loss": 5.575,
       "step": 2577
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017170239076669415,
-      "loss": 2.5931,
+      "learning_rate": 0.0005723413025556472,
+      "loss": 4.7768,
       "step": 2578
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017164056059356964,
-      "loss": 2.1993,
+      "learning_rate": 0.0005721352019785655,
+      "loss": 4.6421,
       "step": 2579
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017157873042044516,
-      "loss": 2.4738,
+      "learning_rate": 0.0005719291014014839,
+      "loss": 4.9369,
       "step": 2580
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017151690024732068,
-      "loss": 3.0736,
+      "learning_rate": 0.0005717230008244023,
+      "loss": 5.3888,
       "step": 2581
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0001714550700741962,
-      "loss": 3.6306,
+      "learning_rate": 0.0005715169002473207,
+      "loss": 6.6137,
       "step": 2582
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0001713932399010717,
-      "loss": 2.8279,
+      "learning_rate": 0.0005713107996702391,
+      "loss": 4.9025,
       "step": 2583
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0001713314097279472,
-      "loss": 2.8936,
+      "learning_rate": 0.0005711046990931574,
+      "loss": 4.7283,
       "step": 2584
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017126957955482274,
-      "loss": 4.0665,
+      "learning_rate": 0.0005708985985160758,
+      "loss": 5.0137,
       "step": 2585
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017120774938169826,
-      "loss": 4.189,
+      "learning_rate": 0.0005706924979389942,
+      "loss": 5.0905,
       "step": 2586
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017114591920857378,
-      "loss": 4.0134,
+      "learning_rate": 0.0005704863973619126,
+      "loss": 5.0072,
       "step": 2587
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017108408903544926,
-      "loss": 3.4283,
+      "learning_rate": 0.0005702802967848309,
+      "loss": 4.9008,
       "step": 2588
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017102225886232478,
-      "loss": 3.2267,
+      "learning_rate": 0.0005700741962077493,
+      "loss": 4.7538,
       "step": 2589
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017096042868920032,
-      "loss": 3.2098,
+      "learning_rate": 0.0005698680956306677,
+      "loss": 4.9587,
       "step": 2590
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017089859851607584,
-      "loss": 3.0284,
+      "learning_rate": 0.0005696619950535861,
+      "loss": 4.807,
       "step": 2591
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017083676834295136,
-      "loss": 3.3748,
+      "learning_rate": 0.0005694558944765045,
+      "loss": 4.9758,
       "step": 2592
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017077493816982685,
-      "loss": 3.2928,
+      "learning_rate": 0.0005692497938994228,
+      "loss": 4.9983,
       "step": 2593
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017071310799670236,
-      "loss": 3.2632,
+      "learning_rate": 0.0005690436933223412,
+      "loss": 4.9617,
       "step": 2594
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.00017065127782357788,
-      "loss": 3.1112,
+      "learning_rate": 0.0005688375927452597,
+      "loss": 4.9015,
       "step": 2595
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00017058944765045342,
-      "loss": 3.0835,
+      "learning_rate": 0.0005686314921681781,
+      "loss": 4.7615,
       "step": 2596
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00017052761747732894,
-      "loss": 2.9482,
+      "learning_rate": 0.0005684253915910965,
+      "loss": 4.9225,
       "step": 2597
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00017046578730420443,
-      "loss": 2.6937,
+      "learning_rate": 0.0005682192910140148,
+      "loss": 5.0747,
       "step": 2598
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00017040395713107995,
-      "loss": 2.6602,
+      "learning_rate": 0.0005680131904369332,
+      "loss": 4.7409,
       "step": 2599
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00017034212695795546,
-      "loss": 3.0561,
+      "learning_rate": 0.0005678070898598516,
+      "loss": 5.337,
       "step": 2600
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.000170280296784831,
-      "loss": 2.8816,
+      "learning_rate": 0.00056760098928277,
+      "loss": 5.0503,
       "step": 2601
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0001702184666117065,
-      "loss": 3.0203,
+      "learning_rate": 0.0005673948887056883,
+      "loss": 4.8812,
       "step": 2602
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.000170156636438582,
-      "loss": 3.2336,
+      "learning_rate": 0.0005671887881286068,
+      "loss": 4.982,
       "step": 2603
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00017009480626545753,
-      "loss": 2.9635,
+      "learning_rate": 0.0005669826875515252,
+      "loss": 4.967,
       "step": 2604
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00017003297609233305,
-      "loss": 2.83,
+      "learning_rate": 0.0005667765869744436,
+      "loss": 5.1395,
       "step": 2605
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00016997114591920856,
-      "loss": 2.5041,
+      "learning_rate": 0.000566570486397362,
+      "loss": 4.6981,
       "step": 2606
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00016990931574608405,
-      "loss": 2.5625,
+      "learning_rate": 0.0005663643858202803,
+      "loss": 4.7929,
       "step": 2607
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0001698474855729596,
-      "loss": 2.7353,
+      "learning_rate": 0.0005661582852431987,
+      "loss": 4.8682,
       "step": 2608
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0001697856553998351,
-      "loss": 2.6612,
+      "learning_rate": 0.0005659521846661171,
+      "loss": 4.5313,
       "step": 2609
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00016972382522671063,
-      "loss": 3.0859,
+      "learning_rate": 0.0005657460840890355,
+      "loss": 5.3326,
       "step": 2610
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00016966199505358612,
-      "loss": 3.2205,
+      "learning_rate": 0.0005655399835119538,
+      "loss": 5.0761,
       "step": 2611
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00016960016488046164,
-      "loss": 2.733,
+      "learning_rate": 0.0005653338829348722,
+      "loss": 4.8847,
       "step": 2612
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00016953833470733715,
-      "loss": 3.1019,
+      "learning_rate": 0.0005651277823577907,
+      "loss": 5.0449,
       "step": 2613
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0001694765045342127,
-      "loss": 2.977,
+      "learning_rate": 0.0005649216817807091,
+      "loss": 5.0013,
       "step": 2614
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0001694146743610882,
-      "loss": 3.089,
+      "learning_rate": 0.0005647155812036275,
+      "loss": 5.3472,
       "step": 2615
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0001693528441879637,
-      "loss": 2.6004,
+      "learning_rate": 0.0005645094806265458,
+      "loss": 4.8797,
       "step": 2616
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016929101401483922,
-      "loss": 2.4707,
+      "learning_rate": 0.0005643033800494642,
+      "loss": 4.7996,
       "step": 2617
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016922918384171474,
-      "loss": 2.5298,
+      "learning_rate": 0.0005640972794723826,
+      "loss": 4.5358,
       "step": 2618
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016916735366859028,
-      "loss": 2.9126,
+      "learning_rate": 0.000563891178895301,
+      "loss": 5.0358,
       "step": 2619
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0001691055234954658,
-      "loss": 2.3508,
+      "learning_rate": 0.0005636850783182194,
+      "loss": 4.6239,
       "step": 2620
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016904369332234128,
-      "loss": 2.6246,
+      "learning_rate": 0.0005634789777411377,
+      "loss": 4.9326,
       "step": 2621
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0001689818631492168,
-      "loss": 2.4502,
+      "learning_rate": 0.0005632728771640561,
+      "loss": 4.9566,
       "step": 2622
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016892003297609232,
-      "loss": 2.4033,
+      "learning_rate": 0.0005630667765869745,
+      "loss": 4.8354,
       "step": 2623
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016885820280296783,
-      "loss": 2.528,
+      "learning_rate": 0.0005628606760098929,
+      "loss": 4.81,
       "step": 2624
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016879637262984332,
-      "loss": 2.6066,
+      "learning_rate": 0.0005626545754328112,
+      "loss": 4.9386,
       "step": 2625
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016873454245671887,
-      "loss": 2.4445,
+      "learning_rate": 0.0005624484748557296,
+      "loss": 4.6555,
       "step": 2626
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016867271228359438,
-      "loss": 2.4114,
+      "learning_rate": 0.000562242374278648,
+      "loss": 4.7439,
       "step": 2627
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0001686108821104699,
-      "loss": 2.6394,
+      "learning_rate": 0.0005620362737015664,
+      "loss": 5.6666,
       "step": 2628
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016854905193734542,
-      "loss": 2.2223,
+      "learning_rate": 0.0005618301731244848,
+      "loss": 4.7649,
       "step": 2629
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0001684872217642209,
-      "loss": 2.5472,
+      "learning_rate": 0.0005616240725474031,
+      "loss": 4.9463,
       "step": 2630
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.00016842539159109642,
-      "loss": 2.2727,
+      "learning_rate": 0.0005614179719703215,
+      "loss": 4.8409,
       "step": 2631
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016836356141797197,
-      "loss": 2.7625,
+      "learning_rate": 0.0005612118713932399,
+      "loss": 5.016,
       "step": 2632
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016830173124484748,
-      "loss": 2.1948,
+      "learning_rate": 0.0005610057708161584,
+      "loss": 4.6575,
       "step": 2633
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016823990107172297,
-      "loss": 2.0612,
+      "learning_rate": 0.0005607996702390767,
+      "loss": 4.4669,
       "step": 2634
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0001681780708985985,
-      "loss": 4.072,
+      "learning_rate": 0.0005605935696619951,
+      "loss": 4.8681,
       "step": 2635
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.000168116240725474,
-      "loss": 4.0679,
+      "learning_rate": 0.0005603874690849135,
+      "loss": 4.9413,
       "step": 2636
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016805441055234955,
-      "loss": 3.7747,
+      "learning_rate": 0.0005601813685078319,
+      "loss": 5.0674,
       "step": 2637
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016799258037922507,
-      "loss": 3.1678,
+      "learning_rate": 0.0005599752679307503,
+      "loss": 5.0034,
       "step": 2638
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016793075020610056,
-      "loss": 2.9582,
+      "learning_rate": 0.0005597691673536686,
+      "loss": 4.8807,
       "step": 2639
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016786892003297607,
-      "loss": 3.0511,
+      "learning_rate": 0.000559563066776587,
+      "loss": 5.1134,
       "step": 2640
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0001678070898598516,
-      "loss": 2.7964,
+      "learning_rate": 0.0005593569661995054,
+      "loss": 5.0428,
       "step": 2641
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0001677452596867271,
-      "loss": 3.1672,
+      "learning_rate": 0.0005591508656224238,
+      "loss": 4.8957,
       "step": 2642
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0001676834295136026,
-      "loss": 2.9085,
+      "learning_rate": 0.0005589447650453421,
+      "loss": 5.0808,
       "step": 2643
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016762159934047814,
-      "loss": 2.781,
+      "learning_rate": 0.0005587386644682605,
+      "loss": 4.8374,
       "step": 2644
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016755976916735366,
-      "loss": 3.0394,
+      "learning_rate": 0.0005585325638911789,
+      "loss": 5.1177,
       "step": 2645
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016749793899422917,
-      "loss": 2.7151,
+      "learning_rate": 0.0005583264633140973,
+      "loss": 4.8372,
       "step": 2646
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0001674361088211047,
-      "loss": 3.15,
+      "learning_rate": 0.0005581203627370157,
+      "loss": 5.0784,
       "step": 2647
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016737427864798018,
-      "loss": 2.593,
+      "learning_rate": 0.000557914262159934,
+      "loss": 4.7165,
       "step": 2648
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.00016731244847485572,
-      "loss": 2.8918,
+      "learning_rate": 0.0005577081615828524,
+      "loss": 4.9549,
       "step": 2649
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016725061830173124,
-      "loss": 2.5949,
+      "learning_rate": 0.0005575020610057708,
+      "loss": 4.7211,
       "step": 2650
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016718878812860676,
-      "loss": 2.9544,
+      "learning_rate": 0.0005572959604286892,
+      "loss": 4.9956,
       "step": 2651
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016712695795548227,
-      "loss": 2.587,
+      "learning_rate": 0.0005570898598516076,
+      "loss": 4.9969,
       "step": 2652
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016706512778235776,
-      "loss": 2.8332,
+      "learning_rate": 0.0005568837592745259,
+      "loss": 5.108,
       "step": 2653
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016700329760923328,
-      "loss": 2.7316,
+      "learning_rate": 0.0005566776586974444,
+      "loss": 5.428,
       "step": 2654
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016694146743610882,
-      "loss": 2.346,
+      "learning_rate": 0.0005564715581203628,
+      "loss": 5.005,
       "step": 2655
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016687963726298434,
-      "loss": 2.8693,
+      "learning_rate": 0.0005562654575432812,
+      "loss": 5.0474,
       "step": 2656
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016681780708985983,
-      "loss": 2.8812,
+      "learning_rate": 0.0005560593569661995,
+      "loss": 4.9852,
       "step": 2657
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016675597691673534,
-      "loss": 2.6721,
+      "learning_rate": 0.0005558532563891179,
+      "loss": 4.8614,
       "step": 2658
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016669414674361086,
-      "loss": 2.7682,
+      "learning_rate": 0.0005556471558120363,
+      "loss": 4.9481,
       "step": 2659
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016663231657048638,
-      "loss": 2.484,
+      "learning_rate": 0.0005554410552349547,
+      "loss": 5.0162,
       "step": 2660
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016657048639736192,
-      "loss": 2.4756,
+      "learning_rate": 0.0005552349546578731,
+      "loss": 4.9602,
       "step": 2661
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0001665086562242374,
-      "loss": 2.5982,
+      "learning_rate": 0.0005550288540807914,
+      "loss": 4.9237,
       "step": 2662
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016644682605111293,
-      "loss": 3.0729,
+      "learning_rate": 0.0005548227535037098,
+      "loss": 5.4024,
       "step": 2663
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016638499587798844,
-      "loss": 2.5105,
+      "learning_rate": 0.0005546166529266282,
+      "loss": 4.7845,
       "step": 2664
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016632316570486396,
-      "loss": 2.4997,
+      "learning_rate": 0.0005544105523495466,
+      "loss": 5.2859,
       "step": 2665
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.00016626133553173945,
-      "loss": 2.8506,
+      "learning_rate": 0.0005542044517724649,
+      "loss": 5.0567,
       "step": 2666
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.000166199505358615,
-      "loss": 2.8584,
+      "learning_rate": 0.0005539983511953833,
+      "loss": 5.3535,
       "step": 2667
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001661376751854905,
-      "loss": 2.395,
+      "learning_rate": 0.0005537922506183017,
+      "loss": 4.9933,
       "step": 2668
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016607584501236603,
-      "loss": 2.5646,
+      "learning_rate": 0.0005535861500412201,
+      "loss": 5.0895,
       "step": 2669
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016601401483924154,
-      "loss": 2.8362,
+      "learning_rate": 0.0005533800494641385,
+      "loss": 5.3002,
       "step": 2670
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016595218466611703,
-      "loss": 2.5413,
+      "learning_rate": 0.0005531739488870568,
+      "loss": 5.2384,
       "step": 2671
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016589035449299255,
-      "loss": 2.5415,
+      "learning_rate": 0.0005529678483099752,
+      "loss": 5.1788,
       "step": 2672
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001658285243198681,
-      "loss": 2.4246,
+      "learning_rate": 0.0005527617477328936,
+      "loss": 4.7467,
       "step": 2673
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001657666941467436,
-      "loss": 2.2838,
+      "learning_rate": 0.000552555647155812,
+      "loss": 4.81,
       "step": 2674
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001657048639736191,
-      "loss": 2.5872,
+      "learning_rate": 0.0005523495465787303,
+      "loss": 5.4366,
       "step": 2675
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016564303380049462,
-      "loss": 3.0265,
+      "learning_rate": 0.0005521434460016488,
+      "loss": 5.5348,
       "step": 2676
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016558120362737013,
-      "loss": 2.1652,
+      "learning_rate": 0.0005519373454245672,
+      "loss": 4.8261,
       "step": 2677
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016551937345424568,
-      "loss": 2.6219,
+      "learning_rate": 0.0005517312448474856,
+      "loss": 5.1766,
       "step": 2678
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001654575432811212,
-      "loss": 2.0934,
+      "learning_rate": 0.000551525144270404,
+      "loss": 4.9959,
       "step": 2679
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016539571310799668,
-      "loss": 1.9016,
+      "learning_rate": 0.0005513190436933223,
+      "loss": 4.6046,
       "step": 2680
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0001653338829348722,
-      "loss": 2.438,
+      "learning_rate": 0.0005511129431162407,
+      "loss": 4.7935,
       "step": 2681
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016527205276174772,
-      "loss": 2.0506,
+      "learning_rate": 0.0005509068425391591,
+      "loss": 4.7336,
       "step": 2682
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016521022258862323,
-      "loss": 2.076,
+      "learning_rate": 0.0005507007419620775,
+      "loss": 4.9861,
       "step": 2683
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.00016514839241549878,
-      "loss": 1.6177,
+      "learning_rate": 0.0005504946413849959,
+      "loss": 4.015,
       "step": 2684
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016508656224237427,
-      "loss": 4.6583,
+      "learning_rate": 0.0005502885408079142,
+      "loss": 5.3946,
       "step": 2685
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016502473206924978,
-      "loss": 4.2,
+      "learning_rate": 0.0005500824402308326,
+      "loss": 4.828,
       "step": 2686
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0001649629018961253,
-      "loss": 3.9585,
+      "learning_rate": 0.000549876339653751,
+      "loss": 4.921,
       "step": 2687
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016490107172300081,
-      "loss": 3.8862,
+      "learning_rate": 0.0005496702390766694,
+      "loss": 4.9531,
       "step": 2688
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0001648392415498763,
-      "loss": 3.3204,
+      "learning_rate": 0.0005494641384995877,
+      "loss": 4.8759,
       "step": 2689
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016477741137675182,
-      "loss": 2.791,
+      "learning_rate": 0.0005492580379225061,
+      "loss": 4.8267,
       "step": 2690
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016471558120362736,
-      "loss": 2.5226,
+      "learning_rate": 0.0005490519373454245,
+      "loss": 4.871,
       "step": 2691
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016465375103050288,
-      "loss": 2.7821,
+      "learning_rate": 0.0005488458367683429,
+      "loss": 5.0513,
       "step": 2692
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0001645919208573784,
-      "loss": 2.6078,
+      "learning_rate": 0.0005486397361912613,
+      "loss": 4.9383,
       "step": 2693
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0001645300906842539,
-      "loss": 2.7697,
+      "learning_rate": 0.0005484336356141796,
+      "loss": 4.8699,
       "step": 2694
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0001644682605111294,
-      "loss": 2.9318,
+      "learning_rate": 0.000548227535037098,
+      "loss": 5.255,
       "step": 2695
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016440643033800495,
-      "loss": 2.5349,
+      "learning_rate": 0.0005480214344600166,
+      "loss": 4.8584,
       "step": 2696
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016434460016488046,
-      "loss": 2.5258,
+      "learning_rate": 0.000547815333882935,
+      "loss": 4.7653,
       "step": 2697
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016428276999175595,
-      "loss": 2.2961,
+      "learning_rate": 0.0005476092333058533,
+      "loss": 4.6637,
       "step": 2698
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016422093981863147,
-      "loss": 2.7035,
+      "learning_rate": 0.0005474031327287717,
+      "loss": 5.0332,
       "step": 2699
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.000164159109645507,
-      "loss": 3.0227,
+      "learning_rate": 0.0005471970321516901,
+      "loss": 5.6386,
       "step": 2700
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0001640972794723825,
-      "loss": 2.5857,
+      "learning_rate": 0.0005469909315746085,
+      "loss": 4.836,
       "step": 2701
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.00016403544929925805,
-      "loss": 2.2756,
+      "learning_rate": 0.0005467848309975269,
+      "loss": 4.5148,
       "step": 2702
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016397361912613354,
-      "loss": 2.3201,
+      "learning_rate": 0.0005465787304204452,
+      "loss": 4.712,
       "step": 2703
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016391178895300905,
-      "loss": 3.1544,
+      "learning_rate": 0.0005463726298433636,
+      "loss": 5.0119,
       "step": 2704
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016384995877988457,
-      "loss": 2.4463,
+      "learning_rate": 0.000546166529266282,
+      "loss": 4.7069,
       "step": 2705
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016378812860676009,
-      "loss": 2.6774,
+      "learning_rate": 0.0005459604286892004,
+      "loss": 5.0023,
       "step": 2706
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016372629843363558,
-      "loss": 2.296,
+      "learning_rate": 0.0005457543281121187,
+      "loss": 4.8547,
       "step": 2707
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0001636644682605111,
-      "loss": 2.3508,
+      "learning_rate": 0.0005455482275350371,
+      "loss": 4.9683,
       "step": 2708
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016360263808738664,
-      "loss": 2.4484,
+      "learning_rate": 0.0005453421269579555,
+      "loss": 4.9358,
       "step": 2709
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016354080791426215,
-      "loss": 2.2664,
+      "learning_rate": 0.0005451360263808739,
+      "loss": 4.7607,
       "step": 2710
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016347897774113767,
-      "loss": 2.0547,
+      "learning_rate": 0.0005449299258037923,
+      "loss": 4.6785,
       "step": 2711
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016341714756801316,
-      "loss": 2.3571,
+      "learning_rate": 0.0005447238252267106,
+      "loss": 4.9307,
       "step": 2712
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016335531739488868,
-      "loss": 2.225,
+      "learning_rate": 0.000544517724649629,
+      "loss": 5.212,
       "step": 2713
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016329348722176422,
-      "loss": 2.7634,
+      "learning_rate": 0.0005443116240725475,
+      "loss": 5.3907,
       "step": 2714
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016323165704863974,
-      "loss": 2.1272,
+      "learning_rate": 0.0005441055234954659,
+      "loss": 4.5373,
       "step": 2715
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016316982687551525,
-      "loss": 2.4398,
+      "learning_rate": 0.0005438994229183843,
+      "loss": 5.2777,
       "step": 2716
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016310799670239074,
-      "loss": 2.6117,
+      "learning_rate": 0.0005436933223413026,
+      "loss": 5.0851,
       "step": 2717
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016304616652926626,
-      "loss": 2.0208,
+      "learning_rate": 0.000543487221764221,
+      "loss": 4.7496,
       "step": 2718
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016298433635614177,
-      "loss": 2.1412,
+      "learning_rate": 0.0005432811211871394,
+      "loss": 4.6809,
       "step": 2719
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.00016292250618301732,
-      "loss": 2.0923,
+      "learning_rate": 0.0005430750206100578,
+      "loss": 5.1033,
       "step": 2720
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0001628606760098928,
-      "loss": 2.2168,
+      "learning_rate": 0.0005428689200329761,
+      "loss": 4.9552,
       "step": 2721
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016279884583676832,
-      "loss": 2.1757,
+      "learning_rate": 0.0005426628194558945,
+      "loss": 4.9282,
       "step": 2722
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016273701566364384,
-      "loss": 2.3183,
+      "learning_rate": 0.0005424567188788129,
+      "loss": 4.9737,
       "step": 2723
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016267518549051936,
-      "loss": 2.3195,
+      "learning_rate": 0.0005422506183017313,
+      "loss": 4.8318,
       "step": 2724
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0001626133553173949,
-      "loss": 2.4852,
+      "learning_rate": 0.0005420445177246497,
+      "loss": 5.1933,
       "step": 2725
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0001625515251442704,
-      "loss": 2.4038,
+      "learning_rate": 0.000541838417147568,
+      "loss": 4.9295,
       "step": 2726
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0001624896949711459,
-      "loss": 2.5227,
+      "learning_rate": 0.0005416323165704864,
+      "loss": 5.0298,
       "step": 2727
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016242786479802142,
-      "loss": 2.1762,
+      "learning_rate": 0.0005414262159934048,
+      "loss": 4.85,
       "step": 2728
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016236603462489694,
-      "loss": 2.408,
+      "learning_rate": 0.0005412201154163232,
+      "loss": 4.9146,
       "step": 2729
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016230420445177243,
-      "loss": 2.6218,
+      "learning_rate": 0.0005410140148392415,
+      "loss": 5.6855,
       "step": 2730
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016224237427864795,
-      "loss": 2.0886,
+      "learning_rate": 0.0005408079142621599,
+      "loss": 4.5711,
       "step": 2731
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0001621805441055235,
-      "loss": 2.227,
+      "learning_rate": 0.0005406018136850783,
+      "loss": 4.3342,
       "step": 2732
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.000162118713932399,
-      "loss": 3.1961,
+      "learning_rate": 0.0005403957131079967,
+      "loss": 5.4346,
       "step": 2733
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016205688375927452,
-      "loss": 1.8819,
+      "learning_rate": 0.0005401896125309152,
+      "loss": 4.5293,
       "step": 2734
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016199505358615,
-      "loss": 4.3616,
+      "learning_rate": 0.0005399835119538334,
+      "loss": 5.8772,
       "step": 2735
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016193322341302553,
-      "loss": 3.9595,
+      "learning_rate": 0.0005397774113767519,
+      "loss": 5.1576,
       "step": 2736
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.00016187139323990105,
-      "loss": 3.8323,
+      "learning_rate": 0.0005395713107996703,
+      "loss": 5.1358,
       "step": 2737
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0001618095630667766,
-      "loss": 3.4704,
+      "learning_rate": 0.0005393652102225887,
+      "loss": 4.8923,
       "step": 2738
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0001617477328936521,
-      "loss": 3.0182,
+      "learning_rate": 0.0005391591096455071,
+      "loss": 4.9491,
       "step": 2739
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0001616859027205276,
-      "loss": 2.6945,
+      "learning_rate": 0.0005389530090684254,
+      "loss": 4.7672,
       "step": 2740
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0001616240725474031,
-      "loss": 2.7088,
+      "learning_rate": 0.0005387469084913438,
+      "loss": 5.1035,
       "step": 2741
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016156224237427863,
-      "loss": 2.5472,
+      "learning_rate": 0.0005385408079142622,
+      "loss": 4.8735,
       "step": 2742
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016150041220115417,
-      "loss": 2.4927,
+      "learning_rate": 0.0005383347073371806,
+      "loss": 4.9367,
       "step": 2743
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016143858202802966,
-      "loss": 2.6783,
+      "learning_rate": 0.0005381286067600989,
+      "loss": 5.1649,
       "step": 2744
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016137675185490518,
-      "loss": 2.7301,
+      "learning_rate": 0.0005379225061830173,
+      "loss": 5.1688,
       "step": 2745
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0001613149216817807,
-      "loss": 2.799,
+      "learning_rate": 0.0005377164056059357,
+      "loss": 5.3375,
       "step": 2746
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0001612530915086562,
-      "loss": 2.4827,
+      "learning_rate": 0.0005375103050288541,
+      "loss": 5.0156,
       "step": 2747
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016119126133553173,
-      "loss": 2.865,
+      "learning_rate": 0.0005373042044517725,
+      "loss": 5.1242,
       "step": 2748
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016112943116240722,
-      "loss": 3.1159,
+      "learning_rate": 0.0005370981038746908,
+      "loss": 5.7294,
       "step": 2749
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016106760098928276,
-      "loss": 3.1228,
+      "learning_rate": 0.0005368920032976092,
+      "loss": 5.8141,
       "step": 2750
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016100577081615828,
-      "loss": 2.5848,
+      "learning_rate": 0.0005366859027205276,
+      "loss": 4.907,
       "step": 2751
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0001609439406430338,
-      "loss": 2.219,
+      "learning_rate": 0.000536479802143446,
+      "loss": 4.7108,
       "step": 2752
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016088211046990928,
-      "loss": 2.0842,
+      "learning_rate": 0.0005362737015663643,
+      "loss": 4.6842,
       "step": 2753
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0001608202802967848,
-      "loss": 2.4874,
+      "learning_rate": 0.0005360676009892827,
+      "loss": 5.1712,
       "step": 2754
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016075845012366034,
-      "loss": 2.7381,
+      "learning_rate": 0.0005358615004122011,
+      "loss": 5.1099,
       "step": 2755
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.00016069661995053586,
-      "loss": 2.3147,
+      "learning_rate": 0.0005356553998351196,
+      "loss": 5.0424,
       "step": 2756
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016063478977741138,
-      "loss": 2.9061,
+      "learning_rate": 0.000535449299258038,
+      "loss": 5.1221,
       "step": 2757
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016057295960428687,
-      "loss": 2.6112,
+      "learning_rate": 0.0005352431986809563,
+      "loss": 5.7373,
       "step": 2758
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016051112943116238,
-      "loss": 2.3287,
+      "learning_rate": 0.0005350370981038747,
+      "loss": 5.2545,
       "step": 2759
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0001604492992580379,
-      "loss": 2.6183,
+      "learning_rate": 0.0005348309975267931,
+      "loss": 5.0077,
       "step": 2760
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016038746908491344,
-      "loss": 2.3325,
+      "learning_rate": 0.0005346248969497115,
+      "loss": 4.8473,
       "step": 2761
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016032563891178893,
-      "loss": 2.298,
+      "learning_rate": 0.0005344187963726298,
+      "loss": 5.0185,
       "step": 2762
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016026380873866445,
-      "loss": 2.2323,
+      "learning_rate": 0.0005342126957955482,
+      "loss": 4.9529,
       "step": 2763
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016020197856553997,
-      "loss": 2.4638,
+      "learning_rate": 0.0005340065952184666,
+      "loss": 5.1013,
       "step": 2764
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00016014014839241548,
-      "loss": 2.042,
+      "learning_rate": 0.000533800494641385,
+      "loss": 4.6892,
       "step": 2765
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.000160078318219291,
-      "loss": 1.7245,
+      "learning_rate": 0.0005335943940643034,
+      "loss": 4.5607,
       "step": 2766
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0001600164880461665,
-      "loss": 2.1383,
+      "learning_rate": 0.0005333882934872217,
+      "loss": 4.9135,
       "step": 2767
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00015995465787304203,
-      "loss": 2.1221,
+      "learning_rate": 0.0005331821929101401,
+      "loss": 5.1268,
       "step": 2768
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00015989282769991755,
-      "loss": 1.6607,
+      "learning_rate": 0.0005329760923330585,
+      "loss": 4.4053,
       "step": 2769
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00015983099752679307,
-      "loss": 2.0488,
+      "learning_rate": 0.0005327699917559769,
+      "loss": 4.7188,
       "step": 2770
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00015976916735366858,
-      "loss": 2.0226,
+      "learning_rate": 0.0005325638911788953,
+      "loss": 4.8435,
       "step": 2771
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00015970733718054407,
-      "loss": 1.9604,
+      "learning_rate": 0.0005323577906018136,
+      "loss": 4.752,
       "step": 2772
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00015964550700741962,
-      "loss": 2.042,
+      "learning_rate": 0.000532151690024732,
+      "loss": 4.8333,
       "step": 2773
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.00015958367683429513,
-      "loss": 1.977,
+      "learning_rate": 0.0005319455894476504,
+      "loss": 5.0294,
       "step": 2774
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015952184666117065,
-      "loss": 1.711,
+      "learning_rate": 0.0005317394888705688,
+      "loss": 4.5488,
       "step": 2775
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015946001648804614,
-      "loss": 1.9545,
+      "learning_rate": 0.0005315333882934871,
+      "loss": 4.9429,
       "step": 2776
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015939818631492166,
-      "loss": 2.1134,
+      "learning_rate": 0.0005313272877164056,
+      "loss": 5.6044,
       "step": 2777
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015933635614179717,
-      "loss": 1.9228,
+      "learning_rate": 0.000531121187139324,
+      "loss": 4.9306,
       "step": 2778
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015927452596867272,
-      "loss": 1.6552,
+      "learning_rate": 0.0005309150865622424,
+      "loss": 4.769,
       "step": 2779
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015921269579554823,
-      "loss": 1.9514,
+      "learning_rate": 0.0005307089859851608,
+      "loss": 4.7281,
       "step": 2780
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015915086562242372,
-      "loss": 1.3734,
+      "learning_rate": 0.0005305028854080791,
+      "loss": 4.1484,
       "step": 2781
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015908903544929924,
-      "loss": 1.5686,
+      "learning_rate": 0.0005302967848309975,
+      "loss": 4.5629,
       "step": 2782
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015902720527617475,
-      "loss": 2.0781,
+      "learning_rate": 0.0005300906842539159,
+      "loss": 4.6512,
       "step": 2783
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0001589653751030503,
-      "loss": 1.5567,
+      "learning_rate": 0.0005298845836768343,
+      "loss": 4.5502,
       "step": 2784
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015890354492992576,
-      "loss": 3.6594,
+      "learning_rate": 0.0005296784830997526,
+      "loss": 5.3159,
       "step": 2785
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0001588417147568013,
-      "loss": 3.4339,
+      "learning_rate": 0.000529472382522671,
+      "loss": 5.0415,
       "step": 2786
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015877988458367682,
-      "loss": 3.297,
+      "learning_rate": 0.0005292662819455894,
+      "loss": 5.0897,
       "step": 2787
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015871805441055234,
-      "loss": 3.7468,
+      "learning_rate": 0.0005290601813685078,
+      "loss": 4.9424,
       "step": 2788
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015865622423742785,
-      "loss": 2.4749,
+      "learning_rate": 0.0005288540807914263,
+      "loss": 4.6518,
       "step": 2789
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.00015859439406430334,
-      "loss": 2.4171,
+      "learning_rate": 0.0005286479802143446,
+      "loss": 4.9314,
       "step": 2790
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0001585325638911789,
-      "loss": 2.3011,
+      "learning_rate": 0.000528441879637263,
+      "loss": 4.7148,
       "step": 2791
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0001584707337180544,
-      "loss": 2.3538,
+      "learning_rate": 0.0005282357790601814,
+      "loss": 5.095,
       "step": 2792
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015840890354492992,
-      "loss": 2.2994,
+      "learning_rate": 0.0005280296784830998,
+      "loss": 4.9162,
       "step": 2793
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0001583470733718054,
-      "loss": 2.3419,
+      "learning_rate": 0.0005278235779060181,
+      "loss": 4.9853,
       "step": 2794
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015828524319868093,
-      "loss": 2.3047,
+      "learning_rate": 0.0005276174773289365,
+      "loss": 4.8752,
       "step": 2795
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015822341302555644,
-      "loss": 2.2876,
+      "learning_rate": 0.000527411376751855,
+      "loss": 4.8529,
       "step": 2796
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.000158161582852432,
-      "loss": 2.3473,
+      "learning_rate": 0.0005272052761747734,
+      "loss": 4.8728,
       "step": 2797
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0001580997526793075,
-      "loss": 2.4376,
+      "learning_rate": 0.0005269991755976918,
+      "loss": 5.0403,
       "step": 2798
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.000158037922506183,
-      "loss": 2.409,
+      "learning_rate": 0.0005267930750206101,
+      "loss": 4.9942,
       "step": 2799
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0001579760923330585,
-      "loss": 2.3342,
+      "learning_rate": 0.0005265869744435285,
+      "loss": 5.5591,
       "step": 2800
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015791426215993403,
-      "loss": 2.0836,
+      "learning_rate": 0.0005263808738664469,
+      "loss": 4.7695,
       "step": 2801
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015785243198680957,
-      "loss": 1.9855,
+      "learning_rate": 0.0005261747732893653,
+      "loss": 4.6509,
       "step": 2802
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0001577906018136851,
-      "loss": 2.0909,
+      "learning_rate": 0.0005259686727122837,
+      "loss": 4.8735,
       "step": 2803
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015772877164056058,
-      "loss": 2.0008,
+      "learning_rate": 0.000525762572135202,
+      "loss": 5.4841,
       "step": 2804
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0001576669414674361,
-      "loss": 2.0894,
+      "learning_rate": 0.0005255564715581204,
+      "loss": 4.9709,
       "step": 2805
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0001576051112943116,
-      "loss": 2.1686,
+      "learning_rate": 0.0005253503709810388,
+      "loss": 5.0337,
       "step": 2806
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015754328112118713,
-      "loss": 3.5704,
+      "learning_rate": 0.0005251442704039572,
+      "loss": 5.4682,
       "step": 2807
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015748145094806262,
-      "loss": 2.2401,
+      "learning_rate": 0.0005249381698268755,
+      "loss": 4.7683,
       "step": 2808
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.00015741962077493816,
-      "loss": 1.7816,
+      "learning_rate": 0.0005247320692497939,
+      "loss": 4.7322,
       "step": 2809
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015735779060181368,
-      "loss": 1.6113,
+      "learning_rate": 0.0005245259686727123,
+      "loss": 4.4981,
       "step": 2810
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0001572959604286892,
-      "loss": 1.9391,
+      "learning_rate": 0.0005243198680956307,
+      "loss": 4.6683,
       "step": 2811
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0001572341302555647,
-      "loss": 1.9826,
+      "learning_rate": 0.0005241137675185491,
+      "loss": 5.1613,
       "step": 2812
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0001571723000824402,
-      "loss": 2.3286,
+      "learning_rate": 0.0005239076669414674,
+      "loss": 5.0958,
       "step": 2813
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015711046990931571,
-      "loss": 2.1256,
+      "learning_rate": 0.0005237015663643858,
+      "loss": 5.0264,
       "step": 2814
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015704863973619126,
-      "loss": 1.9361,
+      "learning_rate": 0.0005234954657873042,
+      "loss": 4.8238,
       "step": 2815
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015698680956306678,
-      "loss": 1.9241,
+      "learning_rate": 0.0005232893652102227,
+      "loss": 4.9071,
       "step": 2816
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015692497938994226,
-      "loss": 1.9624,
+      "learning_rate": 0.000523083264633141,
+      "loss": 4.7696,
       "step": 2817
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015686314921681778,
-      "loss": 1.5856,
+      "learning_rate": 0.0005228771640560594,
+      "loss": 4.6439,
       "step": 2818
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0001568013190436933,
-      "loss": 1.9923,
+      "learning_rate": 0.0005226710634789778,
+      "loss": 4.8542,
       "step": 2819
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015673948887056884,
-      "loss": 1.8694,
+      "learning_rate": 0.0005224649629018962,
+      "loss": 4.6634,
       "step": 2820
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015667765869744436,
-      "loss": 1.9755,
+      "learning_rate": 0.0005222588623248146,
+      "loss": 4.6023,
       "step": 2821
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015661582852431985,
-      "loss": 2.0079,
+      "learning_rate": 0.0005220527617477329,
+      "loss": 5.0339,
       "step": 2822
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015655399835119536,
-      "loss": 1.9367,
+      "learning_rate": 0.0005218466611706513,
+      "loss": 4.8188,
       "step": 2823
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015649216817807088,
-      "loss": 2.0479,
+      "learning_rate": 0.0005216405605935697,
+      "loss": 5.0043,
       "step": 2824
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0001564303380049464,
-      "loss": 2.251,
+      "learning_rate": 0.0005214344600164881,
+      "loss": 5.1227,
       "step": 2825
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015636850783182194,
-      "loss": 1.7367,
+      "learning_rate": 0.0005212283594394065,
+      "loss": 4.7445,
       "step": 2826
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.00015630667765869743,
-      "loss": 2.0058,
+      "learning_rate": 0.0005210222588623248,
+      "loss": 4.9777,
       "step": 2827
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015624484748557295,
-      "loss": 2.1392,
+      "learning_rate": 0.0005208161582852432,
+      "loss": 5.2002,
       "step": 2828
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015618301731244846,
-      "loss": 1.4746,
+      "learning_rate": 0.0005206100577081616,
+      "loss": 4.9054,
       "step": 2829
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015612118713932398,
-      "loss": 2.9054,
+      "learning_rate": 0.00052040395713108,
+      "loss": 5.0842,
       "step": 2830
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015605935696619947,
-      "loss": 1.4517,
+      "learning_rate": 0.0005201978565539983,
+      "loss": 4.2801,
       "step": 2831
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.000155997526793075,
-      "loss": 2.1924,
+      "learning_rate": 0.0005199917559769167,
+      "loss": 4.5727,
       "step": 2832
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015593569661995053,
-      "loss": 1.6485,
+      "learning_rate": 0.0005197856553998351,
+      "loss": 4.4683,
       "step": 2833
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015587386644682605,
-      "loss": 2.2751,
+      "learning_rate": 0.0005195795548227535,
+      "loss": 4.9113,
       "step": 2834
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015581203627370156,
-      "loss": 3.7393,
+      "learning_rate": 0.000519373454245672,
+      "loss": 4.9849,
       "step": 2835
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015575020610057705,
-      "loss": 3.4227,
+      "learning_rate": 0.0005191673536685902,
+      "loss": 4.8494,
       "step": 2836
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015568837592745257,
-      "loss": 3.1555,
+      "learning_rate": 0.0005189612530915087,
+      "loss": 4.6293,
       "step": 2837
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0001556265457543281,
-      "loss": 2.8506,
+      "learning_rate": 0.0005187551525144271,
+      "loss": 4.832,
       "step": 2838
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015556471558120363,
-      "loss": 2.5871,
+      "learning_rate": 0.0005185490519373455,
+      "loss": 4.8129,
       "step": 2839
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015550288540807912,
-      "loss": 2.3073,
+      "learning_rate": 0.0005183429513602638,
+      "loss": 4.6738,
       "step": 2840
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015544105523495464,
-      "loss": 2.9323,
+      "learning_rate": 0.0005181368507831822,
+      "loss": 5.3822,
       "step": 2841
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015537922506183015,
-      "loss": 2.4191,
+      "learning_rate": 0.0005179307502061006,
+      "loss": 5.0753,
       "step": 2842
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00015531739488870567,
-      "loss": 2.1873,
+      "learning_rate": 0.000517724649629019,
+      "loss": 5.0748,
       "step": 2843
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0001552555647155812,
-      "loss": 2.8111,
+      "learning_rate": 0.0005175185490519374,
+      "loss": 5.2758,
       "step": 2844
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0001551937345424567,
-      "loss": 2.2178,
+      "learning_rate": 0.0005173124484748557,
+      "loss": 4.8947,
       "step": 2845
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015513190436933222,
-      "loss": 2.3509,
+      "learning_rate": 0.0005171063478977741,
+      "loss": 4.8727,
       "step": 2846
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015507007419620774,
-      "loss": 2.215,
+      "learning_rate": 0.0005169002473206925,
+      "loss": 5.3107,
       "step": 2847
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015500824402308325,
-      "loss": 2.3248,
+      "learning_rate": 0.0005166941467436109,
+      "loss": 5.0071,
       "step": 2848
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015494641384995874,
-      "loss": 2.1648,
+      "learning_rate": 0.0005164880461665292,
+      "loss": 4.8001,
       "step": 2849
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015488458367683428,
-      "loss": 2.5769,
+      "learning_rate": 0.0005162819455894476,
+      "loss": 4.8467,
       "step": 2850
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0001548227535037098,
-      "loss": 2.2035,
+      "learning_rate": 0.000516075845012366,
+      "loss": 4.9072,
       "step": 2851
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015476092333058532,
-      "loss": 2.2154,
+      "learning_rate": 0.0005158697444352844,
+      "loss": 4.8425,
       "step": 2852
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015469909315746083,
-      "loss": 2.0205,
+      "learning_rate": 0.0005156636438582028,
+      "loss": 5.0271,
       "step": 2853
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015463726298433632,
-      "loss": 1.6901,
+      "learning_rate": 0.0005154575432811211,
+      "loss": 4.6145,
       "step": 2854
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015457543281121184,
-      "loss": 1.7219,
+      "learning_rate": 0.0005152514427040395,
+      "loss": 4.4781,
       "step": 2855
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015451360263808738,
-      "loss": 2.2109,
+      "learning_rate": 0.000515045342126958,
+      "loss": 4.7813,
       "step": 2856
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0001544517724649629,
-      "loss": 1.9504,
+      "learning_rate": 0.0005148392415498764,
+      "loss": 4.8357,
       "step": 2857
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015438994229183842,
-      "loss": 2.0062,
+      "learning_rate": 0.0005146331409727948,
+      "loss": 4.8528,
       "step": 2858
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0001543281121187139,
-      "loss": 1.8702,
+      "learning_rate": 0.0005144270403957131,
+      "loss": 5.5872,
       "step": 2859
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015426628194558942,
-      "loss": 1.8819,
+      "learning_rate": 0.0005142209398186315,
+      "loss": 4.8197,
       "step": 2860
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015420445177246497,
-      "loss": 2.1618,
+      "learning_rate": 0.0005140148392415499,
+      "loss": 5.1407,
       "step": 2861
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015414262159934048,
-      "loss": 2.0189,
+      "learning_rate": 0.0005138087386644683,
+      "loss": 4.8898,
       "step": 2862
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.00015408079142621597,
-      "loss": 2.0976,
+      "learning_rate": 0.0005136026380873866,
+      "loss": 5.065,
       "step": 2863
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0001540189612530915,
-      "loss": 1.9925,
+      "learning_rate": 0.000513396537510305,
+      "loss": 4.9242,
       "step": 2864
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.000153957131079967,
-      "loss": 1.5689,
+      "learning_rate": 0.0005131904369332234,
+      "loss": 4.5865,
       "step": 2865
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015389530090684252,
-      "loss": 2.0997,
+      "learning_rate": 0.0005129843363561418,
+      "loss": 4.9237,
       "step": 2866
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015383347073371807,
-      "loss": 1.9185,
+      "learning_rate": 0.0005127782357790602,
+      "loss": 4.8197,
       "step": 2867
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015377164056059356,
-      "loss": 1.7041,
+      "learning_rate": 0.0005125721352019785,
+      "loss": 4.6222,
       "step": 2868
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015370981038746907,
-      "loss": 1.8922,
+      "learning_rate": 0.0005123660346248969,
+      "loss": 5.5099,
       "step": 2869
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0001536479802143446,
-      "loss": 1.6059,
+      "learning_rate": 0.0005121599340478153,
+      "loss": 4.6911,
       "step": 2870
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0001535861500412201,
-      "loss": 1.7314,
+      "learning_rate": 0.0005119538334707337,
+      "loss": 4.7161,
       "step": 2871
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0001535243198680956,
-      "loss": 1.6381,
+      "learning_rate": 0.000511747732893652,
+      "loss": 4.8067,
       "step": 2872
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0001534624896949711,
-      "loss": 1.9629,
+      "learning_rate": 0.0005115416323165704,
+      "loss": 5.6408,
       "step": 2873
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015340065952184666,
-      "loss": 1.9421,
+      "learning_rate": 0.0005113355317394888,
+      "loss": 4.9481,
       "step": 2874
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015333882934872217,
-      "loss": 2.3741,
+      "learning_rate": 0.0005111294311624072,
+      "loss": 5.2645,
       "step": 2875
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0001532769991755977,
-      "loss": 1.6434,
+      "learning_rate": 0.0005109233305853256,
+      "loss": 4.8979,
       "step": 2876
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015321516900247318,
-      "loss": 2.0476,
+      "learning_rate": 0.000510717230008244,
+      "loss": 4.8281,
       "step": 2877
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0001531533388293487,
-      "loss": 1.8874,
+      "learning_rate": 0.0005105111294311624,
+      "loss": 5.0836,
       "step": 2878
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015309150865622424,
-      "loss": 1.5345,
+      "learning_rate": 0.0005103050288540808,
+      "loss": 4.7873,
       "step": 2879
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015302967848309976,
-      "loss": 1.5439,
+      "learning_rate": 0.0005100989282769992,
+      "loss": 4.5298,
       "step": 2880
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.00015296784830997524,
-      "loss": 1.5081,
+      "learning_rate": 0.0005098928276999175,
+      "loss": 4.7793,
       "step": 2881
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015290601813685076,
-      "loss": 2.7669,
+      "learning_rate": 0.0005096867271228359,
+      "loss": 5.8203,
       "step": 2882
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015284418796372628,
-      "loss": 2.0549,
+      "learning_rate": 0.0005094806265457544,
+      "loss": 4.9854,
       "step": 2883
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015284418796372628,
-      "loss": 1.8911,
+      "learning_rate": 0.0005092745259686728,
+      "loss": 4.6577,
       "step": 2884
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0001527823577906018,
-      "loss": 2.6291,
+      "learning_rate": 0.0005090684253915912,
+      "loss": 5.2622,
       "step": 2885
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015272052761747734,
-      "loss": 2.3613,
+      "learning_rate": 0.0005088623248145095,
+      "loss": 4.8151,
       "step": 2886
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015265869744435283,
-      "loss": 2.4044,
+      "learning_rate": 0.0005086562242374279,
+      "loss": 4.9556,
       "step": 2887
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015259686727122834,
-      "loss": 2.046,
+      "learning_rate": 0.0005084501236603463,
+      "loss": 4.926,
       "step": 2888
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015253503709810386,
-      "loss": 1.9627,
+      "learning_rate": 0.0005082440230832647,
+      "loss": 4.7872,
       "step": 2889
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015247320692497938,
-      "loss": 1.8253,
+      "learning_rate": 0.0005080379225061831,
+      "loss": 4.8647,
       "step": 2890
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015241137675185492,
-      "loss": 1.8892,
+      "learning_rate": 0.0005078318219291014,
+      "loss": 4.793,
       "step": 2891
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015234954657873038,
-      "loss": 2.2374,
+      "learning_rate": 0.0005076257213520198,
+      "loss": 4.9721,
       "step": 2892
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015228771640560593,
-      "loss": 2.5947,
+      "learning_rate": 0.0005074196207749382,
+      "loss": 5.0686,
       "step": 2893
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015222588623248144,
-      "loss": 1.8722,
+      "learning_rate": 0.0005072135201978566,
+      "loss": 4.7767,
       "step": 2894
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015216405605935696,
-      "loss": 2.1414,
+      "learning_rate": 0.0005070074196207749,
+      "loss": 4.9068,
       "step": 2895
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015210222588623245,
-      "loss": 1.9739,
+      "learning_rate": 0.0005068013190436933,
+      "loss": 4.7194,
       "step": 2896
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015204039571310797,
-      "loss": 2.3242,
+      "learning_rate": 0.0005065952184666118,
+      "loss": 4.9561,
       "step": 2897
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0001519785655399835,
-      "loss": 1.8078,
+      "learning_rate": 0.0005063891178895302,
+      "loss": 4.8854,
       "step": 2898
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.00015191673536685903,
-      "loss": 1.6249,
+      "learning_rate": 0.0005061830173124486,
+      "loss": 4.669,
       "step": 2899
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015185490519373454,
-      "loss": 1.8405,
+      "learning_rate": 0.0005059769167353669,
+      "loss": 4.8871,
       "step": 2900
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015179307502061003,
-      "loss": 1.8413,
+      "learning_rate": 0.0005057708161582853,
+      "loss": 4.8804,
       "step": 2901
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015173124484748555,
-      "loss": 1.8677,
+      "learning_rate": 0.0005055647155812037,
+      "loss": 5.0735,
       "step": 2902
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015166941467436107,
-      "loss": 1.8556,
+      "learning_rate": 0.0005053586150041221,
+      "loss": 4.9173,
       "step": 2903
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0001516075845012366,
-      "loss": 2.0828,
+      "learning_rate": 0.0005051525144270404,
+      "loss": 4.9308,
       "step": 2904
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0001515457543281121,
-      "loss": 1.8746,
+      "learning_rate": 0.0005049464138499588,
+      "loss": 4.956,
       "step": 2905
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015148392415498762,
-      "loss": 1.8398,
+      "learning_rate": 0.0005047403132728772,
+      "loss": 4.9552,
       "step": 2906
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015142209398186313,
-      "loss": 1.9337,
+      "learning_rate": 0.0005045342126957956,
+      "loss": 5.2224,
       "step": 2907
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015136026380873865,
-      "loss": 1.9942,
+      "learning_rate": 0.000504328112118714,
+      "loss": 5.0884,
       "step": 2908
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0001512984336356142,
-      "loss": 1.594,
+      "learning_rate": 0.0005041220115416323,
+      "loss": 4.9645,
       "step": 2909
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015123660346248968,
-      "loss": 2.0071,
+      "learning_rate": 0.0005039159109645507,
+      "loss": 5.2955,
       "step": 2910
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0001511747732893652,
-      "loss": 1.8773,
+      "learning_rate": 0.0005037098103874691,
+      "loss": 4.808,
       "step": 2911
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015111294311624072,
-      "loss": 1.5223,
+      "learning_rate": 0.0005035037098103875,
+      "loss": 4.5859,
       "step": 2912
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015105111294311623,
-      "loss": 2.6012,
+      "learning_rate": 0.0005032976092333058,
+      "loss": 5.5587,
       "step": 2913
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015098928276999172,
-      "loss": 2.236,
+      "learning_rate": 0.0005030915086562242,
+      "loss": 4.9497,
       "step": 2914
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015092745259686724,
-      "loss": 1.7925,
+      "learning_rate": 0.0005028854080791426,
+      "loss": 4.7373,
       "step": 2915
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.00015086562242374278,
-      "loss": 1.5674,
+      "learning_rate": 0.000502679307502061,
+      "loss": 4.4535,
       "step": 2916
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0001508037922506183,
-      "loss": 1.3976,
+      "learning_rate": 0.0005024732069249795,
+      "loss": 4.6037,
       "step": 2917
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015074196207749381,
-      "loss": 2.2731,
+      "learning_rate": 0.0005022671063478978,
+      "loss": 5.2585,
       "step": 2918
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0001506801319043693,
-      "loss": 1.5814,
+      "learning_rate": 0.0005020610057708162,
+      "loss": 4.6104,
       "step": 2919
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015061830173124482,
-      "loss": 1.924,
+      "learning_rate": 0.0005018549051937346,
+      "loss": 4.8446,
       "step": 2920
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015055647155812034,
-      "loss": 1.534,
+      "learning_rate": 0.000501648804616653,
+      "loss": 4.651,
       "step": 2921
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015049464138499588,
-      "loss": 1.6826,
+      "learning_rate": 0.0005014427040395714,
+      "loss": 4.4523,
       "step": 2922
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0001504328112118714,
-      "loss": 1.7645,
+      "learning_rate": 0.0005012366034624897,
+      "loss": 5.0576,
       "step": 2923
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0001503709810387469,
-      "loss": 1.5438,
+      "learning_rate": 0.0005010305028854081,
+      "loss": 4.9897,
       "step": 2924
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0001503091508656224,
-      "loss": 1.6638,
+      "learning_rate": 0.0005008244023083265,
+      "loss": 4.9498,
       "step": 2925
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015024732069249792,
-      "loss": 1.9201,
+      "learning_rate": 0.0005006183017312449,
+      "loss": 5.13,
       "step": 2926
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015018549051937346,
-      "loss": 1.7309,
+      "learning_rate": 0.0005004122011541632,
+      "loss": 4.7719,
       "step": 2927
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015012366034624895,
-      "loss": 2.1283,
+      "learning_rate": 0.0005002061005770816,
+      "loss": 4.9221,
       "step": 2928
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015006183017312447,
-      "loss": 1.8584,
+      "learning_rate": 0.0005,
+      "loss": 5.1253,
       "step": 2929
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00015,
-      "loss": 1.9183,
+      "learning_rate": 0.0004997938994229184,
+      "loss": 5.1202,
       "step": 2930
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0001499381698268755,
-      "loss": 1.5168,
+      "learning_rate": 0.0004995877988458368,
+      "loss": 4.5824,
       "step": 2931
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00014987633965375102,
-      "loss": 2.1877,
+      "learning_rate": 0.0004993816982687552,
+      "loss": 4.8614,
       "step": 2932
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00014981450948062654,
-      "loss": 2.6068,
+      "learning_rate": 0.0004991755976916735,
+      "loss": 5.9299,
       "step": 2933
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.00014975267930750205,
-      "loss": 0.9723,
+      "learning_rate": 0.0004989694971145919,
+      "loss": 3.9866,
       "step": 2934
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014969084913437757,
-      "loss": 2.5463,
+      "learning_rate": 0.0004987633965375103,
+      "loss": 4.7628,
       "step": 2935
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014962901896125306,
-      "loss": 3.4756,
+      "learning_rate": 0.0004985572959604287,
+      "loss": 5.5465,
       "step": 2936
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0001495671887881286,
-      "loss": 2.9632,
+      "learning_rate": 0.0004983511953833472,
+      "loss": 4.982,
       "step": 2937
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014950535861500412,
-      "loss": 2.0962,
+      "learning_rate": 0.0004981450948062655,
+      "loss": 4.689,
       "step": 2938
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014944352844187964,
-      "loss": 2.0324,
+      "learning_rate": 0.0004979389942291839,
+      "loss": 4.8461,
       "step": 2939
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014938169826875515,
-      "loss": 2.3003,
+      "learning_rate": 0.0004977328936521023,
+      "loss": 4.9766,
       "step": 2940
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014931986809563064,
-      "loss": 2.027,
+      "learning_rate": 0.0004975267930750207,
+      "loss": 4.7843,
       "step": 2941
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014925803792250619,
-      "loss": 1.7788,
+      "learning_rate": 0.000497320692497939,
+      "loss": 4.7863,
       "step": 2942
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014919620774938168,
-      "loss": 1.9867,
+      "learning_rate": 0.0004971145919208574,
+      "loss": 4.9879,
       "step": 2943
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0001491343775762572,
-      "loss": 2.1919,
+      "learning_rate": 0.0004969084913437758,
+      "loss": 5.1038,
       "step": 2944
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0001490725474031327,
-      "loss": 2.7378,
+      "learning_rate": 0.0004967023907666942,
+      "loss": 5.4049,
       "step": 2945
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014901071723000823,
-      "loss": 1.8344,
+      "learning_rate": 0.0004964962901896126,
+      "loss": 4.7133,
       "step": 2946
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014894888705688374,
-      "loss": 1.8651,
+      "learning_rate": 0.0004962901896125309,
+      "loss": 4.8305,
       "step": 2947
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014888705688375926,
-      "loss": 1.818,
+      "learning_rate": 0.0004960840890354493,
+      "loss": 4.7094,
       "step": 2948
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014882522671063477,
-      "loss": 1.6207,
+      "learning_rate": 0.0004958779884583677,
+      "loss": 4.923,
       "step": 2949
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0001487633965375103,
-      "loss": 1.5668,
+      "learning_rate": 0.0004956718878812861,
+      "loss": 4.5092,
       "step": 2950
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0001487015663643858,
-      "loss": 1.7136,
+      "learning_rate": 0.0004954657873042044,
+      "loss": 4.8073,
       "step": 2951
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.00014863973619126132,
-      "loss": 1.5424,
+      "learning_rate": 0.0004952596867271228,
+      "loss": 4.6826,
       "step": 2952
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014857790601813684,
-      "loss": 1.4458,
+      "learning_rate": 0.0004950535861500412,
+      "loss": 4.526,
       "step": 2953
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014851607584501236,
-      "loss": 2.0629,
+      "learning_rate": 0.0004948474855729596,
+      "loss": 5.1821,
       "step": 2954
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014845424567188787,
-      "loss": 1.6442,
+      "learning_rate": 0.000494641384995878,
+      "loss": 4.5528,
       "step": 2955
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0001483924154987634,
-      "loss": 2.5795,
+      "learning_rate": 0.0004944352844187963,
+      "loss": 5.471,
       "step": 2956
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0001483305853256389,
-      "loss": 1.4049,
+      "learning_rate": 0.0004942291838417147,
+      "loss": 4.4559,
       "step": 2957
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014826875515251442,
-      "loss": 2.0617,
+      "learning_rate": 0.0004940230832646332,
+      "loss": 4.9998,
       "step": 2958
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014820692497938991,
-      "loss": 1.7652,
+      "learning_rate": 0.0004938169826875516,
+      "loss": 5.0525,
       "step": 2959
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014814509480626546,
-      "loss": 1.7015,
+      "learning_rate": 0.0004936108821104699,
+      "loss": 4.7632,
       "step": 2960
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014808326463314095,
-      "loss": 1.6096,
+      "learning_rate": 0.0004934047815333883,
+      "loss": 4.5558,
       "step": 2961
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014802143446001646,
-      "loss": 1.3454,
+      "learning_rate": 0.0004931986809563067,
+      "loss": 4.5775,
       "step": 2962
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.000147959604286892,
-      "loss": 2.5353,
+      "learning_rate": 0.0004929925803792251,
+      "loss": 5.1002,
       "step": 2963
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0001478977741137675,
-      "loss": 2.1222,
+      "learning_rate": 0.0004927864798021435,
+      "loss": 5.4826,
       "step": 2964
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.000147835943940643,
-      "loss": 1.6278,
+      "learning_rate": 0.0004925803792250618,
+      "loss": 4.6726,
       "step": 2965
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014777411376751853,
-      "loss": 1.4907,
+      "learning_rate": 0.0004923742786479802,
+      "loss": 4.5173,
       "step": 2966
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014771228359439405,
-      "loss": 1.6213,
+      "learning_rate": 0.0004921681780708986,
+      "loss": 4.703,
       "step": 2967
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014765045342126956,
-      "loss": 1.5394,
+      "learning_rate": 0.000491962077493817,
+      "loss": 4.6806,
       "step": 2968
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.00014758862324814508,
-      "loss": 1.7878,
+      "learning_rate": 0.0004917559769167354,
+      "loss": 4.8013,
       "step": 2969
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0001475267930750206,
-      "loss": 2.1884,
+      "learning_rate": 0.0004915498763396537,
+      "loss": 4.8185,
       "step": 2970
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0001474649629018961,
-      "loss": 2.1806,
+      "learning_rate": 0.0004913437757625721,
+      "loss": 4.9222,
       "step": 2971
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014740313272877163,
-      "loss": 1.6832,
+      "learning_rate": 0.0004911376751854905,
+      "loss": 5.0195,
       "step": 2972
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014734130255564715,
-      "loss": 1.7217,
+      "learning_rate": 0.0004909315746084089,
+      "loss": 5.2666,
       "step": 2973
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014727947238252266,
-      "loss": 1.8439,
+      "learning_rate": 0.0004907254740313272,
+      "loss": 5.074,
       "step": 2974
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014721764220939818,
-      "loss": 1.6001,
+      "learning_rate": 0.0004905193734542456,
+      "loss": 4.854,
       "step": 2975
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0001471558120362737,
-      "loss": 1.6166,
+      "learning_rate": 0.000490313272877164,
+      "loss": 5.1185,
       "step": 2976
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014709398186314919,
-      "loss": 1.8359,
+      "learning_rate": 0.0004901071723000824,
+      "loss": 4.8451,
       "step": 2977
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014703215169002473,
-      "loss": 1.9809,
+      "learning_rate": 0.0004899010717230009,
+      "loss": 5.6358,
       "step": 2978
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014697032151690025,
-      "loss": 1.8006,
+      "learning_rate": 0.0004896949711459193,
+      "loss": 4.8059,
       "step": 2979
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014690849134377573,
-      "loss": 1.6947,
+      "learning_rate": 0.0004894888705688377,
+      "loss": 4.728,
       "step": 2980
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014684666117065128,
-      "loss": 1.6167,
+      "learning_rate": 0.000489282769991756,
+      "loss": 4.6566,
       "step": 2981
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014678483099752677,
-      "loss": 1.8271,
+      "learning_rate": 0.0004890766694146744,
+      "loss": 4.991,
       "step": 2982
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014672300082440228,
-      "loss": 1.7485,
+      "learning_rate": 0.0004888705688375928,
+      "loss": 5.1569,
       "step": 2983
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0001466611706512778,
-      "loss": 2.2325,
+      "learning_rate": 0.0004886644682605112,
+      "loss": 5.0346,
       "step": 2984
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014659934047815332,
-      "loss": 3.0652,
+      "learning_rate": 0.0004884583676834296,
+      "loss": 5.0383,
       "step": 2985
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014653751030502886,
-      "loss": 2.9615,
+      "learning_rate": 0.0004882522671063479,
+      "loss": 5.0121,
       "step": 2986
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014647568013190435,
-      "loss": 2.6384,
+      "learning_rate": 0.0004880461665292663,
+      "loss": 4.7907,
       "step": 2987
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00014641384995877987,
-      "loss": 2.1878,
+      "learning_rate": 0.00048784006595218465,
+      "loss": 4.6709,
       "step": 2988
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014635201978565538,
-      "loss": 2.4878,
+      "learning_rate": 0.00048763396537510306,
+      "loss": 5.4123,
       "step": 2989
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0001462901896125309,
-      "loss": 1.7273,
+      "learning_rate": 0.0004874278647980214,
+      "loss": 5.0189,
       "step": 2990
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014622835943940642,
-      "loss": 1.9793,
+      "learning_rate": 0.0004872217642209398,
+      "loss": 5.2049,
       "step": 2991
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014616652926628193,
-      "loss": 1.5036,
+      "learning_rate": 0.00048701566364385817,
+      "loss": 4.7734,
       "step": 2992
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014610469909315745,
-      "loss": 2.3646,
+      "learning_rate": 0.0004868095630667766,
+      "loss": 5.6477,
       "step": 2993
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014604286892003297,
-      "loss": 1.9314,
+      "learning_rate": 0.000486603462489695,
+      "loss": 4.7488,
       "step": 2994
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014598103874690848,
-      "loss": 1.8398,
+      "learning_rate": 0.00048639736191261333,
+      "loss": 4.8331,
       "step": 2995
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.000145919208573784,
-      "loss": 1.8153,
+      "learning_rate": 0.00048619126133553174,
+      "loss": 4.7743,
       "step": 2996
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014585737840065952,
-      "loss": 2.2945,
+      "learning_rate": 0.0004859851607584501,
+      "loss": 5.0226,
       "step": 2997
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.000145795548227535,
-      "loss": 1.6662,
+      "learning_rate": 0.0004857790601813685,
+      "loss": 4.6922,
       "step": 2998
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014573371805441055,
-      "loss": 1.6792,
+      "learning_rate": 0.00048557295960428685,
+      "loss": 4.7133,
       "step": 2999
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014567188788128604,
-      "loss": 1.7742,
+      "learning_rate": 0.0004853668590272053,
+      "loss": 4.7653,
       "step": 3000
     },
     {
       "epoch": 1.68,
-      "eval_loss": 2.1782970428466797,
-      "eval_runtime": 1250.1188,
-      "eval_samples_per_second": 2.113,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.2931574771915906,
+      "eval_loss": 5.37747859954834,
+      "eval_runtime": 1246.3228,
+      "eval_samples_per_second": 2.12,
+      "eval_steps_per_second": 0.266,
+      "eval_wer": 1.9548393494644982,
       "step": 3000
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014561005770816158,
-      "loss": 1.6246,
+      "learning_rate": 0.0004851607584501237,
+      "loss": 4.7718,
       "step": 3001
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0001455482275350371,
-      "loss": 2.4096,
+      "learning_rate": 0.00048495465787304207,
+      "loss": 4.922,
       "step": 3002
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0001454863973619126,
-      "loss": 1.4294,
+      "learning_rate": 0.0004847485572959605,
+      "loss": 4.7143,
       "step": 3003
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014542456718878813,
-      "loss": 1.7278,
+      "learning_rate": 0.00048454245671887883,
+      "loss": 4.6927,
       "step": 3004
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014536273701566362,
-      "loss": 1.5225,
+      "learning_rate": 0.00048433635614179724,
+      "loss": 4.6121,
       "step": 3005
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00014530090684253914,
-      "loss": 1.925,
+      "learning_rate": 0.0004841302555647156,
+      "loss": 5.0751,
       "step": 3006
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014523907666941466,
-      "loss": 2.2603,
+      "learning_rate": 0.000483924154987634,
+      "loss": 5.1179,
       "step": 3007
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014517724649629017,
-      "loss": 1.4939,
+      "learning_rate": 0.00048371805441055235,
+      "loss": 4.7184,
       "step": 3008
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0001451154163231657,
-      "loss": 1.4398,
+      "learning_rate": 0.00048351195383347076,
+      "loss": 4.6881,
       "step": 3009
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0001450535861500412,
-      "loss": 1.6482,
+      "learning_rate": 0.00048330585325638916,
+      "loss": 5.1034,
       "step": 3010
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014499175597691672,
-      "loss": 1.6908,
+      "learning_rate": 0.0004830997526793075,
+      "loss": 4.9019,
       "step": 3011
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014492992580379224,
-      "loss": 1.3887,
+      "learning_rate": 0.0004828936521022259,
+      "loss": 4.4907,
       "step": 3012
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014486809563066776,
-      "loss": 1.9958,
+      "learning_rate": 0.0004826875515251443,
+      "loss": 5.4752,
       "step": 3013
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014480626545754327,
-      "loss": 1.6333,
+      "learning_rate": 0.0004824814509480627,
+      "loss": 4.9775,
       "step": 3014
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0001447444352844188,
-      "loss": 1.3493,
+      "learning_rate": 0.00048227535037098103,
+      "loss": 4.58,
       "step": 3015
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0001446826051112943,
-      "loss": 1.7521,
+      "learning_rate": 0.00048206924979389944,
+      "loss": 4.8092,
       "step": 3016
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014462077493816982,
-      "loss": 3.4464,
+      "learning_rate": 0.00048186314921681785,
+      "loss": 6.0983,
       "step": 3017
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014455894476504534,
-      "loss": 1.4186,
+      "learning_rate": 0.0004816570486397362,
+      "loss": 4.6628,
       "step": 3018
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014449711459192085,
-      "loss": 1.683,
+      "learning_rate": 0.0004814509480626546,
+      "loss": 4.9202,
       "step": 3019
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014443528441879637,
-      "loss": 1.4076,
+      "learning_rate": 0.00048124484748557296,
+      "loss": 4.6934,
       "step": 3020
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00014437345424567186,
-      "loss": 1.4857,
+      "learning_rate": 0.00048103874690849137,
+      "loss": 4.7527,
       "step": 3021
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0001443116240725474,
-      "loss": 1.2337,
+      "learning_rate": 0.0004808326463314097,
+      "loss": 4.7332,
       "step": 3022
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0001442497938994229,
-      "loss": 1.8137,
+      "learning_rate": 0.0004806265457543281,
+      "loss": 5.1831,
       "step": 3023
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0001441879637262984,
-      "loss": 1.3033,
+      "learning_rate": 0.0004804204451772465,
+      "loss": 4.6306,
       "step": 3024
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014412613355317393,
-      "loss": 1.6029,
+      "learning_rate": 0.0004802143446001649,
+      "loss": 4.9321,
       "step": 3025
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014406430338004944,
-      "loss": 1.6049,
+      "learning_rate": 0.0004800082440230833,
+      "loss": 4.7831,
       "step": 3026
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014400247320692496,
-      "loss": 1.4596,
+      "learning_rate": 0.00047980214344600164,
+      "loss": 4.6309,
       "step": 3027
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014394064303380048,
-      "loss": 1.9565,
+      "learning_rate": 0.00047959604286892005,
+      "loss": 5.9696,
       "step": 3028
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.000143878812860676,
-      "loss": 1.2809,
+      "learning_rate": 0.0004793899422918384,
+      "loss": 4.7136,
       "step": 3029
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0001438169826875515,
-      "loss": 3.2334,
+      "learning_rate": 0.0004791838417147568,
+      "loss": 5.6978,
       "step": 3030
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014375515251442703,
-      "loss": 1.4871,
+      "learning_rate": 0.00047897774113767516,
+      "loss": 4.7382,
       "step": 3031
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014369332234130254,
-      "loss": 1.6086,
+      "learning_rate": 0.00047877164056059357,
+      "loss": 4.8475,
       "step": 3032
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014363149216817806,
-      "loss": 1.868,
+      "learning_rate": 0.000478565539983512,
+      "loss": 4.7631,
       "step": 3033
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014356966199505358,
-      "loss": 1.7101,
+      "learning_rate": 0.00047835943940643033,
+      "loss": 4.6395,
       "step": 3034
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0001435078318219291,
-      "loss": 2.3239,
+      "learning_rate": 0.00047815333882934873,
+      "loss": 4.832,
       "step": 3035
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0001434460016488046,
-      "loss": 1.9358,
+      "learning_rate": 0.0004779472382522671,
+      "loss": 4.709,
       "step": 3036
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014338417147568013,
-      "loss": 1.9589,
+      "learning_rate": 0.0004777411376751855,
+      "loss": 4.8878,
       "step": 3037
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014332234130255564,
-      "loss": 1.9086,
+      "learning_rate": 0.00047753503709810385,
+      "loss": 4.8311,
       "step": 3038
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014326051112943113,
-      "loss": 1.8231,
+      "learning_rate": 0.00047732893652102225,
+      "loss": 4.8308,
       "step": 3039
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00014319868095630668,
-      "loss": 1.794,
+      "learning_rate": 0.00047712283594394066,
+      "loss": 5.2709,
       "step": 3040
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0001431368507831822,
-      "loss": 1.6722,
+      "learning_rate": 0.000476916735366859,
+      "loss": 4.8662,
       "step": 3041
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014307502061005768,
-      "loss": 1.4658,
+      "learning_rate": 0.0004767106347897774,
+      "loss": 4.7272,
       "step": 3042
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014301319043693323,
-      "loss": 1.692,
+      "learning_rate": 0.00047650453421269577,
+      "loss": 4.7876,
       "step": 3043
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014295136026380872,
-      "loss": 1.6331,
+      "learning_rate": 0.0004762984336356142,
+      "loss": 4.6993,
       "step": 3044
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014288953009068426,
-      "loss": 1.9097,
+      "learning_rate": 0.00047609233305853253,
+      "loss": 4.7552,
       "step": 3045
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014282769991755975,
-      "loss": 1.3912,
+      "learning_rate": 0.000475886232481451,
+      "loss": 4.7379,
       "step": 3046
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014276586974443526,
-      "loss": 1.5813,
+      "learning_rate": 0.00047568013190436934,
+      "loss": 4.6416,
       "step": 3047
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014270403957131078,
-      "loss": 1.8216,
+      "learning_rate": 0.00047547403132728775,
+      "loss": 5.1788,
       "step": 3048
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0001426422093981863,
-      "loss": 1.652,
+      "learning_rate": 0.00047526793075020616,
+      "loss": 4.6729,
       "step": 3049
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014258037922506181,
-      "loss": 1.3672,
+      "learning_rate": 0.0004750618301731245,
+      "loss": 4.5329,
       "step": 3050
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014251854905193733,
-      "loss": 1.707,
+      "learning_rate": 0.0004748557295960429,
+      "loss": 4.5627,
       "step": 3051
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014245671887881285,
-      "loss": 1.5796,
+      "learning_rate": 0.00047464962901896127,
+      "loss": 4.9644,
       "step": 3052
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014239488870568836,
-      "loss": 1.6575,
+      "learning_rate": 0.0004744435284418797,
+      "loss": 4.8746,
       "step": 3053
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014233305853256388,
-      "loss": 1.5738,
+      "learning_rate": 0.00047423742786479803,
+      "loss": 4.7598,
       "step": 3054
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0001422712283594394,
-      "loss": 1.7179,
+      "learning_rate": 0.00047403132728771643,
+      "loss": 4.9776,
       "step": 3055
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014220939818631491,
-      "loss": 1.4782,
+      "learning_rate": 0.00047382522671063484,
+      "loss": 4.8499,
       "step": 3056
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014214756801319043,
-      "loss": 1.7505,
+      "learning_rate": 0.0004736191261335532,
+      "loss": 4.9817,
       "step": 3057
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014208573784006595,
-      "loss": 2.1045,
+      "learning_rate": 0.0004734130255564716,
+      "loss": 4.9736,
       "step": 3058
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00014202390766694146,
-      "loss": 1.3892,
+      "learning_rate": 0.00047320692497938995,
+      "loss": 4.6397,
       "step": 3059
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014196207749381698,
-      "loss": 1.4546,
+      "learning_rate": 0.00047300082440230836,
+      "loss": 4.674,
       "step": 3060
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0001419002473206925,
-      "loss": 1.5867,
+      "learning_rate": 0.0004727947238252267,
+      "loss": 4.5434,
       "step": 3061
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.000141838417147568,
-      "loss": 1.7286,
+      "learning_rate": 0.0004725886232481451,
+      "loss": 4.8244,
       "step": 3062
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014177658697444353,
-      "loss": 1.6877,
+      "learning_rate": 0.00047238252267106347,
+      "loss": 5.0157,
       "step": 3063
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014171475680131902,
-      "loss": 1.5488,
+      "learning_rate": 0.0004721764220939819,
+      "loss": 4.9611,
       "step": 3064
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014165292662819454,
-      "loss": 1.3499,
+      "learning_rate": 0.0004719703215169003,
+      "loss": 4.4803,
       "step": 3065
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014159109645507008,
-      "loss": 1.5511,
+      "learning_rate": 0.00047176422093981864,
+      "loss": 5.0906,
       "step": 3066
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014152926628194557,
-      "loss": 1.3244,
+      "learning_rate": 0.00047155812036273704,
+      "loss": 4.6382,
       "step": 3067
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014146743610882109,
-      "loss": 1.9156,
+      "learning_rate": 0.0004713520197856554,
+      "loss": 5.3363,
       "step": 3068
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0001414056059356966,
-      "loss": 2.2198,
+      "learning_rate": 0.0004711459192085738,
+      "loss": 5.2619,
       "step": 3069
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014134377576257212,
-      "loss": 1.7753,
+      "learning_rate": 0.00047093981863149216,
+      "loss": 4.8895,
       "step": 3070
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014128194558944764,
-      "loss": 1.5019,
+      "learning_rate": 0.00047073371805441056,
+      "loss": 4.8402,
       "step": 3071
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014122011541632315,
-      "loss": 1.5263,
+      "learning_rate": 0.00047052761747732897,
+      "loss": 4.7542,
       "step": 3072
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014115828524319867,
-      "loss": 1.5127,
+      "learning_rate": 0.0004703215169002473,
+      "loss": 4.8944,
       "step": 3073
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014109645507007419,
-      "loss": 1.2696,
+      "learning_rate": 0.00047011541632316573,
+      "loss": 4.8044,
       "step": 3074
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0001410346248969497,
-      "loss": 1.5842,
+      "learning_rate": 0.0004699093157460841,
+      "loss": 5.2358,
       "step": 3075
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014097279472382522,
-      "loss": 3.6353,
+      "learning_rate": 0.0004697032151690025,
+      "loss": 6.0982,
       "step": 3076
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00014091096455070074,
-      "loss": 1.2686,
+      "learning_rate": 0.00046949711459192084,
+      "loss": 4.9189,
       "step": 3077
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014084913437757625,
-      "loss": 1.3171,
+      "learning_rate": 0.00046929101401483925,
+      "loss": 5.1576,
       "step": 3078
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014078730420445177,
-      "loss": 1.9729,
+      "learning_rate": 0.0004690849134377576,
+      "loss": 5.5019,
       "step": 3079
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014072547403132726,
-      "loss": 2.0167,
+      "learning_rate": 0.000468878812860676,
+      "loss": 5.232,
       "step": 3080
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0001406636438582028,
-      "loss": 1.1428,
+      "learning_rate": 0.0004686727122835944,
+      "loss": 4.5148,
       "step": 3081
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014060181368507832,
-      "loss": 1.6653,
+      "learning_rate": 0.00046846661170651276,
+      "loss": 5.0405,
       "step": 3082
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0001405399835119538,
-      "loss": 1.7054,
+      "learning_rate": 0.00046826051112943117,
+      "loss": 4.7981,
       "step": 3083
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014047815333882935,
-      "loss": 3.1613,
+      "learning_rate": 0.0004680544105523495,
+      "loss": 8.6501,
       "step": 3084
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014041632316570484,
-      "loss": 3.1964,
+      "learning_rate": 0.00046784830997526793,
+      "loss": 5.0449,
       "step": 3085
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014035449299258036,
-      "loss": 2.5686,
+      "learning_rate": 0.0004676422093981863,
+      "loss": 4.9747,
       "step": 3086
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014029266281945587,
-      "loss": 2.3223,
+      "learning_rate": 0.0004674361088211047,
+      "loss": 4.8174,
       "step": 3087
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0001402308326463314,
-      "loss": 2.3609,
+      "learning_rate": 0.0004672300082440231,
+      "loss": 4.945,
       "step": 3088
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0001401690024732069,
-      "loss": 2.0663,
+      "learning_rate": 0.00046702390766694145,
+      "loss": 4.9772,
       "step": 3089
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014010717230008242,
-      "loss": 1.6501,
+      "learning_rate": 0.00046681780708985986,
+      "loss": 4.5791,
       "step": 3090
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00014004534212695794,
-      "loss": 1.6523,
+      "learning_rate": 0.0004666117065127782,
+      "loss": 4.9822,
       "step": 3091
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00013998351195383346,
-      "loss": 1.4749,
+      "learning_rate": 0.0004664056059356966,
+      "loss": 5.1276,
       "step": 3092
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00013992168178070897,
-      "loss": 1.2648,
+      "learning_rate": 0.000466199505358615,
+      "loss": 4.6729,
       "step": 3093
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0001398598516075845,
-      "loss": 1.8888,
+      "learning_rate": 0.00046599340478153343,
+      "loss": 5.0321,
       "step": 3094
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00013979802143446,
-      "loss": 1.641,
+      "learning_rate": 0.0004657873042044518,
+      "loss": 5.0349,
       "step": 3095
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013973619126133552,
-      "loss": 2.0332,
+      "learning_rate": 0.0004655812036273702,
+      "loss": 4.955,
       "step": 3096
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013967436108821104,
-      "loss": 1.7942,
+      "learning_rate": 0.0004653751030502886,
+      "loss": 4.9472,
       "step": 3097
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013961253091508656,
-      "loss": 1.6728,
+      "learning_rate": 0.00046516900247320695,
+      "loss": 4.6128,
       "step": 3098
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013955070074196207,
-      "loss": 1.3839,
+      "learning_rate": 0.00046496290189612535,
+      "loss": 4.6475,
       "step": 3099
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0001394888705688376,
-      "loss": 1.8967,
+      "learning_rate": 0.0004647568013190437,
+      "loss": 4.9007,
       "step": 3100
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013942704039571308,
-      "loss": 1.7344,
+      "learning_rate": 0.0004645507007419621,
+      "loss": 5.0667,
       "step": 3101
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013936521022258862,
-      "loss": 1.5467,
+      "learning_rate": 0.00046434460016488046,
+      "loss": 4.9717,
       "step": 3102
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0001393033800494641,
-      "loss": 1.8048,
+      "learning_rate": 0.00046413849958779887,
+      "loss": 5.3151,
       "step": 3103
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013924154987633963,
-      "loss": 1.5366,
+      "learning_rate": 0.0004639323990107173,
+      "loss": 4.615,
       "step": 3104
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013917971970321517,
-      "loss": 2.0292,
+      "learning_rate": 0.00046372629843363563,
+      "loss": 5.5692,
       "step": 3105
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013911788953009066,
-      "loss": 1.7513,
+      "learning_rate": 0.00046352019785655404,
+      "loss": 5.0304,
       "step": 3106
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0001390560593569662,
-      "loss": 1.4473,
+      "learning_rate": 0.0004633140972794724,
+      "loss": 4.8689,
       "step": 3107
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0001389942291838417,
-      "loss": 1.6301,
+      "learning_rate": 0.0004631079967023908,
+      "loss": 4.6382,
       "step": 3108
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0001389323990107172,
-      "loss": 1.5719,
+      "learning_rate": 0.00046290189612530915,
+      "loss": 4.8791,
       "step": 3109
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013887056883759273,
-      "loss": 1.8869,
+      "learning_rate": 0.00046269579554822756,
+      "loss": 5.1656,
       "step": 3110
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013880873866446825,
-      "loss": 1.4468,
+      "learning_rate": 0.0004624896949711459,
+      "loss": 4.749,
       "step": 3111
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013874690849134376,
-      "loss": 1.9468,
+      "learning_rate": 0.0004622835943940643,
+      "loss": 5.302,
       "step": 3112
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00013868507831821928,
-      "loss": 1.4547,
+      "learning_rate": 0.0004620774938169827,
+      "loss": 4.8968,
       "step": 3113
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0001386232481450948,
-      "loss": 1.5174,
+      "learning_rate": 0.0004618713932399011,
+      "loss": 4.7596,
       "step": 3114
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0001385614179719703,
-      "loss": 1.7212,
+      "learning_rate": 0.0004616652926628195,
+      "loss": 4.8576,
       "step": 3115
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013849958779884583,
-      "loss": 1.407,
+      "learning_rate": 0.00046145919208573783,
+      "loss": 4.6059,
       "step": 3116
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013843775762572134,
-      "loss": 1.2841,
+      "learning_rate": 0.00046125309150865624,
+      "loss": 4.6033,
       "step": 3117
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013837592745259686,
-      "loss": 1.383,
+      "learning_rate": 0.0004610469909315746,
+      "loss": 4.6653,
       "step": 3118
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013831409727947235,
-      "loss": 1.2556,
+      "learning_rate": 0.000460840890354493,
+      "loss": 4.7146,
       "step": 3119
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0001382522671063479,
-      "loss": 1.4223,
+      "learning_rate": 0.0004606347897774114,
+      "loss": 4.7012,
       "step": 3120
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0001381904369332234,
-      "loss": 1.5587,
+      "learning_rate": 0.00046042868920032976,
+      "loss": 5.0652,
       "step": 3121
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013812860676009893,
-      "loss": 1.2685,
+      "learning_rate": 0.00046022258862324817,
+      "loss": 4.7243,
       "step": 3122
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013806677658697444,
-      "loss": 1.3272,
+      "learning_rate": 0.0004600164880461665,
+      "loss": 4.7944,
       "step": 3123
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013800494641384993,
-      "loss": 1.4054,
+      "learning_rate": 0.0004598103874690849,
+      "loss": 4.9057,
       "step": 3124
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013794311624072548,
-      "loss": 1.2659,
+      "learning_rate": 0.0004596042868920033,
+      "loss": 4.7848,
       "step": 3125
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013788128606760097,
-      "loss": 1.3789,
+      "learning_rate": 0.0004593981863149217,
+      "loss": 4.7799,
       "step": 3126
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013781945589447648,
-      "loss": 1.4552,
+      "learning_rate": 0.00045919208573784004,
+      "loss": 4.6397,
       "step": 3127
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.000137757625721352,
-      "loss": 0.9915,
+      "learning_rate": 0.00045898598516075844,
+      "loss": 4.5853,
       "step": 3128
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013769579554822752,
-      "loss": 1.3683,
+      "learning_rate": 0.00045877988458367685,
+      "loss": 4.6781,
       "step": 3129
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00013763396537510303,
-      "loss": 1.2754,
+      "learning_rate": 0.0004585737840065952,
+      "loss": 4.5539,
       "step": 3130
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013757213520197855,
-      "loss": 1.2967,
+      "learning_rate": 0.0004583676834295136,
+      "loss": 4.9331,
       "step": 3131
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013751030502885407,
-      "loss": 2.7258,
+      "learning_rate": 0.00045816158285243196,
+      "loss": 5.8975,
       "step": 3132
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013744847485572958,
-      "loss": 1.1477,
+      "learning_rate": 0.00045795548227535037,
+      "loss": 4.6423,
       "step": 3133
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0001373866446826051,
-      "loss": 1.2742,
+      "learning_rate": 0.0004577493816982687,
+      "loss": 4.3845,
       "step": 3134
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013732481450948062,
-      "loss": 2.335,
+      "learning_rate": 0.0004575432811211871,
+      "loss": 5.4291,
       "step": 3135
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013726298433635613,
-      "loss": 2.1805,
+      "learning_rate": 0.00045733718054410553,
+      "loss": 5.2008,
       "step": 3136
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013720115416323165,
-      "loss": 1.7483,
+      "learning_rate": 0.0004571310799670239,
+      "loss": 5.0435,
       "step": 3137
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013713932399010717,
-      "loss": 1.5639,
+      "learning_rate": 0.0004569249793899423,
+      "loss": 4.9647,
       "step": 3138
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013707749381698268,
-      "loss": 1.3947,
+      "learning_rate": 0.00045671887881286065,
+      "loss": 4.8151,
       "step": 3139
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0001370156636438582,
-      "loss": 1.5721,
+      "learning_rate": 0.0004565127782357791,
+      "loss": 4.8969,
       "step": 3140
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013695383347073372,
-      "loss": 1.8825,
+      "learning_rate": 0.00045630667765869746,
+      "loss": 5.0369,
       "step": 3141
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0001368920032976092,
-      "loss": 2.1878,
+      "learning_rate": 0.00045610057708161587,
+      "loss": 5.3135,
       "step": 3142
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013683017312448475,
-      "loss": 1.6823,
+      "learning_rate": 0.0004558944765045342,
+      "loss": 4.9621,
       "step": 3143
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013676834295136024,
-      "loss": 1.7178,
+      "learning_rate": 0.0004556883759274526,
+      "loss": 4.8998,
       "step": 3144
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00013670651277823575,
-      "loss": 1.282,
+      "learning_rate": 0.00045548227535037103,
+      "loss": 4.5532,
       "step": 3145
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0001366446826051113,
-      "loss": 1.2401,
+      "learning_rate": 0.0004552761747732894,
+      "loss": 4.5489,
       "step": 3146
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0001365828524319868,
-      "loss": 1.5828,
+      "learning_rate": 0.0004550700741962078,
+      "loss": 4.7167,
       "step": 3147
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0001365210222588623,
-      "loss": 1.4762,
+      "learning_rate": 0.00045486397361912614,
+      "loss": 4.7675,
       "step": 3148
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013645919208573782,
-      "loss": 1.3242,
+      "learning_rate": 0.00045465787304204455,
+      "loss": 4.7578,
       "step": 3149
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013639736191261334,
-      "loss": 1.6829,
+      "learning_rate": 0.0004544517724649629,
+      "loss": 4.8245,
       "step": 3150
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013633553173948885,
-      "loss": 1.4577,
+      "learning_rate": 0.0004542456718878813,
+      "loss": 4.7631,
       "step": 3151
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013627370156636437,
-      "loss": 1.865,
+      "learning_rate": 0.0004540395713107997,
+      "loss": 5.2156,
       "step": 3152
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0001362118713932399,
-      "loss": 1.453,
+      "learning_rate": 0.00045383347073371807,
+      "loss": 4.8006,
       "step": 3153
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0001361500412201154,
-      "loss": 2.0336,
+      "learning_rate": 0.0004536273701566365,
+      "loss": 5.3038,
       "step": 3154
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013608821104699092,
-      "loss": 1.454,
+      "learning_rate": 0.0004534212695795548,
+      "loss": 4.785,
       "step": 3155
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013602638087386644,
-      "loss": 1.0897,
+      "learning_rate": 0.00045321516900247323,
+      "loss": 4.4697,
       "step": 3156
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013596455070074195,
-      "loss": 1.6493,
+      "learning_rate": 0.0004530090684253916,
+      "loss": 4.9542,
       "step": 3157
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013590272052761747,
-      "loss": 1.1906,
+      "learning_rate": 0.00045280296784831,
+      "loss": 4.7452,
       "step": 3158
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.000135840890354493,
-      "loss": 1.1785,
+      "learning_rate": 0.0004525968672712284,
+      "loss": 4.7631,
       "step": 3159
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0001357790601813685,
-      "loss": 1.5536,
+      "learning_rate": 0.00045239076669414675,
+      "loss": 4.9631,
       "step": 3160
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013571723000824402,
-      "loss": 1.3546,
+      "learning_rate": 0.00045218466611706516,
+      "loss": 4.7429,
       "step": 3161
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013565539983511954,
-      "loss": 1.4581,
+      "learning_rate": 0.0004519785655399835,
+      "loss": 4.557,
       "step": 3162
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013559356966199503,
-      "loss": 1.1637,
+      "learning_rate": 0.0004517724649629019,
+      "loss": 4.8137,
       "step": 3163
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013553173948887057,
-      "loss": 1.9406,
+      "learning_rate": 0.00045156636438582027,
+      "loss": 5.2951,
       "step": 3164
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013546990931574606,
-      "loss": 1.6998,
+      "learning_rate": 0.0004513602638087387,
+      "loss": 5.1023,
       "step": 3165
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00013540807914262158,
-      "loss": 1.2904,
+      "learning_rate": 0.00045115416323165703,
+      "loss": 4.5265,
       "step": 3166
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0001353462489694971,
-      "loss": 1.8389,
+      "learning_rate": 0.00045094806265457544,
+      "loss": 4.8865,
       "step": 3167
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0001352844187963726,
-      "loss": 1.3262,
+      "learning_rate": 0.00045074196207749384,
+      "loss": 4.7221,
       "step": 3168
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013522258862324815,
-      "loss": 1.1118,
+      "learning_rate": 0.0004505358615004122,
+      "loss": 4.6497,
       "step": 3169
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013516075845012364,
-      "loss": 1.2192,
+      "learning_rate": 0.0004503297609233306,
+      "loss": 4.6578,
       "step": 3170
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013509892827699916,
-      "loss": 1.5904,
+      "learning_rate": 0.00045012366034624895,
+      "loss": 4.9729,
       "step": 3171
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013503709810387468,
-      "loss": 1.1388,
+      "learning_rate": 0.00044991755976916736,
+      "loss": 4.5774,
       "step": 3172
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0001349752679307502,
-      "loss": 1.4391,
+      "learning_rate": 0.0004497114591920857,
+      "loss": 4.8821,
       "step": 3173
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0001349134377576257,
-      "loss": 1.2622,
+      "learning_rate": 0.0004495053586150041,
+      "loss": 4.6988,
       "step": 3174
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013485160758450123,
-      "loss": 1.1688,
+      "learning_rate": 0.0004492992580379225,
+      "loss": 4.8031,
       "step": 3175
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013478977741137674,
-      "loss": 1.6527,
+      "learning_rate": 0.0004490931574608409,
+      "loss": 4.9744,
       "step": 3176
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013472794723825226,
-      "loss": 1.3917,
+      "learning_rate": 0.0004488870568837593,
+      "loss": 5.0486,
       "step": 3177
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013466611706512778,
-      "loss": 0.9887,
+      "learning_rate": 0.00044868095630667764,
+      "loss": 4.7017,
       "step": 3178
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0001346042868920033,
-      "loss": 1.435,
+      "learning_rate": 0.00044847485572959605,
+      "loss": 4.7223,
       "step": 3179
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0001345424567188788,
-      "loss": 1.434,
+      "learning_rate": 0.0004482687551525144,
+      "loss": 5.1026,
       "step": 3180
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0001344806265457543,
-      "loss": 2.8895,
+      "learning_rate": 0.0004480626545754328,
+      "loss": 5.2464,
       "step": 3181
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013441879637262984,
-      "loss": 0.7265,
+      "learning_rate": 0.00044785655399835116,
+      "loss": 4.194,
       "step": 3182
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013435696619950533,
-      "loss": 2.1806,
+      "learning_rate": 0.00044765045342126956,
+      "loss": 4.8353,
       "step": 3183
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00013429513602638087,
-      "loss": 1.5174,
+      "learning_rate": 0.00044744435284418797,
+      "loss": 4.6733,
       "step": 3184
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0001342333058532564,
-      "loss": 2.0754,
+      "learning_rate": 0.0004472382522671063,
+      "loss": 4.8886,
       "step": 3185
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013417147568013188,
-      "loss": 2.2174,
+      "learning_rate": 0.0004470321516900248,
+      "loss": 4.8368,
       "step": 3186
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013410964550700742,
-      "loss": 1.9918,
+      "learning_rate": 0.00044682605111294314,
+      "loss": 4.8884,
       "step": 3187
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013404781533388291,
-      "loss": 1.6941,
+      "learning_rate": 0.00044661995053586154,
+      "loss": 5.0182,
       "step": 3188
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013398598516075843,
-      "loss": 1.7642,
+      "learning_rate": 0.0004464138499587799,
+      "loss": 5.0725,
       "step": 3189
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013392415498763395,
-      "loss": 1.5937,
+      "learning_rate": 0.0004462077493816983,
+      "loss": 5.0742,
       "step": 3190
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013386232481450946,
-      "loss": 1.4389,
+      "learning_rate": 0.0004460016488046167,
+      "loss": 4.7844,
       "step": 3191
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013380049464138498,
-      "loss": 1.8774,
+      "learning_rate": 0.00044579554822753506,
+      "loss": 5.1153,
       "step": 3192
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0001337386644682605,
-      "loss": 1.4407,
+      "learning_rate": 0.00044558944765045347,
+      "loss": 4.9863,
       "step": 3193
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.000133676834295136,
-      "loss": 1.711,
+      "learning_rate": 0.0004453833470733718,
+      "loss": 4.9356,
       "step": 3194
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013361500412201153,
-      "loss": 1.7068,
+      "learning_rate": 0.00044517724649629023,
+      "loss": 4.9145,
       "step": 3195
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013355317394888705,
-      "loss": 2.0738,
+      "learning_rate": 0.0004449711459192086,
+      "loss": 5.1526,
       "step": 3196
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013349134377576256,
-      "loss": 1.4413,
+      "learning_rate": 0.000444765045342127,
+      "loss": 4.6843,
       "step": 3197
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013342951360263808,
-      "loss": 1.6506,
+      "learning_rate": 0.00044455894476504534,
+      "loss": 5.0706,
       "step": 3198
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0001333676834295136,
-      "loss": 1.9682,
+      "learning_rate": 0.00044435284418796375,
+      "loss": 5.4601,
       "step": 3199
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0001333058532563891,
-      "loss": 2.0144,
+      "learning_rate": 0.00044414674361088215,
+      "loss": 5.3174,
       "step": 3200
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013324402308326463,
-      "loss": 1.4265,
+      "learning_rate": 0.0004439406430338005,
+      "loss": 4.7116,
       "step": 3201
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00013318219291014015,
-      "loss": 1.2685,
+      "learning_rate": 0.0004437345424567189,
+      "loss": 4.8322,
       "step": 3202
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013312036273701566,
-      "loss": 1.6608,
+      "learning_rate": 0.00044352844187963726,
+      "loss": 4.9297,
       "step": 3203
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013305853256389115,
-      "loss": 1.6064,
+      "learning_rate": 0.00044332234130255567,
+      "loss": 4.9091,
       "step": 3204
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0001329967023907667,
-      "loss": 1.525,
+      "learning_rate": 0.000443116240725474,
+      "loss": 5.2406,
       "step": 3205
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013293487221764219,
-      "loss": 1.525,
+      "learning_rate": 0.00044291014014839243,
+      "loss": 4.9367,
       "step": 3206
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0001328730420445177,
-      "loss": 1.4271,
+      "learning_rate": 0.00044270403957131084,
+      "loss": 4.8079,
       "step": 3207
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013281121187139325,
-      "loss": 1.3807,
+      "learning_rate": 0.0004424979389942292,
+      "loss": 4.8947,
       "step": 3208
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013274938169826873,
-      "loss": 1.4944,
+      "learning_rate": 0.0004422918384171476,
+      "loss": 4.8248,
       "step": 3209
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013268755152514425,
-      "loss": 2.3034,
+      "learning_rate": 0.00044208573784006595,
+      "loss": 5.2378,
       "step": 3210
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013262572135201977,
-      "loss": 1.4234,
+      "learning_rate": 0.00044187963726298435,
+      "loss": 4.9258,
       "step": 3211
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013256389117889528,
-      "loss": 1.1132,
+      "learning_rate": 0.0004416735366859027,
+      "loss": 4.6727,
       "step": 3212
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0001325020610057708,
-      "loss": 1.6649,
+      "learning_rate": 0.0004414674361088211,
+      "loss": 4.7168,
       "step": 3213
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013244023083264632,
-      "loss": 1.9015,
+      "learning_rate": 0.00044126133553173947,
+      "loss": 5.2165,
       "step": 3214
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013237840065952183,
-      "loss": 1.7327,
+      "learning_rate": 0.0004410552349546579,
+      "loss": 4.8579,
       "step": 3215
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013231657048639735,
-      "loss": 1.6875,
+      "learning_rate": 0.0004408491343775763,
+      "loss": 4.8613,
       "step": 3216
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013225474031327287,
-      "loss": 1.3126,
+      "learning_rate": 0.00044064303380049463,
+      "loss": 4.6757,
       "step": 3217
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013219291014014838,
-      "loss": 1.2485,
+      "learning_rate": 0.00044043693322341304,
+      "loss": 4.6173,
       "step": 3218
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0001321310799670239,
-      "loss": 1.3606,
+      "learning_rate": 0.0004402308326463314,
+      "loss": 4.7392,
       "step": 3219
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00013206924979389942,
-      "loss": 1.291,
+      "learning_rate": 0.0004400247320692498,
+      "loss": 4.6896,
       "step": 3220
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013200741962077493,
-      "loss": 1.6499,
+      "learning_rate": 0.00043981863149216815,
+      "loss": 5.211,
       "step": 3221
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013194558944765042,
-      "loss": 1.3838,
+      "learning_rate": 0.00043961253091508656,
+      "loss": 4.8211,
       "step": 3222
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013188375927452597,
-      "loss": 1.1887,
+      "learning_rate": 0.00043940643033800496,
+      "loss": 4.5102,
       "step": 3223
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013182192910140148,
-      "loss": 1.1953,
+      "learning_rate": 0.0004392003297609233,
+      "loss": 4.8341,
       "step": 3224
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013176009892827697,
-      "loss": 1.1084,
+      "learning_rate": 0.0004389942291838417,
+      "loss": 4.6401,
       "step": 3225
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013169826875515252,
-      "loss": 1.5219,
+      "learning_rate": 0.0004387881286067601,
+      "loss": 5.2141,
       "step": 3226
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.000131636438582028,
-      "loss": 1.7765,
+      "learning_rate": 0.0004385820280296785,
+      "loss": 5.369,
       "step": 3227
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013157460840890355,
-      "loss": 1.2312,
+      "learning_rate": 0.00043837592745259683,
+      "loss": 4.6513,
       "step": 3228
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013151277823577904,
-      "loss": 1.0189,
+      "learning_rate": 0.00043816982687551524,
+      "loss": 4.6688,
       "step": 3229
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013145094806265456,
-      "loss": 1.2815,
+      "learning_rate": 0.0004379637262984336,
+      "loss": 4.4774,
       "step": 3230
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013138911788953007,
-      "loss": 1.0913,
+      "learning_rate": 0.000437757625721352,
+      "loss": 4.4189,
       "step": 3231
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0001313272877164056,
-      "loss": 1.902,
+      "learning_rate": 0.00043755152514427046,
+      "loss": 4.6987,
       "step": 3232
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0001312654575432811,
-      "loss": 1.2164,
+      "learning_rate": 0.0004373454245671888,
+      "loss": 5.2783,
       "step": 3233
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013120362737015662,
-      "loss": 1.5057,
+      "learning_rate": 0.0004371393239901072,
+      "loss": 4.6701,
       "step": 3234
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013114179719703214,
-      "loss": 2.6914,
+      "learning_rate": 0.0004369332234130256,
+      "loss": 5.7257,
       "step": 3235
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013107996702390766,
-      "loss": 1.7072,
+      "learning_rate": 0.000436727122835944,
+      "loss": 5.0093,
       "step": 3236
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00013101813685078317,
-      "loss": 1.7128,
+      "learning_rate": 0.00043652102225886233,
+      "loss": 4.7871,
       "step": 3237
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0001309563066776587,
-      "loss": 1.6177,
+      "learning_rate": 0.00043631492168178074,
+      "loss": 4.7474,
       "step": 3238
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0001308944765045342,
-      "loss": 1.9365,
+      "learning_rate": 0.00043610882110469915,
+      "loss": 5.1011,
       "step": 3239
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013083264633140972,
-      "loss": 1.4471,
+      "learning_rate": 0.0004359027205276175,
+      "loss": 4.7328,
       "step": 3240
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013077081615828524,
-      "loss": 1.3531,
+      "learning_rate": 0.0004356966199505359,
+      "loss": 4.6768,
       "step": 3241
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013070898598516076,
-      "loss": 1.6068,
+      "learning_rate": 0.00043549051937345426,
+      "loss": 5.5402,
       "step": 3242
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013064715581203624,
-      "loss": 1.815,
+      "learning_rate": 0.00043528441879637266,
+      "loss": 5.0311,
       "step": 3243
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0001305853256389118,
-      "loss": 1.7216,
+      "learning_rate": 0.000435078318219291,
+      "loss": 4.999,
       "step": 3244
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013052349546578728,
-      "loss": 1.3446,
+      "learning_rate": 0.0004348722176422094,
+      "loss": 4.944,
       "step": 3245
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013046166529266282,
-      "loss": 1.3431,
+      "learning_rate": 0.0004346661170651278,
+      "loss": 4.8487,
       "step": 3246
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0001303998351195383,
-      "loss": 1.4656,
+      "learning_rate": 0.0004344600164880462,
+      "loss": 4.9389,
       "step": 3247
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013033800494641383,
-      "loss": 1.3035,
+      "learning_rate": 0.0004342539159109646,
+      "loss": 4.7366,
       "step": 3248
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013027617477328937,
-      "loss": 1.103,
+      "learning_rate": 0.00043404781533388294,
+      "loss": 4.5171,
       "step": 3249
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013021434460016486,
-      "loss": 1.4287,
+      "learning_rate": 0.00043384171475680135,
+      "loss": 4.7362,
       "step": 3250
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00013015251442704038,
-      "loss": 1.3625,
+      "learning_rate": 0.0004336356141797197,
+      "loss": 4.7415,
       "step": 3251
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0001300906842539159,
-      "loss": 1.3452,
+      "learning_rate": 0.0004334295136026381,
+      "loss": 4.8461,
       "step": 3252
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0001300288540807914,
-      "loss": 1.4014,
+      "learning_rate": 0.00043322341302555646,
+      "loss": 5.076,
       "step": 3253
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00012996702390766693,
-      "loss": 1.3353,
+      "learning_rate": 0.00043301731244847487,
+      "loss": 4.828,
       "step": 3254
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00012990519373454244,
-      "loss": 1.1733,
+      "learning_rate": 0.0004328112118713933,
+      "loss": 4.9678,
       "step": 3255
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012984336356141796,
-      "loss": 1.0528,
+      "learning_rate": 0.0004326051112943116,
+      "loss": 4.6477,
       "step": 3256
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012978153338829348,
-      "loss": 1.2415,
+      "learning_rate": 0.00043239901071723003,
+      "loss": 4.6926,
       "step": 3257
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.000129719703215169,
-      "loss": 1.7309,
+      "learning_rate": 0.0004321929101401484,
+      "loss": 4.9184,
       "step": 3258
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0001296578730420445,
-      "loss": 1.4214,
+      "learning_rate": 0.0004319868095630668,
+      "loss": 4.8205,
       "step": 3259
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012959604286892003,
-      "loss": 1.5736,
+      "learning_rate": 0.00043178070898598514,
+      "loss": 5.1338,
       "step": 3260
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012953421269579554,
-      "loss": 1.1891,
+      "learning_rate": 0.00043157460840890355,
+      "loss": 5.0879,
       "step": 3261
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012947238252267106,
-      "loss": 2.0012,
+      "learning_rate": 0.00043136850783182196,
+      "loss": 5.5863,
       "step": 3262
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012941055234954658,
-      "loss": 1.3927,
+      "learning_rate": 0.0004311624072547403,
+      "loss": 4.9713,
       "step": 3263
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0001293487221764221,
-      "loss": 1.0889,
+      "learning_rate": 0.0004309563066776587,
+      "loss": 4.6285,
       "step": 3264
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0001292868920032976,
-      "loss": 1.6544,
+      "learning_rate": 0.00043075020610057707,
+      "loss": 5.0367,
       "step": 3265
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0001292250618301731,
-      "loss": 1.2592,
+      "learning_rate": 0.0004305441055234955,
+      "loss": 4.6544,
       "step": 3266
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012916323165704864,
-      "loss": 0.9291,
+      "learning_rate": 0.00043033800494641383,
+      "loss": 4.5245,
       "step": 3267
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012910140148392413,
-      "loss": 1.0154,
+      "learning_rate": 0.00043013190436933224,
+      "loss": 4.6522,
       "step": 3268
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012903957131079965,
-      "loss": 1.1709,
+      "learning_rate": 0.0004299258037922506,
+      "loss": 4.8567,
       "step": 3269
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012897774113767517,
-      "loss": 1.541,
+      "learning_rate": 0.000429719703215169,
+      "loss": 5.1076,
       "step": 3270
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012891591096455068,
-      "loss": 1.3322,
+      "learning_rate": 0.0004295136026380874,
+      "loss": 4.6998,
       "step": 3271
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0001288540807914262,
-      "loss": 1.4356,
+      "learning_rate": 0.00042930750206100575,
+      "loss": 4.6232,
       "step": 3272
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00012879225061830172,
-      "loss": 1.3169,
+      "learning_rate": 0.00042910140148392416,
+      "loss": 4.7186,
       "step": 3273
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012873042044517723,
-      "loss": 1.5669,
+      "learning_rate": 0.0004288953009068425,
+      "loss": 4.803,
       "step": 3274
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012866859027205275,
-      "loss": 1.5042,
+      "learning_rate": 0.0004286892003297609,
+      "loss": 5.0615,
       "step": 3275
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012860676009892826,
-      "loss": 1.338,
+      "learning_rate": 0.00042848309975267927,
+      "loss": 4.8424,
       "step": 3276
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012854492992580378,
-      "loss": 1.2794,
+      "learning_rate": 0.0004282769991755977,
+      "loss": 4.3099,
       "step": 3277
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0001284830997526793,
-      "loss": 1.334,
+      "learning_rate": 0.0004280708985985161,
+      "loss": 5.034,
       "step": 3278
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012842126957955481,
-      "loss": 1.4943,
+      "learning_rate": 0.0004278647980214345,
+      "loss": 4.8562,
       "step": 3279
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012835943940643033,
-      "loss": 1.0833,
+      "learning_rate": 0.0004276586974443529,
+      "loss": 4.4245,
       "step": 3280
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012829760923330585,
-      "loss": 1.6253,
+      "learning_rate": 0.00042745259686727125,
+      "loss": 4.8343,
       "step": 3281
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012823577906018136,
-      "loss": 0.8064,
+      "learning_rate": 0.00042724649629018966,
+      "loss": 4.3519,
       "step": 3282
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012817394888705688,
-      "loss": 2.2131,
+      "learning_rate": 0.000427040395713108,
+      "loss": 5.2552,
       "step": 3283
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012811211871393237,
-      "loss": 1.1135,
+      "learning_rate": 0.0004268342951360264,
+      "loss": 4.2039,
       "step": 3284
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012805028854080791,
-      "loss": 1.991,
+      "learning_rate": 0.00042662819455894477,
+      "loss": 4.5619,
       "step": 3285
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0001279884583676834,
-      "loss": 2.6617,
+      "learning_rate": 0.0004264220939818632,
+      "loss": 5.1431,
       "step": 3286
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012792662819455892,
-      "loss": 2.2329,
+      "learning_rate": 0.0004262159934047816,
+      "loss": 4.6661,
       "step": 3287
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012786479802143446,
-      "loss": 1.7129,
+      "learning_rate": 0.00042600989282769994,
+      "loss": 4.4522,
       "step": 3288
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00012780296784830995,
-      "loss": 2.377,
+      "learning_rate": 0.00042580379225061834,
+      "loss": 5.4291,
       "step": 3289
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0001277411376751855,
-      "loss": 1.7962,
+      "learning_rate": 0.0004255976916735367,
+      "loss": 4.8745,
       "step": 3290
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.000127679307502061,
-      "loss": 1.757,
+      "learning_rate": 0.0004253915910964551,
+      "loss": 5.3573,
       "step": 3291
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0001276174773289365,
-      "loss": 1.4148,
+      "learning_rate": 0.00042518549051937345,
+      "loss": 4.8001,
       "step": 3292
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012755564715581202,
-      "loss": 1.436,
+      "learning_rate": 0.00042497938994229186,
+      "loss": 4.7845,
       "step": 3293
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012749381698268754,
-      "loss": 1.1924,
+      "learning_rate": 0.00042477328936521027,
+      "loss": 4.6345,
       "step": 3294
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012743198680956305,
-      "loss": 1.3487,
+      "learning_rate": 0.0004245671887881286,
+      "loss": 4.9105,
       "step": 3295
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012737015663643857,
-      "loss": 1.4301,
+      "learning_rate": 0.000424361088211047,
+      "loss": 4.7087,
       "step": 3296
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012730832646331409,
-      "loss": 1.282,
+      "learning_rate": 0.0004241549876339654,
+      "loss": 4.8645,
       "step": 3297
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0001272464962901896,
-      "loss": 1.5351,
+      "learning_rate": 0.0004239488870568838,
+      "loss": 4.7795,
       "step": 3298
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012718466611706512,
-      "loss": 1.7937,
+      "learning_rate": 0.00042374278647980214,
+      "loss": 5.0883,
       "step": 3299
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012712283594394064,
-      "loss": 1.394,
+      "learning_rate": 0.00042353668590272054,
+      "loss": 4.7145,
       "step": 3300
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012706100577081615,
-      "loss": 1.0501,
+      "learning_rate": 0.0004233305853256389,
+      "loss": 4.8431,
       "step": 3301
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012699917559769164,
-      "loss": 1.196,
+      "learning_rate": 0.0004231244847485573,
+      "loss": 4.5773,
       "step": 3302
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012693734542456719,
-      "loss": 1.4739,
+      "learning_rate": 0.0004229183841714757,
+      "loss": 4.8097,
       "step": 3303
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0001268755152514427,
-      "loss": 1.3933,
+      "learning_rate": 0.00042271228359439406,
+      "loss": 4.6245,
       "step": 3304
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012681368507831822,
-      "loss": 1.5522,
+      "learning_rate": 0.00042250618301731247,
+      "loss": 4.8282,
       "step": 3305
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012675185490519374,
-      "loss": 1.3975,
+      "learning_rate": 0.0004223000824402308,
+      "loss": 4.5493,
       "step": 3306
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012669002473206922,
-      "loss": 1.3529,
+      "learning_rate": 0.00042209398186314923,
+      "loss": 4.8475,
       "step": 3307
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012662819455894477,
-      "loss": 1.3622,
+      "learning_rate": 0.0004218878812860676,
+      "loss": 4.9035,
       "step": 3308
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00012656636438582026,
-      "loss": 1.3954,
+      "learning_rate": 0.000421681780708986,
+      "loss": 4.8557,
       "step": 3309
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012650453421269577,
-      "loss": 1.4709,
+      "learning_rate": 0.0004214756801319044,
+      "loss": 4.9576,
       "step": 3310
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012644270403957132,
-      "loss": 1.3034,
+      "learning_rate": 0.00042126957955482275,
+      "loss": 4.7932,
       "step": 3311
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0001263808738664468,
-      "loss": 1.0846,
+      "learning_rate": 0.00042106347897774115,
+      "loss": 4.5029,
       "step": 3312
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012631904369332232,
-      "loss": 1.2364,
+      "learning_rate": 0.0004208573784006595,
+      "loss": 4.6352,
       "step": 3313
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012625721352019784,
-      "loss": 1.3262,
+      "learning_rate": 0.0004206512778235779,
+      "loss": 4.9154,
       "step": 3314
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012619538334707336,
-      "loss": 1.3242,
+      "learning_rate": 0.00042044517724649627,
+      "loss": 4.7988,
       "step": 3315
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012613355317394887,
-      "loss": 1.3979,
+      "learning_rate": 0.00042023907666941467,
+      "loss": 4.9443,
       "step": 3316
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0001260717230008244,
-      "loss": 1.0526,
+      "learning_rate": 0.000420032976092333,
+      "loss": 4.7826,
       "step": 3317
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0001260098928276999,
-      "loss": 1.3563,
+      "learning_rate": 0.00041982687551525143,
+      "loss": 4.7367,
       "step": 3318
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012594806265457542,
-      "loss": 1.1772,
+      "learning_rate": 0.00041962077493816984,
+      "loss": 4.5704,
       "step": 3319
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012588623248145094,
-      "loss": 1.2558,
+      "learning_rate": 0.0004194146743610882,
+      "loss": 4.7499,
       "step": 3320
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012582440230832646,
-      "loss": 0.9659,
+      "learning_rate": 0.0004192085737840066,
+      "loss": 4.6772,
       "step": 3321
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012576257213520197,
-      "loss": 1.219,
+      "learning_rate": 0.00041900247320692495,
+      "loss": 4.7127,
       "step": 3322
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0001257007419620775,
-      "loss": 1.6515,
+      "learning_rate": 0.00041879637262984336,
+      "loss": 5.0778,
       "step": 3323
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.000125638911788953,
-      "loss": 1.1709,
+      "learning_rate": 0.0004185902720527617,
+      "loss": 4.7434,
       "step": 3324
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0001255770816158285,
-      "loss": 1.1336,
+      "learning_rate": 0.0004183841714756801,
+      "loss": 4.7699,
       "step": 3325
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012551525144270404,
-      "loss": 1.4649,
+      "learning_rate": 0.0004181780708985986,
+      "loss": 5.3764,
       "step": 3326
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00012545342126957956,
-      "loss": 1.4599,
+      "learning_rate": 0.00041797197032151693,
+      "loss": 4.9099,
       "step": 3327
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012539159109645505,
-      "loss": 1.1356,
+      "learning_rate": 0.00041776586974443534,
+      "loss": 4.5252,
       "step": 3328
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001253297609233306,
-      "loss": 1.133,
+      "learning_rate": 0.0004175597691673537,
+      "loss": 4.7284,
       "step": 3329
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012526793075020608,
-      "loss": 1.3054,
+      "learning_rate": 0.0004173536685902721,
+      "loss": 4.7711,
       "step": 3330
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001252061005770816,
-      "loss": 1.3394,
+      "learning_rate": 0.00041714756801319045,
+      "loss": 4.6967,
       "step": 3331
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001251442704039571,
-      "loss": 1.2932,
+      "learning_rate": 0.00041694146743610885,
+      "loss": 5.0961,
       "step": 3332
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012508244023083263,
-      "loss": 1.4126,
+      "learning_rate": 0.0004167353668590272,
+      "loss": 5.1053,
       "step": 3333
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012502061005770815,
-      "loss": 1.5745,
+      "learning_rate": 0.0004165292662819456,
+      "loss": 4.3585,
       "step": 3334
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012495877988458366,
-      "loss": 2.1851,
+      "learning_rate": 0.000416323165704864,
+      "loss": 4.8844,
       "step": 3335
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012489694971145918,
-      "loss": 1.9141,
+      "learning_rate": 0.00041611706512778237,
+      "loss": 5.0525,
       "step": 3336
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001248351195383347,
-      "loss": 2.4509,
+      "learning_rate": 0.0004159109645507008,
+      "loss": 5.2952,
       "step": 3337
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001247732893652102,
-      "loss": 1.9225,
+      "learning_rate": 0.00041570486397361913,
+      "loss": 4.8484,
       "step": 3338
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012471145919208573,
-      "loss": 1.5331,
+      "learning_rate": 0.00041549876339653754,
+      "loss": 4.5423,
       "step": 3339
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012464962901896125,
-      "loss": 1.4024,
+      "learning_rate": 0.0004152926628194559,
+      "loss": 4.7553,
       "step": 3340
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012458779884583676,
-      "loss": 1.3268,
+      "learning_rate": 0.0004150865622423743,
+      "loss": 4.5723,
       "step": 3341
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00012452596867271228,
-      "loss": 1.3334,
+      "learning_rate": 0.0004148804616652927,
+      "loss": 4.8167,
       "step": 3342
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001244641384995878,
-      "loss": 1.7794,
+      "learning_rate": 0.00041467436108821106,
+      "loss": 4.9743,
       "step": 3343
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0001244023083264633,
-      "loss": 1.0735,
+      "learning_rate": 0.00041446826051112946,
+      "loss": 4.5933,
       "step": 3344
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012434047815333883,
-      "loss": 1.5357,
+      "learning_rate": 0.0004142621599340478,
+      "loss": 5.5062,
       "step": 3345
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012427864798021432,
-      "loss": 1.2546,
+      "learning_rate": 0.0004140560593569662,
+      "loss": 4.4583,
       "step": 3346
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012421681780708986,
-      "loss": 1.1049,
+      "learning_rate": 0.0004138499587798846,
+      "loss": 4.3661,
       "step": 3347
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012415498763396535,
-      "loss": 1.4227,
+      "learning_rate": 0.000413643858202803,
+      "loss": 4.6886,
       "step": 3348
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012409315746084087,
-      "loss": 1.3007,
+      "learning_rate": 0.00041343775762572133,
+      "loss": 4.5952,
       "step": 3349
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012403132728771638,
-      "loss": 1.2593,
+      "learning_rate": 0.00041323165704863974,
+      "loss": 4.7484,
       "step": 3350
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0001239694971145919,
-      "loss": 1.4166,
+      "learning_rate": 0.00041302555647155815,
+      "loss": 4.8139,
       "step": 3351
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012390766694146744,
-      "loss": 1.1771,
+      "learning_rate": 0.0004128194558944765,
+      "loss": 4.7335,
       "step": 3352
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012384583676834293,
-      "loss": 1.1309,
+      "learning_rate": 0.0004126133553173949,
+      "loss": 4.6977,
       "step": 3353
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012378400659521845,
-      "loss": 1.6252,
+      "learning_rate": 0.00041240725474031326,
+      "loss": 4.9789,
       "step": 3354
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012372217642209397,
-      "loss": 1.2847,
+      "learning_rate": 0.00041220115416323167,
+      "loss": 4.8396,
       "step": 3355
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012366034624896948,
-      "loss": 1.3757,
+      "learning_rate": 0.00041199505358615,
+      "loss": 4.6689,
       "step": 3356
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.000123598516075845,
-      "loss": 1.3846,
+      "learning_rate": 0.0004117889530090684,
+      "loss": 5.142,
       "step": 3357
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012353668590272052,
-      "loss": 1.0802,
+      "learning_rate": 0.00041158285243198683,
+      "loss": 4.5445,
       "step": 3358
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012347485572959603,
-      "loss": 1.4154,
+      "learning_rate": 0.0004113767518549052,
+      "loss": 5.269,
       "step": 3359
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012341302555647155,
-      "loss": 1.0709,
+      "learning_rate": 0.0004111706512778236,
+      "loss": 5.2971,
       "step": 3360
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012335119538334707,
-      "loss": 1.9649,
+      "learning_rate": 0.00041096455070074194,
+      "loss": 5.1702,
       "step": 3361
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00012328936521022258,
-      "loss": 1.2734,
+      "learning_rate": 0.00041075845012366035,
+      "loss": 4.9609,
       "step": 3362
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0001232275350370981,
-      "loss": 1.12,
+      "learning_rate": 0.0004105523495465787,
+      "loss": 4.5996,
       "step": 3363
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0001231657048639736,
-      "loss": 1.1303,
+      "learning_rate": 0.0004103462489694971,
+      "loss": 4.7091,
       "step": 3364
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012310387469084913,
-      "loss": 1.5005,
+      "learning_rate": 0.00041014014839241546,
+      "loss": 5.1736,
       "step": 3365
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012304204451772462,
-      "loss": 1.4304,
+      "learning_rate": 0.00040993404781533387,
+      "loss": 4.8153,
       "step": 3366
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012298021434460017,
-      "loss": 1.255,
+      "learning_rate": 0.0004097279472382523,
+      "loss": 4.9472,
       "step": 3367
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012291838417147568,
-      "loss": 1.4336,
+      "learning_rate": 0.00040952184666117063,
+      "loss": 5.0991,
       "step": 3368
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012285655399835117,
-      "loss": 1.2612,
+      "learning_rate": 0.00040931574608408903,
+      "loss": 4.9006,
       "step": 3369
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012279472382522672,
-      "loss": 1.179,
+      "learning_rate": 0.0004091096455070074,
+      "loss": 4.7975,
       "step": 3370
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0001227328936521022,
-      "loss": 1.2965,
+      "learning_rate": 0.0004089035449299258,
+      "loss": 5.0763,
       "step": 3371
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012267106347897772,
-      "loss": 1.4477,
+      "learning_rate": 0.00040869744435284415,
+      "loss": 4.9482,
       "step": 3372
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012260923330585324,
-      "loss": 0.9841,
+      "learning_rate": 0.0004084913437757626,
+      "loss": 4.6109,
       "step": 3373
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012254740313272875,
-      "loss": 0.9838,
+      "learning_rate": 0.000408285243198681,
+      "loss": 4.9554,
       "step": 3374
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012248557295960427,
-      "loss": 1.8337,
+      "learning_rate": 0.00040807914262159937,
+      "loss": 5.4849,
       "step": 3375
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0001224237427864798,
-      "loss": 1.1102,
+      "learning_rate": 0.00040787304204451777,
+      "loss": 4.6041,
       "step": 3376
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0001223619126133553,
-      "loss": 1.1328,
+      "learning_rate": 0.0004076669414674361,
+      "loss": 5.0535,
       "step": 3377
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012230008244023082,
-      "loss": 1.2854,
+      "learning_rate": 0.00040746084089035453,
+      "loss": 4.5565,
       "step": 3378
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012223825226710634,
-      "loss": 0.9674,
+      "learning_rate": 0.0004072547403132729,
+      "loss": 4.4918,
       "step": 3379
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00012217642209398185,
-      "loss": 0.8814,
+      "learning_rate": 0.0004070486397361913,
+      "loss": 4.3907,
       "step": 3380
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012211459192085737,
-      "loss": 2.2954,
+      "learning_rate": 0.0004068425391591097,
+      "loss": 6.3704,
       "step": 3381
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012205276174773289,
-      "loss": 1.3387,
+      "learning_rate": 0.00040663643858202805,
+      "loss": 4.7241,
       "step": 3382
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012199093157460839,
-      "loss": 1.4521,
+      "learning_rate": 0.00040643033800494646,
+      "loss": 4.9549,
       "step": 3383
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012192910140148392,
-      "loss": 1.3801,
+      "learning_rate": 0.0004062242374278648,
+      "loss": 4.4062,
       "step": 3384
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012186727122835942,
-      "loss": 1.5212,
+      "learning_rate": 0.0004060181368507832,
+      "loss": 4.7854,
       "step": 3385
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012180544105523495,
-      "loss": 1.2222,
+      "learning_rate": 0.00040581203627370157,
+      "loss": 4.8542,
       "step": 3386
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012174361088211046,
-      "loss": 1.5097,
+      "learning_rate": 0.00040560593569662,
+      "loss": 4.9975,
       "step": 3387
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012168178070898597,
-      "loss": 1.8296,
+      "learning_rate": 0.00040539983511953833,
+      "loss": 5.1747,
       "step": 3388
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012161995053586148,
-      "loss": 1.4567,
+      "learning_rate": 0.00040519373454245673,
+      "loss": 4.7826,
       "step": 3389
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.000121558120362737,
-      "loss": 1.298,
+      "learning_rate": 0.00040498763396537514,
+      "loss": 5.008,
       "step": 3390
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012149629018961252,
-      "loss": 1.4837,
+      "learning_rate": 0.0004047815333882935,
+      "loss": 4.9458,
       "step": 3391
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012143446001648804,
-      "loss": 1.1635,
+      "learning_rate": 0.0004045754328112119,
+      "loss": 4.6367,
       "step": 3392
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012137262984336356,
-      "loss": 1.1003,
+      "learning_rate": 0.00040436933223413025,
+      "loss": 4.6591,
       "step": 3393
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012131079967023906,
-      "loss": 1.3598,
+      "learning_rate": 0.00040416323165704866,
+      "loss": 4.9568,
       "step": 3394
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012124896949711459,
-      "loss": 1.2039,
+      "learning_rate": 0.000403957131079967,
+      "loss": 4.8437,
       "step": 3395
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012118713932399009,
-      "loss": 1.5189,
+      "learning_rate": 0.0004037510305028854,
+      "loss": 5.0203,
       "step": 3396
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012112530915086561,
-      "loss": 1.0661,
+      "learning_rate": 0.0004035449299258038,
+      "loss": 4.5651,
       "step": 3397
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00012106347897774114,
-      "loss": 1.3072,
+      "learning_rate": 0.0004033388293487222,
+      "loss": 5.0757,
       "step": 3398
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012100164880461664,
-      "loss": 1.0874,
+      "learning_rate": 0.0004031327287716406,
+      "loss": 4.6867,
       "step": 3399
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012093981863149216,
-      "loss": 1.1885,
+      "learning_rate": 0.00040292662819455894,
+      "loss": 4.5968,
       "step": 3400
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012087798845836768,
-      "loss": 1.2352,
+      "learning_rate": 0.00040272052761747734,
+      "loss": 4.6518,
       "step": 3401
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012081615828524319,
-      "loss": 1.0468,
+      "learning_rate": 0.0004025144270403957,
+      "loss": 4.5181,
       "step": 3402
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0001207543281121187,
-      "loss": 1.3422,
+      "learning_rate": 0.0004023083264633141,
+      "loss": 4.9078,
       "step": 3403
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012069249793899423,
-      "loss": 1.3134,
+      "learning_rate": 0.00040210222588623246,
+      "loss": 4.7953,
       "step": 3404
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012063066776586973,
-      "loss": 1.0336,
+      "learning_rate": 0.00040189612530915086,
+      "loss": 4.6249,
       "step": 3405
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012056883759274525,
-      "loss": 1.1262,
+      "learning_rate": 0.00040169002473206927,
+      "loss": 4.7323,
       "step": 3406
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012050700741962078,
-      "loss": 0.9348,
+      "learning_rate": 0.0004014839241549876,
+      "loss": 4.7391,
       "step": 3407
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012044517724649628,
-      "loss": 1.4128,
+      "learning_rate": 0.00040127782357790603,
+      "loss": 5.1484,
       "step": 3408
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0001203833470733718,
-      "loss": 1.771,
+      "learning_rate": 0.0004010717230008244,
+      "loss": 5.1443,
       "step": 3409
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012032151690024731,
-      "loss": 1.2512,
+      "learning_rate": 0.0004008656224237428,
+      "loss": 4.8747,
       "step": 3410
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012025968672712283,
-      "loss": 1.133,
+      "learning_rate": 0.00040065952184666114,
+      "loss": 4.756,
       "step": 3411
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012019785655399833,
-      "loss": 1.1623,
+      "learning_rate": 0.00040045342126957955,
+      "loss": 4.8372,
       "step": 3412
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012013602638087386,
-      "loss": 1.4217,
+      "learning_rate": 0.00040024732069249795,
+      "loss": 4.4609,
       "step": 3413
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012007419620774938,
-      "loss": 1.4916,
+      "learning_rate": 0.0004000412201154163,
+      "loss": 4.6454,
       "step": 3414
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00012001236603462488,
-      "loss": 1.3487,
+      "learning_rate": 0.0003998351195383347,
+      "loss": 4.7215,
       "step": 3415
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00011995053586150041,
-      "loss": 0.9573,
+      "learning_rate": 0.00039962901896125306,
+      "loss": 4.6686,
       "step": 3416
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011988870568837591,
-      "loss": 1.1224,
+      "learning_rate": 0.00039942291838417147,
+      "loss": 4.7053,
       "step": 3417
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011982687551525143,
-      "loss": 1.067,
+      "learning_rate": 0.0003992168178070898,
+      "loss": 4.5054,
       "step": 3418
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011976504534212695,
-      "loss": 0.8829,
+      "learning_rate": 0.0003990107172300083,
+      "loss": 4.6862,
       "step": 3419
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011970321516900246,
-      "loss": 1.1192,
+      "learning_rate": 0.00039880461665292664,
+      "loss": 4.5206,
       "step": 3420
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011964138499587797,
-      "loss": 1.3218,
+      "learning_rate": 0.00039859851607584504,
+      "loss": 4.9256,
       "step": 3421
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0001195795548227535,
-      "loss": 0.9506,
+      "learning_rate": 0.00039839241549876345,
+      "loss": 4.8188,
       "step": 3422
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011951772464962901,
-      "loss": 1.2099,
+      "learning_rate": 0.0003981863149216818,
+      "loss": 4.7555,
       "step": 3423
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011945589447650452,
-      "loss": 1.0526,
+      "learning_rate": 0.0003979802143446002,
+      "loss": 4.5959,
       "step": 3424
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011939406430338005,
-      "loss": 1.1151,
+      "learning_rate": 0.00039777411376751856,
+      "loss": 4.6231,
       "step": 3425
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011933223413025555,
-      "loss": 0.9555,
+      "learning_rate": 0.00039756801319043697,
+      "loss": 4.5784,
       "step": 3426
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011927040395713107,
-      "loss": 1.0532,
+      "learning_rate": 0.0003973619126133553,
+      "loss": 4.4765,
       "step": 3427
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011920857378400658,
-      "loss": 1.2026,
+      "learning_rate": 0.00039715581203627373,
+      "loss": 4.8259,
       "step": 3428
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0001191467436108821,
-      "loss": 0.9166,
+      "learning_rate": 0.00039694971145919213,
+      "loss": 4.3661,
       "step": 3429
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011908491343775763,
-      "loss": 1.6136,
+      "learning_rate": 0.0003967436108821105,
+      "loss": 5.837,
       "step": 3430
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011902308326463313,
-      "loss": 0.7415,
+      "learning_rate": 0.0003965375103050289,
+      "loss": 4.5403,
       "step": 3431
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011896125309150865,
-      "loss": 1.6366,
+      "learning_rate": 0.00039633140972794725,
+      "loss": 5.1954,
       "step": 3432
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011889942291838415,
-      "loss": 1.3726,
+      "learning_rate": 0.00039612530915086565,
+      "loss": 4.274,
       "step": 3433
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00011883759274525968,
-      "loss": 1.6397,
+      "learning_rate": 0.000395919208573784,
+      "loss": 4.4951,
       "step": 3434
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011877576257213519,
-      "loss": 1.9551,
+      "learning_rate": 0.0003957131079967024,
+      "loss": 4.7472,
       "step": 3435
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0001187139323990107,
-      "loss": 2.3767,
+      "learning_rate": 0.00039550700741962076,
+      "loss": 4.7832,
       "step": 3436
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011865210222588622,
-      "loss": 1.5948,
+      "learning_rate": 0.00039530090684253917,
+      "loss": 4.7386,
       "step": 3437
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011859027205276174,
-      "loss": 1.3994,
+      "learning_rate": 0.0003950948062654576,
+      "loss": 4.7036,
       "step": 3438
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011852844187963727,
-      "loss": 1.2472,
+      "learning_rate": 0.00039488870568837593,
+      "loss": 4.6697,
       "step": 3439
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011846661170651277,
-      "loss": 1.1264,
+      "learning_rate": 0.00039468260511129434,
+      "loss": 4.7119,
       "step": 3440
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011840478153338828,
-      "loss": 2.2506,
+      "learning_rate": 0.0003944765045342127,
+      "loss": 5.2962,
       "step": 3441
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011834295136026379,
-      "loss": 1.2318,
+      "learning_rate": 0.0003942704039571311,
+      "loss": 4.8954,
       "step": 3442
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011828112118713932,
-      "loss": 1.3146,
+      "learning_rate": 0.00039406430338004945,
+      "loss": 5.0259,
       "step": 3443
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011821929101401482,
-      "loss": 1.4527,
+      "learning_rate": 0.00039385820280296786,
+      "loss": 4.7879,
       "step": 3444
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011815746084089035,
-      "loss": 1.1627,
+      "learning_rate": 0.00039365210222588626,
+      "loss": 4.8812,
       "step": 3445
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011809563066776587,
-      "loss": 1.2172,
+      "learning_rate": 0.0003934460016488046,
+      "loss": 4.8308,
       "step": 3446
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011803380049464137,
-      "loss": 1.2584,
+      "learning_rate": 0.000393239901071723,
+      "loss": 5.0648,
       "step": 3447
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0001179719703215169,
-      "loss": 1.0986,
+      "learning_rate": 0.0003930338004946414,
+      "loss": 4.6062,
       "step": 3448
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0001179101401483924,
-      "loss": 1.1608,
+      "learning_rate": 0.0003928276999175598,
+      "loss": 4.8874,
       "step": 3449
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011784830997526792,
-      "loss": 1.298,
+      "learning_rate": 0.00039262159934047813,
+      "loss": 5.0936,
       "step": 3450
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011778647980214342,
-      "loss": 1.1799,
+      "learning_rate": 0.00039241549876339654,
+      "loss": 4.5317,
       "step": 3451
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00011772464962901895,
-      "loss": 1.1447,
+      "learning_rate": 0.0003922093981863149,
+      "loss": 4.8123,
       "step": 3452
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011766281945589446,
-      "loss": 1.006,
+      "learning_rate": 0.0003920032976092333,
+      "loss": 4.3349,
       "step": 3453
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011760098928276999,
-      "loss": 1.2399,
+      "learning_rate": 0.0003917971970321517,
+      "loss": 4.7756,
       "step": 3454
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0001175391591096455,
-      "loss": 1.0699,
+      "learning_rate": 0.00039159109645507006,
+      "loss": 4.5968,
       "step": 3455
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.000117477328936521,
-      "loss": 0.8493,
+      "learning_rate": 0.00039138499587798846,
+      "loss": 4.4312,
       "step": 3456
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011741549876339654,
-      "loss": 1.3963,
+      "learning_rate": 0.0003911788953009068,
+      "loss": 4.8211,
       "step": 3457
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011735366859027204,
-      "loss": 1.2714,
+      "learning_rate": 0.0003909727947238252,
+      "loss": 4.9525,
       "step": 3458
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011729183841714756,
-      "loss": 1.2309,
+      "learning_rate": 0.0003907666941467436,
+      "loss": 4.8558,
       "step": 3459
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011723000824402306,
-      "loss": 1.1558,
+      "learning_rate": 0.000390560593569662,
+      "loss": 4.5851,
       "step": 3460
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011716817807089859,
-      "loss": 1.2222,
+      "learning_rate": 0.0003903544929925804,
+      "loss": 4.4795,
       "step": 3461
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0001171063478977741,
-      "loss": 0.9112,
+      "learning_rate": 0.00039014839241549874,
+      "loss": 4.2778,
       "step": 3462
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011704451772464962,
-      "loss": 1.3143,
+      "learning_rate": 0.00038994229183841715,
+      "loss": 4.8323,
       "step": 3463
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011698268755152514,
-      "loss": 1.1644,
+      "learning_rate": 0.0003897361912613355,
+      "loss": 4.5137,
       "step": 3464
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011692085737840064,
-      "loss": 1.1347,
+      "learning_rate": 0.0003895300906842539,
+      "loss": 4.9396,
       "step": 3465
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011685902720527617,
-      "loss": 1.1918,
+      "learning_rate": 0.0003893239901071723,
+      "loss": 5.0272,
       "step": 3466
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011679719703215168,
-      "loss": 1.1075,
+      "learning_rate": 0.0003891178895300907,
+      "loss": 4.7508,
       "step": 3467
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011673536685902719,
-      "loss": 0.9503,
+      "learning_rate": 0.0003889117889530091,
+      "loss": 4.433,
       "step": 3468
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00011667353668590271,
-      "loss": 1.0416,
+      "learning_rate": 0.0003887056883759275,
+      "loss": 4.4877,
       "step": 3469
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011661170651277823,
-      "loss": 1.4829,
+      "learning_rate": 0.0003884995877988459,
+      "loss": 5.4642,
       "step": 3470
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011654987633965374,
-      "loss": 0.9819,
+      "learning_rate": 0.00038829348722176424,
+      "loss": 4.4374,
       "step": 3471
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011648804616652926,
-      "loss": 1.0256,
+      "learning_rate": 0.00038808738664468265,
+      "loss": 4.611,
       "step": 3472
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011642621599340477,
-      "loss": 1.4037,
+      "learning_rate": 0.000387881286067601,
+      "loss": 5.8008,
       "step": 3473
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011636438582028028,
-      "loss": 1.9408,
+      "learning_rate": 0.0003876751854905194,
+      "loss": 5.458,
       "step": 3474
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011630255564715581,
-      "loss": 1.0198,
+      "learning_rate": 0.00038746908491343776,
+      "loss": 4.496,
       "step": 3475
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011624072547403131,
-      "loss": 0.9613,
+      "learning_rate": 0.00038726298433635616,
+      "loss": 4.5265,
       "step": 3476
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011617889530090683,
-      "loss": 1.2666,
+      "learning_rate": 0.00038705688375927457,
+      "loss": 4.8058,
       "step": 3477
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011611706512778236,
-      "loss": 1.3603,
+      "learning_rate": 0.0003868507831821929,
+      "loss": 4.735,
       "step": 3478
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011605523495465786,
-      "loss": 1.5743,
+      "learning_rate": 0.00038664468260511133,
+      "loss": 5.0505,
       "step": 3479
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011599340478153338,
-      "loss": 1.4339,
+      "learning_rate": 0.0003864385820280297,
+      "loss": 4.7965,
       "step": 3480
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0001159315746084089,
-      "loss": 1.0952,
+      "learning_rate": 0.0003862324814509481,
+      "loss": 4.9624,
       "step": 3481
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011586974443528441,
-      "loss": 1.2582,
+      "learning_rate": 0.00038602638087386644,
+      "loss": 4.3876,
       "step": 3482
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011580791426215991,
-      "loss": 0.9257,
+      "learning_rate": 0.00038582028029678485,
+      "loss": 4.846,
       "step": 3483
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011574608408903544,
-      "loss": 1.3527,
+      "learning_rate": 0.00038561417971970326,
+      "loss": 4.651,
       "step": 3484
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011568425391591096,
-      "loss": 1.6639,
+      "learning_rate": 0.0003854080791426216,
+      "loss": 4.9595,
       "step": 3485
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00011562242374278646,
-      "loss": 1.5074,
+      "learning_rate": 0.00038520197856554,
+      "loss": 4.6494,
       "step": 3486
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.000115560593569662,
-      "loss": 1.7117,
+      "learning_rate": 0.00038499587798845837,
+      "loss": 4.8342,
       "step": 3487
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0001154987633965375,
-      "loss": 1.2899,
+      "learning_rate": 0.0003847897774113768,
+      "loss": 4.728,
       "step": 3488
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011543693322341301,
-      "loss": 1.5523,
+      "learning_rate": 0.0003845836768342951,
+      "loss": 4.9452,
       "step": 3489
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011537510305028853,
-      "loss": 1.8009,
+      "learning_rate": 0.00038437757625721353,
+      "loss": 5.6787,
       "step": 3490
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011531327287716405,
-      "loss": 1.2658,
+      "learning_rate": 0.0003841714756801319,
+      "loss": 4.6806,
       "step": 3491
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011525144270403955,
-      "loss": 1.4039,
+      "learning_rate": 0.0003839653751030503,
+      "loss": 4.8563,
       "step": 3492
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011518961253091508,
-      "loss": 1.3114,
+      "learning_rate": 0.0003837592745259687,
+      "loss": 4.8826,
       "step": 3493
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0001151277823577906,
-      "loss": 1.4445,
+      "learning_rate": 0.00038355317394888705,
+      "loss": 4.7766,
       "step": 3494
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0001150659521846661,
-      "loss": 1.326,
+      "learning_rate": 0.00038334707337180546,
+      "loss": 4.8236,
       "step": 3495
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011500412201154163,
-      "loss": 1.3273,
+      "learning_rate": 0.0003831409727947238,
+      "loss": 4.9607,
       "step": 3496
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011494229183841713,
-      "loss": 1.1979,
+      "learning_rate": 0.0003829348722176422,
+      "loss": 4.7433,
       "step": 3497
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011488046166529265,
-      "loss": 1.2051,
+      "learning_rate": 0.00038272877164056057,
+      "loss": 4.6359,
       "step": 3498
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011481863149216817,
-      "loss": 0.9845,
+      "learning_rate": 0.000382522671063479,
+      "loss": 4.7336,
       "step": 3499
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011475680131904368,
-      "loss": 1.0824,
+      "learning_rate": 0.0003823165704863974,
+      "loss": 4.566,
       "step": 3500
     },
     {
       "epoch": 1.96,
-      "eval_loss": 1.5747624635696411,
-      "eval_runtime": 1250.4918,
-      "eval_samples_per_second": 2.113,
+      "eval_loss": 5.284274101257324,
+      "eval_runtime": 1249.3998,
+      "eval_samples_per_second": 2.115,
       "eval_steps_per_second": 0.265,
-      "eval_wer": 1.1541848472828242,
+      "eval_wer": 1.9802261007536692,
       "step": 3500
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011469497114591921,
-      "loss": 1.5026,
+      "learning_rate": 0.00038211046990931574,
+      "loss": 4.8114,
       "step": 3501
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011463314097279472,
-      "loss": 2.1573,
+      "learning_rate": 0.00038190436933223414,
+      "loss": 6.0071,
       "step": 3502
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011457131079967023,
-      "loss": 1.3399,
+      "learning_rate": 0.0003816982687551525,
+      "loss": 4.852,
       "step": 3503
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011450948062654573,
-      "loss": 1.0647,
+      "learning_rate": 0.0003814921681780709,
+      "loss": 4.3603,
       "step": 3504
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00011444765045342127,
-      "loss": 0.8481,
+      "learning_rate": 0.00038128606760098925,
+      "loss": 4.451,
       "step": 3505
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011438582028029677,
-      "loss": 1.1068,
+      "learning_rate": 0.00038107996702390766,
+      "loss": 4.8503,
       "step": 3506
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0001143239901071723,
-      "loss": 1.4013,
+      "learning_rate": 0.000380873866446826,
+      "loss": 4.6641,
       "step": 3507
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0001142621599340478,
-      "loss": 1.5757,
+      "learning_rate": 0.0003806677658697444,
+      "loss": 4.9771,
       "step": 3508
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011420032976092332,
-      "loss": 1.1413,
+      "learning_rate": 0.0003804616652926628,
+      "loss": 4.9483,
       "step": 3509
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011413849958779885,
-      "loss": 1.5788,
+      "learning_rate": 0.0003802555647155812,
+      "loss": 5.0402,
       "step": 3510
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011407666941467435,
-      "loss": 0.7362,
+      "learning_rate": 0.0003800494641384996,
+      "loss": 4.2775,
       "step": 3511
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011401483924154987,
-      "loss": 1.1358,
+      "learning_rate": 0.00037984336356141794,
+      "loss": 4.7486,
       "step": 3512
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011395300906842537,
-      "loss": 1.2194,
+      "learning_rate": 0.0003796372629843364,
+      "loss": 4.8403,
       "step": 3513
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0001138911788953009,
-      "loss": 0.8492,
+      "learning_rate": 0.00037943116240725475,
+      "loss": 4.4698,
       "step": 3514
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0001138293487221764,
-      "loss": 1.1213,
+      "learning_rate": 0.00037922506183017316,
+      "loss": 4.7063,
       "step": 3515
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011376751854905193,
-      "loss": 0.9231,
+      "learning_rate": 0.00037901896125309157,
+      "loss": 4.5009,
       "step": 3516
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011370568837592745,
-      "loss": 1.1022,
+      "learning_rate": 0.0003788128606760099,
+      "loss": 4.4723,
       "step": 3517
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011364385820280295,
-      "loss": 1.2717,
+      "learning_rate": 0.0003786067600989283,
+      "loss": 4.8936,
       "step": 3518
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011358202802967848,
-      "loss": 0.9696,
+      "learning_rate": 0.0003784006595218467,
+      "loss": 4.5395,
       "step": 3519
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011352019785655399,
-      "loss": 0.85,
+      "learning_rate": 0.0003781945589447651,
+      "loss": 4.5365,
       "step": 3520
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0001134583676834295,
-      "loss": 1.0896,
+      "learning_rate": 0.00037798845836768344,
+      "loss": 4.6528,
       "step": 3521
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011339653751030502,
-      "loss": 1.2867,
+      "learning_rate": 0.00037778235779060184,
+      "loss": 4.8759,
       "step": 3522
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00011333470733718054,
-      "loss": 1.2267,
+      "learning_rate": 0.0003775762572135202,
+      "loss": 4.8592,
       "step": 3523
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011327287716405604,
-      "loss": 1.07,
+      "learning_rate": 0.0003773701566364386,
+      "loss": 5.1568,
       "step": 3524
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011321104699093157,
-      "loss": 1.162,
+      "learning_rate": 0.000377164056059357,
+      "loss": 4.769,
       "step": 3525
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011314921681780709,
-      "loss": 1.1973,
+      "learning_rate": 0.00037695795548227536,
+      "loss": 4.8461,
       "step": 3526
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011308738664468259,
-      "loss": 1.1363,
+      "learning_rate": 0.00037675185490519377,
+      "loss": 4.8091,
       "step": 3527
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011302555647155812,
-      "loss": 1.2676,
+      "learning_rate": 0.0003765457543281121,
+      "loss": 4.93,
       "step": 3528
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011296372629843362,
-      "loss": 0.9488,
+      "learning_rate": 0.0003763396537510305,
+      "loss": 4.5999,
       "step": 3529
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011290189612530914,
-      "loss": 0.8753,
+      "learning_rate": 0.0003761335531739489,
+      "loss": 4.5166,
       "step": 3530
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011284006595218466,
-      "loss": 0.9292,
+      "learning_rate": 0.0003759274525968673,
+      "loss": 4.8218,
       "step": 3531
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011277823577906017,
-      "loss": 0.8495,
+      "learning_rate": 0.0003757213520197857,
+      "loss": 4.739,
       "step": 3532
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011271640560593569,
-      "loss": 1.3275,
+      "learning_rate": 0.00037551525144270405,
+      "loss": 4.2466,
       "step": 3533
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0001126545754328112,
-      "loss": 1.5902,
+      "learning_rate": 0.00037530915086562245,
+      "loss": 4.2381,
       "step": 3534
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011259274525968672,
-      "loss": 1.7392,
+      "learning_rate": 0.0003751030502885408,
+      "loss": 4.7059,
       "step": 3535
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011253091508656223,
-      "loss": 1.6502,
+      "learning_rate": 0.0003748969497114592,
+      "loss": 4.7721,
       "step": 3536
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011246908491343776,
-      "loss": 1.5861,
+      "learning_rate": 0.00037469084913437756,
+      "loss": 4.7028,
       "step": 3537
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011240725474031326,
-      "loss": 1.2973,
+      "learning_rate": 0.00037448474855729597,
+      "loss": 4.5897,
       "step": 3538
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011234542456718877,
-      "loss": 1.1034,
+      "learning_rate": 0.0003742786479802143,
+      "loss": 4.6412,
       "step": 3539
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011228359439406429,
-      "loss": 1.5568,
+      "learning_rate": 0.00037407254740313273,
+      "loss": 4.6575,
       "step": 3540
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00011222176422093981,
-      "loss": 1.2654,
+      "learning_rate": 0.00037386644682605114,
+      "loss": 5.1197,
       "step": 3541
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011215993404781532,
-      "loss": 1.208,
+      "learning_rate": 0.0003736603462489695,
+      "loss": 4.4617,
       "step": 3542
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011209810387469084,
-      "loss": 1.5909,
+      "learning_rate": 0.0003734542456718879,
+      "loss": 5.1106,
       "step": 3543
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011203627370156636,
-      "loss": 1.0455,
+      "learning_rate": 0.00037324814509480625,
+      "loss": 4.5566,
       "step": 3544
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011197444352844186,
-      "loss": 1.0822,
+      "learning_rate": 0.00037304204451772465,
+      "loss": 4.4198,
       "step": 3545
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011191261335531739,
-      "loss": 1.1583,
+      "learning_rate": 0.000372835943940643,
+      "loss": 4.6055,
       "step": 3546
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0001118507831821929,
-      "loss": 1.3616,
+      "learning_rate": 0.0003726298433635614,
+      "loss": 5.1071,
       "step": 3547
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011178895300906841,
-      "loss": 1.1636,
+      "learning_rate": 0.0003724237427864798,
+      "loss": 4.9423,
       "step": 3548
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011172712283594394,
-      "loss": 0.8173,
+      "learning_rate": 0.00037221764220939817,
+      "loss": 4.5571,
       "step": 3549
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011166529266281944,
-      "loss": 1.0227,
+      "learning_rate": 0.0003720115416323166,
+      "loss": 4.6288,
       "step": 3550
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011160346248969496,
-      "loss": 1.2641,
+      "learning_rate": 0.00037180544105523493,
+      "loss": 4.9229,
       "step": 3551
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011154163231657048,
-      "loss": 1.0284,
+      "learning_rate": 0.00037159934047815334,
+      "loss": 4.4566,
       "step": 3552
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.000111479802143446,
-      "loss": 1.0452,
+      "learning_rate": 0.0003713932399010717,
+      "loss": 4.8379,
       "step": 3553
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0001114179719703215,
-      "loss": 1.1806,
+      "learning_rate": 0.0003711871393239901,
+      "loss": 5.0964,
       "step": 3554
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011135614179719703,
-      "loss": 1.3042,
+      "learning_rate": 0.00037098103874690845,
+      "loss": 5.0211,
       "step": 3555
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011129431162407253,
-      "loss": 1.5433,
+      "learning_rate": 0.00037077493816982686,
+      "loss": 5.2793,
       "step": 3556
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011123248145094805,
-      "loss": 1.0189,
+      "learning_rate": 0.00037056883759274526,
+      "loss": 4.647,
       "step": 3557
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011117065127782358,
-      "loss": 1.128,
+      "learning_rate": 0.0003703627370156636,
+      "loss": 4.7218,
       "step": 3558
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00011110882110469908,
-      "loss": 0.8223,
+      "learning_rate": 0.0003701566364385821,
+      "loss": 4.9586,
       "step": 3559
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011104699093157461,
-      "loss": 0.6847,
+      "learning_rate": 0.00036995053586150043,
+      "loss": 4.3742,
       "step": 3560
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011098516075845011,
-      "loss": 1.0183,
+      "learning_rate": 0.00036974443528441884,
+      "loss": 4.5387,
       "step": 3561
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011092333058532563,
-      "loss": 1.0051,
+      "learning_rate": 0.0003695383347073372,
+      "loss": 4.7167,
       "step": 3562
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011086150041220113,
-      "loss": 1.2309,
+      "learning_rate": 0.0003693322341302556,
+      "loss": 4.7,
       "step": 3563
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011079967023907666,
-      "loss": 1.303,
+      "learning_rate": 0.000369126133553174,
+      "loss": 5.0233,
       "step": 3564
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011073784006595218,
-      "loss": 0.9803,
+      "learning_rate": 0.00036892003297609235,
+      "loss": 4.5137,
       "step": 3565
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011067600989282768,
-      "loss": 1.575,
+      "learning_rate": 0.00036871393239901076,
+      "loss": 5.0293,
       "step": 3566
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011061417971970321,
-      "loss": 1.8148,
+      "learning_rate": 0.0003685078318219291,
+      "loss": 5.1896,
       "step": 3567
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011055234954657872,
-      "loss": 0.5387,
+      "learning_rate": 0.0003683017312448475,
+      "loss": 3.5698,
       "step": 3568
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011049051937345425,
-      "loss": 1.2341,
+      "learning_rate": 0.0003680956306677659,
+      "loss": 4.3443,
       "step": 3569
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011042868920032975,
-      "loss": 0.879,
+      "learning_rate": 0.0003678895300906843,
+      "loss": 4.1845,
       "step": 3570
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011036685902720526,
-      "loss": 0.9203,
+      "learning_rate": 0.00036768342951360263,
+      "loss": 4.2753,
       "step": 3571
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011030502885408077,
-      "loss": 0.8894,
+      "learning_rate": 0.00036747732893652104,
+      "loss": 4.1534,
       "step": 3572
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0001102431986809563,
-      "loss": 0.8385,
+      "learning_rate": 0.00036727122835943945,
+      "loss": 4.1776,
       "step": 3573
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011018136850783181,
-      "loss": 0.7348,
+      "learning_rate": 0.0003670651277823578,
+      "loss": 4.0363,
       "step": 3574
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011011953833470733,
-      "loss": 0.8468,
+      "learning_rate": 0.0003668590272052762,
+      "loss": 4.3003,
       "step": 3575
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00011005770816158285,
-      "loss": 0.8348,
+      "learning_rate": 0.00036665292662819456,
+      "loss": 4.1835,
       "step": 3576
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010999587798845835,
-      "loss": 0.9994,
+      "learning_rate": 0.00036644682605111296,
+      "loss": 4.5585,
       "step": 3577
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010993404781533388,
-      "loss": 0.963,
+      "learning_rate": 0.0003662407254740313,
+      "loss": 4.1136,
       "step": 3578
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010987221764220938,
-      "loss": 0.871,
+      "learning_rate": 0.0003660346248969497,
+      "loss": 4.4215,
       "step": 3579
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0001098103874690849,
-      "loss": 0.9256,
+      "learning_rate": 0.00036582852431986813,
+      "loss": 4.1625,
       "step": 3580
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010974855729596043,
-      "loss": 1.153,
+      "learning_rate": 0.0003656224237427865,
+      "loss": 4.3837,
       "step": 3581
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010968672712283593,
-      "loss": 0.9803,
+      "learning_rate": 0.0003654163231657049,
+      "loss": 4.1832,
       "step": 3582
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010962489694971145,
-      "loss": 0.8623,
+      "learning_rate": 0.00036521022258862324,
+      "loss": 4.1196,
       "step": 3583
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010956306677658697,
-      "loss": 0.806,
+      "learning_rate": 0.00036500412201154165,
+      "loss": 4.435,
       "step": 3584
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010950123660346248,
-      "loss": 0.986,
+      "learning_rate": 0.00036479802143446,
+      "loss": 4.296,
       "step": 3585
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010943940643033799,
-      "loss": 1.0548,
+      "learning_rate": 0.0003645919208573784,
+      "loss": 4.3019,
       "step": 3586
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010937757625721352,
-      "loss": 0.9531,
+      "learning_rate": 0.00036438582028029676,
+      "loss": 4.3063,
       "step": 3587
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010931574608408902,
-      "loss": 0.811,
+      "learning_rate": 0.00036417971970321517,
+      "loss": 4.079,
       "step": 3588
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010925391591096454,
-      "loss": 0.7422,
+      "learning_rate": 0.0003639736191261336,
+      "loss": 4.239,
       "step": 3589
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010919208573784007,
-      "loss": 0.8014,
+      "learning_rate": 0.0003637675185490519,
+      "loss": 3.9906,
       "step": 3590
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010913025556471557,
-      "loss": 1.1404,
+      "learning_rate": 0.00036356141797197033,
+      "loss": 4.4964,
       "step": 3591
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010906842539159109,
-      "loss": 0.8893,
+      "learning_rate": 0.0003633553173948887,
+      "loss": 4.4445,
       "step": 3592
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0001090065952184666,
-      "loss": 0.9405,
+      "learning_rate": 0.0003631492168178071,
+      "loss": 4.166,
       "step": 3593
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00010894476504534212,
-      "loss": 0.7626,
+      "learning_rate": 0.00036294311624072544,
+      "loss": 4.1784,
       "step": 3594
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010888293487221762,
-      "loss": 0.8145,
+      "learning_rate": 0.00036273701566364385,
+      "loss": 4.0959,
       "step": 3595
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010882110469909315,
-      "loss": 0.6417,
+      "learning_rate": 0.00036253091508656226,
+      "loss": 4.2517,
       "step": 3596
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010875927452596867,
-      "loss": 0.8161,
+      "learning_rate": 0.0003623248145094806,
+      "loss": 4.0979,
       "step": 3597
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010869744435284417,
-      "loss": 0.8391,
+      "learning_rate": 0.000362118713932399,
+      "loss": 4.257,
       "step": 3598
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0001086356141797197,
-      "loss": 0.9895,
+      "learning_rate": 0.00036191261335531737,
+      "loss": 4.0202,
       "step": 3599
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0001085737840065952,
-      "loss": 0.8809,
+      "learning_rate": 0.0003617065127782358,
+      "loss": 4.2162,
       "step": 3600
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010851195383347072,
-      "loss": 1.0176,
+      "learning_rate": 0.00036150041220115413,
+      "loss": 4.0788,
       "step": 3601
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010845012366034624,
-      "loss": 0.5425,
+      "learning_rate": 0.00036129431162407253,
+      "loss": 4.0035,
       "step": 3602
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010838829348722176,
-      "loss": 0.7205,
+      "learning_rate": 0.00036108821104699094,
+      "loss": 3.9209,
       "step": 3603
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010832646331409727,
-      "loss": 0.5854,
+      "learning_rate": 0.0003608821104699093,
+      "loss": 4.0792,
       "step": 3604
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010826463314097279,
-      "loss": 0.854,
+      "learning_rate": 0.0003606760098928277,
+      "loss": 4.2221,
       "step": 3605
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0001082028029678483,
-      "loss": 0.8426,
+      "learning_rate": 0.0003604699093157461,
+      "loss": 4.2138,
       "step": 3606
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010814097279472381,
-      "loss": 0.6911,
+      "learning_rate": 0.0003602638087386645,
+      "loss": 3.9808,
       "step": 3607
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010807914262159934,
-      "loss": 0.7362,
+      "learning_rate": 0.00036005770816158287,
+      "loss": 4.286,
       "step": 3608
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010801731244847484,
-      "loss": 1.3427,
+      "learning_rate": 0.0003598516075845013,
+      "loss": 4.5028,
       "step": 3609
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010795548227535036,
-      "loss": 1.0471,
+      "learning_rate": 0.0003596455070074196,
+      "loss": 4.2353,
       "step": 3610
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010789365210222587,
-      "loss": 0.6173,
+      "learning_rate": 0.00035943940643033803,
+      "loss": 3.8068,
       "step": 3611
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00010783182192910139,
-      "loss": 0.6174,
+      "learning_rate": 0.00035923330585325644,
+      "loss": 3.7772,
       "step": 3612
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010776999175597692,
-      "loss": 0.6617,
+      "learning_rate": 0.0003590272052761748,
+      "loss": 3.7642,
       "step": 3613
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010770816158285242,
-      "loss": 0.7814,
+      "learning_rate": 0.0003588211046990932,
+      "loss": 3.945,
       "step": 3614
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010764633140972794,
-      "loss": 0.4923,
+      "learning_rate": 0.00035861500412201155,
+      "loss": 3.143,
       "step": 3615
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010758450123660344,
-      "loss": 0.9003,
+      "learning_rate": 0.00035840890354492996,
+      "loss": 3.5583,
       "step": 3616
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010752267106347897,
-      "loss": 0.9167,
+      "learning_rate": 0.0003582028029678483,
+      "loss": 3.5837,
       "step": 3617
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010746084089035448,
-      "loss": 0.6365,
+      "learning_rate": 0.0003579967023907667,
+      "loss": 3.359,
       "step": 3618
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010739901071723,
-      "loss": 1.1644,
+      "learning_rate": 0.0003577906018136851,
+      "loss": 4.4598,
       "step": 3619
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010733718054410552,
-      "loss": 0.9657,
+      "learning_rate": 0.0003575845012366035,
+      "loss": 4.1168,
       "step": 3620
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010727535037098103,
-      "loss": 1.0774,
+      "learning_rate": 0.0003573784006595219,
+      "loss": 4.1852,
       "step": 3621
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010721352019785656,
-      "loss": 1.112,
+      "learning_rate": 0.00035717230008244024,
+      "loss": 4.433,
       "step": 3622
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010715169002473206,
-      "loss": 0.9265,
+      "learning_rate": 0.00035696619950535864,
+      "loss": 4.1676,
       "step": 3623
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010708985985160758,
-      "loss": 0.9188,
+      "learning_rate": 0.000356760098928277,
+      "loss": 4.1231,
       "step": 3624
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010702802967848308,
-      "loss": 1.0193,
+      "learning_rate": 0.0003565539983511954,
+      "loss": 4.4675,
       "step": 3625
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010696619950535861,
-      "loss": 1.0382,
+      "learning_rate": 0.00035634789777411375,
+      "loss": 4.467,
       "step": 3626
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010690436933223411,
-      "loss": 1.1375,
+      "learning_rate": 0.00035614179719703216,
+      "loss": 4.369,
       "step": 3627
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010684253915910963,
-      "loss": 1.2558,
+      "learning_rate": 0.00035593569661995057,
+      "loss": 4.6344,
       "step": 3628
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010678070898598516,
-      "loss": 0.9981,
+      "learning_rate": 0.0003557295960428689,
+      "loss": 4.3623,
       "step": 3629
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00010671887881286066,
-      "loss": 0.9096,
+      "learning_rate": 0.0003555234954657873,
+      "loss": 4.0455,
       "step": 3630
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010665704863973619,
-      "loss": 0.6124,
+      "learning_rate": 0.0003553173948887057,
+      "loss": 4.0276,
       "step": 3631
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0001065952184666117,
-      "loss": 0.7903,
+      "learning_rate": 0.0003551112943116241,
+      "loss": 4.1344,
       "step": 3632
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010653338829348721,
-      "loss": 1.0635,
+      "learning_rate": 0.00035490519373454244,
+      "loss": 4.9253,
       "step": 3633
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010647155812036272,
-      "loss": 0.9322,
+      "learning_rate": 0.00035469909315746084,
+      "loss": 4.1643,
       "step": 3634
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010640972794723825,
-      "loss": 0.9928,
+      "learning_rate": 0.00035449299258037925,
+      "loss": 4.3861,
       "step": 3635
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010634789777411376,
-      "loss": 0.8787,
+      "learning_rate": 0.0003542868920032976,
+      "loss": 4.3731,
       "step": 3636
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010628606760098928,
-      "loss": 0.9839,
+      "learning_rate": 0.000354080791426216,
+      "loss": 4.3804,
       "step": 3637
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0001062242374278648,
-      "loss": 0.788,
+      "learning_rate": 0.00035387469084913436,
+      "loss": 4.2411,
       "step": 3638
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0001061624072547403,
-      "loss": 0.6273,
+      "learning_rate": 0.00035366859027205277,
+      "loss": 4.0534,
       "step": 3639
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010610057708161583,
-      "loss": 0.5279,
+      "learning_rate": 0.0003534624896949711,
+      "loss": 4.1232,
       "step": 3640
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010603874690849133,
-      "loss": 0.5808,
+      "learning_rate": 0.00035325638911788953,
+      "loss": 3.9911,
       "step": 3641
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010597691673536685,
-      "loss": 0.6929,
+      "learning_rate": 0.0003530502885408079,
+      "loss": 3.9408,
       "step": 3642
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010591508656224235,
-      "loss": 0.7645,
+      "learning_rate": 0.0003528441879637263,
+      "loss": 4.178,
       "step": 3643
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010585325638911788,
-      "loss": 0.7972,
+      "learning_rate": 0.0003526380873866447,
+      "loss": 4.2293,
       "step": 3644
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0001057914262159934,
-      "loss": 0.5677,
+      "learning_rate": 0.00035243198680956305,
+      "loss": 4.1888,
       "step": 3645
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010572959604286891,
-      "loss": 0.6504,
+      "learning_rate": 0.00035222588623248145,
+      "loss": 4.1129,
       "step": 3646
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010566776586974443,
-      "loss": 0.9745,
+      "learning_rate": 0.0003520197856553998,
+      "loss": 4.1418,
       "step": 3647
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00010560593569661993,
-      "loss": 0.9127,
+      "learning_rate": 0.0003518136850783182,
+      "loss": 3.975,
       "step": 3648
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010554410552349546,
-      "loss": 0.6573,
+      "learning_rate": 0.00035160758450123657,
+      "loss": 4.4257,
       "step": 3649
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010548227535037097,
-      "loss": 0.4833,
+      "learning_rate": 0.00035140148392415497,
+      "loss": 3.9152,
       "step": 3650
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010542044517724648,
-      "loss": 0.9949,
+      "learning_rate": 0.0003511953833470734,
+      "loss": 4.6934,
       "step": 3651
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010535861500412201,
-      "loss": 0.8461,
+      "learning_rate": 0.00035098928276999173,
+      "loss": 4.3512,
       "step": 3652
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010529678483099752,
-      "loss": 0.6781,
+      "learning_rate": 0.0003507831821929102,
+      "loss": 4.2817,
       "step": 3653
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010523495465787303,
-      "loss": 0.5278,
+      "learning_rate": 0.00035057708161582854,
+      "loss": 3.7464,
       "step": 3654
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010517312448474855,
-      "loss": 0.5457,
+      "learning_rate": 0.00035037098103874695,
+      "loss": 3.9491,
       "step": 3655
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010511129431162407,
-      "loss": 0.7186,
+      "learning_rate": 0.0003501648804616653,
+      "loss": 4.0878,
       "step": 3656
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010504946413849957,
-      "loss": 0.645,
+      "learning_rate": 0.0003499587798845837,
+      "loss": 4.1872,
       "step": 3657
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0001049876339653751,
-      "loss": 0.82,
+      "learning_rate": 0.00034975267930750206,
+      "loss": 3.8181,
       "step": 3658
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0001049258037922506,
-      "loss": 0.6596,
+      "learning_rate": 0.00034954657873042047,
+      "loss": 3.9885,
       "step": 3659
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010486397361912612,
-      "loss": 1.1496,
+      "learning_rate": 0.0003493404781533389,
+      "loss": 4.2296,
       "step": 3660
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010480214344600165,
-      "loss": 0.811,
+      "learning_rate": 0.00034913437757625723,
+      "loss": 4.1406,
       "step": 3661
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010474031327287715,
-      "loss": 0.6877,
+      "learning_rate": 0.00034892827699917564,
+      "loss": 3.8549,
       "step": 3662
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010467848309975267,
-      "loss": 0.7998,
+      "learning_rate": 0.000348722176422094,
+      "loss": 3.8693,
       "step": 3663
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00010461665292662819,
-      "loss": 0.6826,
+      "learning_rate": 0.0003485160758450124,
+      "loss": 3.7375,
       "step": 3664
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0001045548227535037,
-      "loss": 1.0569,
+      "learning_rate": 0.00034830997526793075,
+      "loss": 4.0972,
       "step": 3665
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0001044929925803792,
-      "loss": 0.6775,
+      "learning_rate": 0.00034810387469084915,
+      "loss": 3.646,
       "step": 3666
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010443116240725474,
-      "loss": 0.7254,
+      "learning_rate": 0.00034789777411376756,
+      "loss": 3.8753,
       "step": 3667
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010436933223413025,
-      "loss": 0.8499,
+      "learning_rate": 0.0003476916735366859,
+      "loss": 3.5197,
       "step": 3668
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010430750206100575,
-      "loss": 1.1334,
+      "learning_rate": 0.0003474855729596043,
+      "loss": 4.305,
       "step": 3669
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010424567188788129,
-      "loss": 1.3901,
+      "learning_rate": 0.00034727947238252267,
+      "loss": 4.5513,
       "step": 3670
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010418384171475679,
-      "loss": 0.9059,
+      "learning_rate": 0.0003470733718054411,
+      "loss": 4.0203,
       "step": 3671
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0001041220115416323,
-      "loss": 0.9797,
+      "learning_rate": 0.00034686727122835943,
+      "loss": 4.0658,
       "step": 3672
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010406018136850782,
-      "loss": 0.8441,
+      "learning_rate": 0.00034666117065127784,
+      "loss": 4.0953,
       "step": 3673
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010399835119538334,
-      "loss": 0.6684,
+      "learning_rate": 0.0003464550700741962,
+      "loss": 4.2018,
       "step": 3674
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010393652102225884,
-      "loss": 1.0553,
+      "learning_rate": 0.0003462489694971146,
+      "loss": 4.8505,
       "step": 3675
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010387469084913437,
-      "loss": 0.7871,
+      "learning_rate": 0.000346042868920033,
+      "loss": 4.1872,
       "step": 3676
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010381286067600989,
-      "loss": 1.0444,
+      "learning_rate": 0.00034583676834295136,
+      "loss": 4.406,
       "step": 3677
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010375103050288539,
-      "loss": 0.7706,
+      "learning_rate": 0.00034563066776586976,
+      "loss": 4.1894,
       "step": 3678
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010368920032976092,
-      "loss": 0.9763,
+      "learning_rate": 0.0003454245671887881,
+      "loss": 4.0363,
       "step": 3679
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010362737015663642,
-      "loss": 0.9676,
+      "learning_rate": 0.0003452184666117065,
+      "loss": 4.238,
       "step": 3680
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010356553998351194,
-      "loss": 0.7273,
+      "learning_rate": 0.0003450123660346249,
+      "loss": 4.2913,
       "step": 3681
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010350370981038746,
-      "loss": 0.9525,
+      "learning_rate": 0.0003448062654575433,
+      "loss": 4.1583,
       "step": 3682
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00010344187963726297,
-      "loss": 0.7204,
+      "learning_rate": 0.0003446001648804617,
+      "loss": 4.2548,
       "step": 3683
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0001033800494641385,
-      "loss": 0.9597,
+      "learning_rate": 0.00034439406430338004,
+      "loss": 4.3861,
       "step": 3684
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.000103318219291014,
-      "loss": 0.933,
+      "learning_rate": 0.00034418796372629845,
+      "loss": 4.3047,
       "step": 3685
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010325638911788952,
-      "loss": 0.8609,
+      "learning_rate": 0.0003439818631492168,
+      "loss": 4.4504,
       "step": 3686
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010319455894476503,
-      "loss": 0.7305,
+      "learning_rate": 0.0003437757625721352,
+      "loss": 4.3347,
       "step": 3687
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010313272877164056,
-      "loss": 0.8329,
+      "learning_rate": 0.00034356966199505356,
+      "loss": 4.3435,
       "step": 3688
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010307089859851606,
-      "loss": 0.7676,
+      "learning_rate": 0.00034336356141797197,
+      "loss": 4.0876,
       "step": 3689
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010300906842539159,
-      "loss": 1.0242,
+      "learning_rate": 0.0003431574608408903,
+      "loss": 4.1959,
       "step": 3690
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010294723825226709,
-      "loss": 0.7415,
+      "learning_rate": 0.0003429513602638087,
+      "loss": 4.0557,
       "step": 3691
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010288540807914261,
-      "loss": 0.8097,
+      "learning_rate": 0.00034274525968672713,
+      "loss": 4.1933,
       "step": 3692
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010282357790601814,
-      "loss": 0.7476,
+      "learning_rate": 0.0003425391591096455,
+      "loss": 3.9794,
       "step": 3693
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010276174773289364,
-      "loss": 0.9283,
+      "learning_rate": 0.0003423330585325639,
+      "loss": 4.2046,
       "step": 3694
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010269991755976916,
-      "loss": 0.8518,
+      "learning_rate": 0.00034212695795548224,
+      "loss": 4.257,
       "step": 3695
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010263808738664466,
-      "loss": 0.5327,
+      "learning_rate": 0.00034192085737840065,
+      "loss": 3.9712,
       "step": 3696
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010257625721352019,
-      "loss": 0.8108,
+      "learning_rate": 0.000341714756801319,
+      "loss": 4.3585,
       "step": 3697
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0001025144270403957,
-      "loss": 0.6647,
+      "learning_rate": 0.0003415086562242374,
+      "loss": 4.0659,
       "step": 3698
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010245259686727123,
-      "loss": 0.641,
+      "learning_rate": 0.00034130255564715587,
+      "loss": 4.0655,
       "step": 3699
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010239076669414674,
-      "loss": 0.7637,
+      "learning_rate": 0.0003410964550700742,
+      "loss": 4.2117,
       "step": 3700
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00010232893652102224,
-      "loss": 1.0368,
+      "learning_rate": 0.00034089035449299263,
+      "loss": 4.484,
       "step": 3701
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010226710634789778,
-      "loss": 0.7253,
+      "learning_rate": 0.000340684253915911,
+      "loss": 3.9193,
       "step": 3702
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010220527617477328,
-      "loss": 0.6837,
+      "learning_rate": 0.0003404781533388294,
+      "loss": 4.0548,
       "step": 3703
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0001021434460016488,
-      "loss": 0.7409,
+      "learning_rate": 0.00034027205276174774,
+      "loss": 4.1632,
       "step": 3704
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010208161582852431,
-      "loss": 0.6376,
+      "learning_rate": 0.00034006595218466615,
+      "loss": 4.0142,
       "step": 3705
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010201978565539983,
-      "loss": 0.9984,
+      "learning_rate": 0.00033985985160758455,
+      "loss": 4.0628,
       "step": 3706
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010195795548227534,
-      "loss": 0.6658,
+      "learning_rate": 0.0003396537510305029,
+      "loss": 4.2197,
       "step": 3707
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010189612530915086,
-      "loss": 0.7051,
+      "learning_rate": 0.0003394476504534213,
+      "loss": 4.0214,
       "step": 3708
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010183429513602638,
-      "loss": 0.6733,
+      "learning_rate": 0.00033924154987633967,
+      "loss": 4.1325,
       "step": 3709
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010177246496290188,
-      "loss": 0.9357,
+      "learning_rate": 0.00033903544929925807,
+      "loss": 4.1346,
       "step": 3710
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010171063478977741,
-      "loss": 0.9878,
+      "learning_rate": 0.0003388293487221764,
+      "loss": 4.122,
       "step": 3711
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010164880461665291,
-      "loss": 0.4639,
+      "learning_rate": 0.00033862324814509483,
+      "loss": 5.1289,
       "step": 3712
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010158697444352843,
-      "loss": 0.6799,
+      "learning_rate": 0.0003384171475680132,
+      "loss": 3.9975,
       "step": 3713
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010152514427040395,
-      "loss": 0.8157,
+      "learning_rate": 0.0003382110469909316,
+      "loss": 3.9169,
       "step": 3714
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010146331409727946,
-      "loss": 0.8153,
+      "learning_rate": 0.00033800494641385,
+      "loss": 4.2691,
       "step": 3715
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010140148392415498,
-      "loss": 0.5227,
+      "learning_rate": 0.00033779884583676835,
+      "loss": 3.5229,
       "step": 3716
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0001013396537510305,
-      "loss": 0.6815,
+      "learning_rate": 0.00033759274525968676,
+      "loss": 3.5943,
       "step": 3717
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010127782357790601,
-      "loss": 0.7376,
+      "learning_rate": 0.0003373866446826051,
+      "loss": 3.3199,
       "step": 3718
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00010121599340478152,
-      "loss": 1.1916,
+      "learning_rate": 0.0003371805441055235,
+      "loss": 4.5585,
       "step": 3719
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010115416323165705,
-      "loss": 1.1366,
+      "learning_rate": 0.00033697444352844187,
+      "loss": 4.6449,
       "step": 3720
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010109233305853255,
-      "loss": 0.815,
+      "learning_rate": 0.0003367683429513603,
+      "loss": 4.0208,
       "step": 3721
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010103050288540807,
-      "loss": 0.8922,
+      "learning_rate": 0.0003365622423742787,
+      "loss": 4.1108,
       "step": 3722
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0001009686727122836,
-      "loss": 1.3352,
+      "learning_rate": 0.00033635614179719703,
+      "loss": 4.9957,
       "step": 3723
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0001009068425391591,
-      "loss": 0.8772,
+      "learning_rate": 0.00033615004122011544,
+      "loss": 4.1105,
       "step": 3724
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010084501236603462,
-      "loss": 0.782,
+      "learning_rate": 0.0003359439406430338,
+      "loss": 4.023,
       "step": 3725
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010078318219291013,
-      "loss": 0.6514,
+      "learning_rate": 0.0003357378400659522,
+      "loss": 4.2332,
       "step": 3726
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010072135201978565,
-      "loss": 0.5486,
+      "learning_rate": 0.00033553173948887055,
+      "loss": 4.1649,
       "step": 3727
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010065952184666115,
-      "loss": 0.7977,
+      "learning_rate": 0.00033532563891178896,
+      "loss": 4.0826,
       "step": 3728
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010059769167353668,
-      "loss": 0.9642,
+      "learning_rate": 0.0003351195383347073,
+      "loss": 4.3062,
       "step": 3729
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010053586150041219,
-      "loss": 0.9947,
+      "learning_rate": 0.0003349134377576257,
+      "loss": 4.1424,
       "step": 3730
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0001004740313272877,
-      "loss": 0.7633,
+      "learning_rate": 0.0003347073371805441,
+      "loss": 3.8809,
       "step": 3731
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010041220115416323,
-      "loss": 0.8767,
+      "learning_rate": 0.0003345012366034625,
+      "loss": 4.1399,
       "step": 3732
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010035037098103874,
-      "loss": 1.0048,
+      "learning_rate": 0.0003342951360263809,
+      "loss": 4.1754,
       "step": 3733
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010028854080791425,
-      "loss": 1.1582,
+      "learning_rate": 0.00033408903544929924,
+      "loss": 4.3943,
       "step": 3734
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010022671063478977,
-      "loss": 0.6784,
+      "learning_rate": 0.00033388293487221764,
+      "loss": 4.0475,
       "step": 3735
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010016488046166528,
-      "loss": 0.977,
+      "learning_rate": 0.000333676834295136,
+      "loss": 4.2241,
       "step": 3736
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00010010305028854079,
-      "loss": 0.6059,
+      "learning_rate": 0.0003334707337180544,
+      "loss": 4.1113,
       "step": 3737
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00010004122011541632,
-      "loss": 0.8456,
+      "learning_rate": 0.0003332646331409728,
+      "loss": 4.2762,
       "step": 3738
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.997938994229183e-05,
-      "loss": 0.9081,
+      "learning_rate": 0.00033305853256389116,
+      "loss": 4.274,
       "step": 3739
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.991755976916734e-05,
-      "loss": 0.8143,
+      "learning_rate": 0.00033285243198680957,
+      "loss": 4.2894,
       "step": 3740
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.985572959604287e-05,
-      "loss": 0.6713,
+      "learning_rate": 0.0003326463314097279,
+      "loss": 4.1549,
       "step": 3741
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.979389942291837e-05,
-      "loss": 0.7606,
+      "learning_rate": 0.00033244023083264633,
+      "loss": 4.2425,
       "step": 3742
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.97320692497939e-05,
-      "loss": 0.5626,
+      "learning_rate": 0.0003322341302555647,
+      "loss": 4.1323,
       "step": 3743
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.96702390766694e-05,
-      "loss": 0.891,
+      "learning_rate": 0.0003320280296784831,
+      "loss": 4.278,
       "step": 3744
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.960840890354492e-05,
-      "loss": 0.6974,
+      "learning_rate": 0.00033182192910140144,
+      "loss": 4.2883,
       "step": 3745
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.954657873042042e-05,
-      "loss": 0.818,
+      "learning_rate": 0.0003316158285243199,
+      "loss": 4.3033,
       "step": 3746
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.948474855729595e-05,
-      "loss": 0.6353,
+      "learning_rate": 0.0003314097279472383,
+      "loss": 3.8966,
       "step": 3747
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.942291838417147e-05,
-      "loss": 0.7257,
+      "learning_rate": 0.00033120362737015666,
+      "loss": 4.07,
       "step": 3748
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.936108821104697e-05,
-      "loss": 0.6591,
+      "learning_rate": 0.00033099752679307507,
+      "loss": 4.0953,
       "step": 3749
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.92992580379225e-05,
-      "loss": 0.7392,
+      "learning_rate": 0.0003307914262159934,
+      "loss": 4.3498,
       "step": 3750
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.9237427864798e-05,
-      "loss": 0.6865,
+      "learning_rate": 0.0003305853256389118,
+      "loss": 4.2417,
       "step": 3751
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.917559769167354e-05,
-      "loss": 0.6424,
+      "learning_rate": 0.0003303792250618302,
+      "loss": 4.1761,
       "step": 3752
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.911376751854904e-05,
-      "loss": 0.7472,
+      "learning_rate": 0.0003301731244847486,
+      "loss": 4.017,
       "step": 3753
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.905193734542456e-05,
-      "loss": 0.8368,
+      "learning_rate": 0.000329967023907667,
+      "loss": 3.9064,
       "step": 3754
     },
     {
       "epoch": 2.1,
-      "learning_rate": 9.899010717230009e-05,
-      "loss": 0.8038,
+      "learning_rate": 0.00032976092333058534,
+      "loss": 4.352,
       "step": 3755
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.892827699917559e-05,
-      "loss": 0.7289,
+      "learning_rate": 0.00032955482275350375,
+      "loss": 4.0681,
       "step": 3756
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.88664468260511e-05,
-      "loss": 0.6893,
+      "learning_rate": 0.0003293487221764221,
+      "loss": 4.1873,
       "step": 3757
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.880461665292662e-05,
-      "loss": 1.5788,
+      "learning_rate": 0.0003291426215993405,
+      "loss": 4.3458,
       "step": 3758
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.874278647980214e-05,
-      "loss": 0.8699,
+      "learning_rate": 0.00032893652102225886,
+      "loss": 4.5754,
       "step": 3759
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.868095630667764e-05,
-      "loss": 0.9759,
+      "learning_rate": 0.00032873042044517727,
+      "loss": 4.1492,
       "step": 3760
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.861912613355317e-05,
-      "loss": 0.6857,
+      "learning_rate": 0.0003285243198680956,
+      "loss": 3.9552,
       "step": 3761
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.855729596042868e-05,
-      "loss": 0.6974,
+      "learning_rate": 0.00032831821929101403,
+      "loss": 3.9674,
       "step": 3762
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.849546578730419e-05,
-      "loss": 0.7134,
+      "learning_rate": 0.00032811211871393243,
+      "loss": 3.8928,
       "step": 3763
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.843363561417972e-05,
-      "loss": 0.8554,
+      "learning_rate": 0.0003279060181368508,
+      "loss": 3.7404,
       "step": 3764
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.837180544105523e-05,
-      "loss": 0.5514,
+      "learning_rate": 0.0003276999175597692,
+      "loss": 3.686,
       "step": 3765
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.830997526793074e-05,
-      "loss": 0.6494,
+      "learning_rate": 0.00032749381698268755,
+      "loss": 3.8057,
       "step": 3766
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.824814509480626e-05,
-      "loss": 1.1257,
+      "learning_rate": 0.00032728771640560595,
+      "loss": 4.0004,
       "step": 3767
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.818631492168177e-05,
-      "loss": 0.6081,
+      "learning_rate": 0.0003270816158285243,
+      "loss": 3.5113,
       "step": 3768
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.812448474855728e-05,
-      "loss": 1.1838,
+      "learning_rate": 0.0003268755152514427,
+      "loss": 4.5058,
       "step": 3769
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.806265457543281e-05,
-      "loss": 0.9657,
+      "learning_rate": 0.0003266694146743611,
+      "loss": 4.0737,
       "step": 3770
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.800082440230832e-05,
-      "loss": 0.9871,
+      "learning_rate": 0.00032646331409727947,
+      "loss": 4.2785,
       "step": 3771
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.793899422918383e-05,
-      "loss": 1.2063,
+      "learning_rate": 0.0003262572135201979,
+      "loss": 4.4197,
       "step": 3772
     },
     {
       "epoch": 2.11,
-      "learning_rate": 9.787716405605936e-05,
-      "loss": 0.8225,
+      "learning_rate": 0.00032605111294311623,
+      "loss": 4.1764,
       "step": 3773
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.781533388293486e-05,
-      "loss": 0.727,
+      "learning_rate": 0.00032584501236603464,
+      "loss": 4.003,
       "step": 3774
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.775350370981038e-05,
-      "loss": 0.7561,
+      "learning_rate": 0.000325638911788953,
+      "loss": 4.0184,
       "step": 3775
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.76916735366859e-05,
-      "loss": 0.8661,
+      "learning_rate": 0.0003254328112118714,
+      "loss": 4.2844,
       "step": 3776
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.762984336356141e-05,
-      "loss": 1.0145,
+      "learning_rate": 0.00032522671063478975,
+      "loss": 4.2929,
       "step": 3777
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.756801319043691e-05,
-      "loss": 0.8968,
+      "learning_rate": 0.00032502061005770816,
+      "loss": 4.3295,
       "step": 3778
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.750618301731244e-05,
-      "loss": 0.7627,
+      "learning_rate": 0.00032481450948062656,
+      "loss": 3.911,
       "step": 3779
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.744435284418796e-05,
-      "loss": 0.6281,
+      "learning_rate": 0.0003246084089035449,
+      "loss": 3.95,
       "step": 3780
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.738252267106346e-05,
-      "loss": 0.706,
+      "learning_rate": 0.0003244023083264633,
+      "loss": 4.0873,
       "step": 3781
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.7320692497939e-05,
-      "loss": 1.347,
+      "learning_rate": 0.0003241962077493817,
+      "loss": 4.7243,
       "step": 3782
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.72588623248145e-05,
-      "loss": 0.4964,
+      "learning_rate": 0.0003239901071723001,
+      "loss": 3.7764,
       "step": 3783
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.719703215169001e-05,
-      "loss": 0.6322,
+      "learning_rate": 0.00032378400659521843,
+      "loss": 4.0955,
       "step": 3784
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.713520197856553e-05,
-      "loss": 0.7905,
+      "learning_rate": 0.00032357790601813684,
+      "loss": 4.3724,
       "step": 3785
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.707337180544105e-05,
-      "loss": 0.6993,
+      "learning_rate": 0.00032337180544105525,
+      "loss": 4.1733,
       "step": 3786
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.701154163231656e-05,
-      "loss": 0.7376,
+      "learning_rate": 0.0003231657048639736,
+      "loss": 4.2521,
       "step": 3787
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.694971145919208e-05,
-      "loss": 0.817,
+      "learning_rate": 0.000322959604286892,
+      "loss": 4.3336,
       "step": 3788
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.68878812860676e-05,
-      "loss": 0.7222,
+      "learning_rate": 0.00032275350370981036,
+      "loss": 3.9601,
       "step": 3789
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.68260511129431e-05,
-      "loss": 0.8335,
+      "learning_rate": 0.00032254740313272876,
+      "loss": 4.5063,
       "step": 3790
     },
     {
       "epoch": 2.12,
-      "learning_rate": 9.676422093981863e-05,
-      "loss": 0.7472,
+      "learning_rate": 0.0003223413025556471,
+      "loss": 4.1958,
       "step": 3791
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.670239076669413e-05,
-      "loss": 1.0913,
+      "learning_rate": 0.0003221352019785655,
+      "loss": 4.6148,
       "step": 3792
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.664056059356965e-05,
-      "loss": 0.7216,
+      "learning_rate": 0.00032192910140148393,
+      "loss": 4.2743,
       "step": 3793
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.657873042044517e-05,
-      "loss": 0.7249,
+      "learning_rate": 0.00032172300082440234,
+      "loss": 4.1043,
       "step": 3794
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.651690024732068e-05,
-      "loss": 0.7877,
+      "learning_rate": 0.00032151690024732074,
+      "loss": 3.9861,
       "step": 3795
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.645507007419621e-05,
-      "loss": 1.0259,
+      "learning_rate": 0.0003213107996702391,
+      "loss": 4.5033,
       "step": 3796
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.639323990107172e-05,
-      "loss": 0.7594,
+      "learning_rate": 0.0003211046990931575,
+      "loss": 4.3829,
       "step": 3797
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.633140972794723e-05,
-      "loss": 0.7132,
+      "learning_rate": 0.00032089859851607586,
+      "loss": 4.158,
       "step": 3798
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.626957955482273e-05,
-      "loss": 0.5905,
+      "learning_rate": 0.00032069249793899426,
+      "loss": 4.2529,
       "step": 3799
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.620774938169827e-05,
-      "loss": 0.8592,
+      "learning_rate": 0.0003204863973619126,
+      "loss": 4.2216,
       "step": 3800
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.614591920857377e-05,
-      "loss": 0.7598,
+      "learning_rate": 0.000320280296784831,
+      "loss": 3.9512,
       "step": 3801
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.608408903544928e-05,
-      "loss": 0.8512,
+      "learning_rate": 0.00032007419620774943,
+      "loss": 4.2462,
       "step": 3802
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.602225886232481e-05,
-      "loss": 0.6166,
+      "learning_rate": 0.0003198680956306678,
+      "loss": 4.0295,
       "step": 3803
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.596042868920032e-05,
-      "loss": 0.8629,
+      "learning_rate": 0.0003196619950535862,
+      "loss": 4.1878,
       "step": 3804
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.589859851607585e-05,
-      "loss": 0.4908,
+      "learning_rate": 0.00031945589447650454,
+      "loss": 4.0144,
       "step": 3805
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.583676834295135e-05,
-      "loss": 0.63,
+      "learning_rate": 0.00031924979389942295,
+      "loss": 4.2387,
       "step": 3806
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.577493816982687e-05,
-      "loss": 0.493,
+      "learning_rate": 0.0003190436933223413,
+      "loss": 3.769,
       "step": 3807
     },
     {
       "epoch": 2.13,
-      "learning_rate": 9.571310799670237e-05,
-      "loss": 0.5088,
+      "learning_rate": 0.0003188375927452597,
+      "loss": 3.8161,
       "step": 3808
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.56512778235779e-05,
-      "loss": 0.7085,
+      "learning_rate": 0.00031863149216817806,
+      "loss": 3.7658,
       "step": 3809
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.55894476504534e-05,
-      "loss": 0.5936,
+      "learning_rate": 0.00031842539159109646,
+      "loss": 3.9501,
       "step": 3810
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.552761747732892e-05,
-      "loss": 0.4903,
+      "learning_rate": 0.00031821929101401487,
+      "loss": 3.8618,
       "step": 3811
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.546578730420445e-05,
-      "loss": 0.3999,
+      "learning_rate": 0.0003180131904369332,
+      "loss": 3.3782,
       "step": 3812
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.540395713107995e-05,
-      "loss": 0.5083,
+      "learning_rate": 0.00031780708985985163,
+      "loss": 4.4266,
       "step": 3813
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.534212695795548e-05,
-      "loss": 0.6137,
+      "learning_rate": 0.00031760098928277,
+      "loss": 3.6482,
       "step": 3814
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.528029678483099e-05,
-      "loss": 0.449,
+      "learning_rate": 0.0003173948887056884,
+      "loss": 3.4938,
       "step": 3815
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.52184666117065e-05,
-      "loss": 0.4703,
+      "learning_rate": 0.00031718878812860674,
+      "loss": 3.5221,
       "step": 3816
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.5156636438582e-05,
-      "loss": 0.6464,
+      "learning_rate": 0.00031698268755152515,
+      "loss": 3.2281,
       "step": 3817
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.509480626545754e-05,
-      "loss": 1.8582,
+      "learning_rate": 0.00031677658697444356,
+      "loss": 3.5781,
       "step": 3818
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.503297609233305e-05,
-      "loss": 1.2814,
+      "learning_rate": 0.0003165704863973619,
+      "loss": 4.3927,
       "step": 3819
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.497114591920857e-05,
-      "loss": 1.1357,
+      "learning_rate": 0.0003163643858202803,
+      "loss": 4.317,
       "step": 3820
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.490931574608409e-05,
-      "loss": 1.1002,
+      "learning_rate": 0.00031615828524319867,
+      "loss": 4.4545,
       "step": 3821
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.484748557295959e-05,
-      "loss": 0.8158,
+      "learning_rate": 0.0003159521846661171,
+      "loss": 4.1876,
       "step": 3822
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.478565539983512e-05,
-      "loss": 0.8858,
+      "learning_rate": 0.0003157460840890354,
+      "loss": 4.4242,
       "step": 3823
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.472382522671062e-05,
-      "loss": 0.8931,
+      "learning_rate": 0.00031553998351195383,
+      "loss": 4.2528,
       "step": 3824
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.466199505358614e-05,
-      "loss": 1.3111,
+      "learning_rate": 0.00031533388293487224,
+      "loss": 4.3391,
       "step": 3825
     },
     {
       "epoch": 2.14,
-      "learning_rate": 9.460016488046167e-05,
-      "loss": 0.587,
+      "learning_rate": 0.0003151277823577906,
+      "loss": 4.0592,
       "step": 3826
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.453833470733717e-05,
-      "loss": 0.981,
+      "learning_rate": 0.000314921681780709,
+      "loss": 4.736,
       "step": 3827
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.447650453421269e-05,
-      "loss": 0.7606,
+      "learning_rate": 0.00031471558120362735,
+      "loss": 4.1938,
       "step": 3828
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.44146743610882e-05,
-      "loss": 0.8735,
+      "learning_rate": 0.00031450948062654576,
+      "loss": 4.2883,
       "step": 3829
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.435284418796372e-05,
-      "loss": 0.8937,
+      "learning_rate": 0.0003143033800494641,
+      "loss": 4.1106,
       "step": 3830
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.429101401483923e-05,
-      "loss": 0.9434,
+      "learning_rate": 0.0003140972794723825,
+      "loss": 4.4614,
       "step": 3831
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.422918384171476e-05,
-      "loss": 0.7974,
+      "learning_rate": 0.00031389117889530087,
+      "loss": 3.9693,
       "step": 3832
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.416735366859026e-05,
-      "loss": 0.6284,
+      "learning_rate": 0.0003136850783182193,
+      "loss": 4.0741,
       "step": 3833
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.410552349546577e-05,
-      "loss": 0.7072,
+      "learning_rate": 0.0003134789777411377,
+      "loss": 4.1548,
       "step": 3834
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.40436933223413e-05,
-      "loss": 0.6145,
+      "learning_rate": 0.00031327287716405604,
+      "loss": 4.186,
       "step": 3835
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.398186314921681e-05,
-      "loss": 0.6098,
+      "learning_rate": 0.00031306677658697444,
+      "loss": 4.0764,
       "step": 3836
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.392003297609232e-05,
-      "loss": 0.6779,
+      "learning_rate": 0.0003128606760098928,
+      "loss": 4.0459,
       "step": 3837
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.385820280296784e-05,
-      "loss": 0.6018,
+      "learning_rate": 0.0003126545754328112,
+      "loss": 4.1743,
       "step": 3838
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.379637262984336e-05,
-      "loss": 0.6414,
+      "learning_rate": 0.00031244847485572955,
+      "loss": 3.9454,
       "step": 3839
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.373454245671886e-05,
-      "loss": 0.8067,
+      "learning_rate": 0.000312242374278648,
+      "loss": 4.226,
       "step": 3840
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.367271228359439e-05,
-      "loss": 0.8342,
+      "learning_rate": 0.0003120362737015664,
+      "loss": 4.0656,
       "step": 3841
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.361088211046991e-05,
-      "loss": 0.837,
+      "learning_rate": 0.0003118301731244848,
+      "loss": 4.1275,
       "step": 3842
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.354905193734541e-05,
-      "loss": 0.6352,
+      "learning_rate": 0.0003116240725474032,
+      "loss": 4.1272,
       "step": 3843
     },
     {
       "epoch": 2.15,
-      "learning_rate": 9.348722176422094e-05,
-      "loss": 0.5188,
+      "learning_rate": 0.00031141797197032153,
+      "loss": 4.0491,
       "step": 3844
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.342539159109644e-05,
-      "loss": 0.5758,
+      "learning_rate": 0.00031121187139323994,
+      "loss": 3.8922,
       "step": 3845
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.336356141797196e-05,
-      "loss": 0.613,
+      "learning_rate": 0.0003110057708161583,
+      "loss": 4.1425,
       "step": 3846
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.330173124484748e-05,
-      "loss": 0.8749,
+      "learning_rate": 0.0003107996702390767,
+      "loss": 4.2908,
       "step": 3847
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.3239901071723e-05,
-      "loss": 0.8672,
+      "learning_rate": 0.00031059356966199505,
+      "loss": 4.4668,
       "step": 3848
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.31780708985985e-05,
-      "loss": 0.6558,
+      "learning_rate": 0.00031038746908491346,
+      "loss": 3.9677,
       "step": 3849
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.311624072547403e-05,
-      "loss": 0.5274,
+      "learning_rate": 0.00031018136850783186,
+      "loss": 3.9712,
       "step": 3850
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.305441055234954e-05,
-      "loss": 0.4864,
+      "learning_rate": 0.0003099752679307502,
+      "loss": 3.9558,
       "step": 3851
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.299258037922505e-05,
-      "loss": 0.7488,
+      "learning_rate": 0.0003097691673536686,
+      "loss": 4.0367,
       "step": 3852
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.293075020610058e-05,
-      "loss": 0.6068,
+      "learning_rate": 0.000309563066776587,
+      "loss": 3.8838,
       "step": 3853
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.286892003297608e-05,
-      "loss": 0.5794,
+      "learning_rate": 0.0003093569661995054,
+      "loss": 3.8805,
       "step": 3854
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.28070898598516e-05,
-      "loss": 0.801,
+      "learning_rate": 0.00030915086562242374,
+      "loss": 4.2344,
       "step": 3855
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.274525968672711e-05,
-      "loss": 0.6759,
+      "learning_rate": 0.00030894476504534214,
+      "loss": 3.9473,
       "step": 3856
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.268342951360263e-05,
-      "loss": 0.689,
+      "learning_rate": 0.00030873866446826055,
+      "loss": 4.0824,
       "step": 3857
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.262159934047816e-05,
-      "loss": 0.7509,
+      "learning_rate": 0.0003085325638911789,
+      "loss": 3.9773,
       "step": 3858
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.255976916735366e-05,
-      "loss": 0.7839,
+      "learning_rate": 0.0003083264633140973,
+      "loss": 4.2702,
       "step": 3859
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.249793899422918e-05,
-      "loss": 0.7896,
+      "learning_rate": 0.00030812036273701566,
+      "loss": 4.087,
       "step": 3860
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.243610882110468e-05,
-      "loss": 0.6097,
+      "learning_rate": 0.00030791426215993407,
+      "loss": 3.8625,
       "step": 3861
     },
     {
       "epoch": 2.16,
-      "learning_rate": 9.237427864798021e-05,
-      "loss": 0.4065,
+      "learning_rate": 0.0003077081615828524,
+      "loss": 3.2908,
       "step": 3862
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.231244847485572e-05,
-      "loss": 0.3825,
+      "learning_rate": 0.0003075020610057708,
+      "loss": 3.4741,
       "step": 3863
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.225061830173123e-05,
-      "loss": 0.7447,
+      "learning_rate": 0.0003072959604286892,
+      "loss": 3.8052,
       "step": 3864
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.218878812860675e-05,
-      "loss": 0.6771,
+      "learning_rate": 0.0003070898598516076,
+      "loss": 3.6967,
       "step": 3865
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.212695795548226e-05,
-      "loss": 0.5546,
+      "learning_rate": 0.000306883759274526,
+      "loss": 3.5192,
       "step": 3866
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.20651277823578e-05,
-      "loss": 1.4525,
+      "learning_rate": 0.00030667765869744435,
+      "loss": 4.0731,
       "step": 3867
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.20032976092333e-05,
-      "loss": 0.9018,
+      "learning_rate": 0.00030647155812036275,
+      "loss": 3.9215,
       "step": 3868
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.194146743610881e-05,
-      "loss": 1.2456,
+      "learning_rate": 0.0003062654575432811,
+      "loss": 4.4112,
       "step": 3869
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.187963726298432e-05,
-      "loss": 1.4193,
+      "learning_rate": 0.0003060593569661995,
+      "loss": 4.559,
       "step": 3870
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.181780708985985e-05,
-      "loss": 1.0643,
+      "learning_rate": 0.00030585325638911786,
+      "loss": 4.2582,
       "step": 3871
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.175597691673535e-05,
-      "loss": 1.0649,
+      "learning_rate": 0.00030564715581203627,
+      "loss": 4.3368,
       "step": 3872
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.169414674361088e-05,
-      "loss": 0.9031,
+      "learning_rate": 0.0003054410552349547,
+      "loss": 4.2757,
       "step": 3873
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.16323165704864e-05,
-      "loss": 0.9528,
+      "learning_rate": 0.00030523495465787303,
+      "loss": 4.3322,
       "step": 3874
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.15704863973619e-05,
-      "loss": 0.8572,
+      "learning_rate": 0.00030502885408079144,
+      "loss": 4.2184,
       "step": 3875
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.150865622423743e-05,
-      "loss": 0.7077,
+      "learning_rate": 0.0003048227535037098,
+      "loss": 4.3487,
       "step": 3876
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.144682605111293e-05,
-      "loss": 0.8899,
+      "learning_rate": 0.0003046166529266282,
+      "loss": 4.1783,
       "step": 3877
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.138499587798845e-05,
-      "loss": 0.794,
+      "learning_rate": 0.00030441055234954655,
+      "loss": 4.0487,
       "step": 3878
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.132316570486395e-05,
-      "loss": 0.7972,
+      "learning_rate": 0.00030420445177246495,
+      "loss": 4.2616,
       "step": 3879
     },
     {
       "epoch": 2.17,
-      "learning_rate": 9.126133553173948e-05,
-      "loss": 1.0305,
+      "learning_rate": 0.0003039983511953833,
+      "loss": 4.5808,
       "step": 3880
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.119950535861499e-05,
-      "loss": 0.6703,
+      "learning_rate": 0.0003037922506183017,
+      "loss": 4.0817,
       "step": 3881
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.113767518549052e-05,
-      "loss": 0.9461,
+      "learning_rate": 0.0003035861500412201,
+      "loss": 4.2913,
       "step": 3882
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.107584501236603e-05,
-      "loss": 0.8642,
+      "learning_rate": 0.00030338004946413847,
+      "loss": 4.2897,
       "step": 3883
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.101401483924154e-05,
-      "loss": 0.8082,
+      "learning_rate": 0.0003031739488870569,
+      "loss": 4.3845,
       "step": 3884
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.095218466611707e-05,
-      "loss": 0.4991,
+      "learning_rate": 0.00030296784830997523,
+      "loss": 4.0265,
       "step": 3885
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.089035449299257e-05,
-      "loss": 0.6336,
+      "learning_rate": 0.0003027617477328937,
+      "loss": 4.0364,
       "step": 3886
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.082852431986809e-05,
-      "loss": 0.6302,
+      "learning_rate": 0.00030255564715581205,
+      "loss": 4.1108,
       "step": 3887
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.07666941467436e-05,
-      "loss": 0.6483,
+      "learning_rate": 0.00030234954657873045,
+      "loss": 4.1764,
       "step": 3888
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.070486397361912e-05,
-      "loss": 0.6162,
+      "learning_rate": 0.00030214344600164886,
+      "loss": 3.988,
       "step": 3889
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.064303380049464e-05,
-      "loss": 0.9101,
+      "learning_rate": 0.0003019373454245672,
+      "loss": 4.2859,
       "step": 3890
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.058120362737015e-05,
-      "loss": 0.7221,
+      "learning_rate": 0.0003017312448474856,
+      "loss": 3.9113,
       "step": 3891
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.051937345424567e-05,
-      "loss": 0.5979,
+      "learning_rate": 0.00030152514427040397,
+      "loss": 4.0745,
       "step": 3892
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.045754328112117e-05,
-      "loss": 0.6143,
+      "learning_rate": 0.0003013190436933224,
+      "loss": 4.0313,
       "step": 3893
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.03957131079967e-05,
-      "loss": 0.6394,
+      "learning_rate": 0.00030111294311624073,
+      "loss": 4.1866,
       "step": 3894
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.03338829348722e-05,
-      "loss": 0.6932,
+      "learning_rate": 0.00030090684253915914,
+      "loss": 3.8588,
       "step": 3895
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.027205276174772e-05,
-      "loss": 0.7751,
+      "learning_rate": 0.0003007007419620775,
+      "loss": 4.1562,
       "step": 3896
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.021022258862324e-05,
-      "loss": 0.5579,
+      "learning_rate": 0.0003004946413849959,
+      "loss": 4.1709,
       "step": 3897
     },
     {
       "epoch": 2.18,
-      "learning_rate": 9.014839241549876e-05,
-      "loss": 0.6385,
+      "learning_rate": 0.0003002885408079143,
+      "loss": 4.0835,
       "step": 3898
     },
     {
       "epoch": 2.19,
-      "learning_rate": 9.008656224237427e-05,
-      "loss": 0.7777,
+      "learning_rate": 0.00030008244023083265,
+      "loss": 3.7711,
       "step": 3899
     },
     {
       "epoch": 2.19,
-      "learning_rate": 9.002473206924979e-05,
-      "loss": 0.5636,
+      "learning_rate": 0.00029987633965375106,
+      "loss": 3.9256,
       "step": 3900
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.99629018961253e-05,
-      "loss": 0.4397,
+      "learning_rate": 0.0002996702390766694,
+      "loss": 3.9953,
       "step": 3901
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.990107172300081e-05,
-      "loss": 0.5608,
+      "learning_rate": 0.0002994641384995878,
+      "loss": 4.0807,
       "step": 3902
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.983924154987634e-05,
-      "loss": 1.0103,
+      "learning_rate": 0.00029925803792250617,
+      "loss": 4.0887,
       "step": 3903
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.977741137675184e-05,
-      "loss": 0.9285,
+      "learning_rate": 0.0002990519373454246,
+      "loss": 4.3249,
       "step": 3904
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.971558120362736e-05,
-      "loss": 0.7764,
+      "learning_rate": 0.000298845836768343,
+      "loss": 4.0313,
       "step": 3905
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.965375103050289e-05,
-      "loss": 0.5187,
+      "learning_rate": 0.00029863973619126134,
+      "loss": 4.0025,
       "step": 3906
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.959192085737839e-05,
-      "loss": 0.7632,
+      "learning_rate": 0.00029843363561417975,
+      "loss": 4.0359,
       "step": 3907
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.953009068425391e-05,
-      "loss": 1.2399,
+      "learning_rate": 0.0002982275350370981,
+      "loss": 4.0512,
       "step": 3908
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.946826051112942e-05,
-      "loss": 0.7088,
+      "learning_rate": 0.0002980214344600165,
+      "loss": 4.3657,
       "step": 3909
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.940643033800494e-05,
-      "loss": 1.9809,
+      "learning_rate": 0.00029781533388293486,
+      "loss": 4.6396,
       "step": 3910
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.934460016488044e-05,
-      "loss": 0.7403,
+      "learning_rate": 0.00029760923330585326,
+      "loss": 4.1532,
       "step": 3911
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.928276999175597e-05,
-      "loss": 0.3709,
+      "learning_rate": 0.0002974031327287716,
+      "loss": 3.8258,
       "step": 3912
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.922093981863148e-05,
-      "loss": 0.5473,
+      "learning_rate": 0.00029719703215169,
+      "loss": 3.8341,
       "step": 3913
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.9159109645507e-05,
-      "loss": 0.4652,
+      "learning_rate": 0.00029699093157460843,
+      "loss": 3.8641,
       "step": 3914
     },
     {
       "epoch": 2.19,
-      "learning_rate": 8.909727947238252e-05,
-      "loss": 0.9471,
+      "learning_rate": 0.0002967848309975268,
+      "loss": 3.8049,
       "step": 3915
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.903544929925803e-05,
-      "loss": 0.3047,
+      "learning_rate": 0.0002965787304204452,
+      "loss": 3.3339,
       "step": 3916
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.897361912613354e-05,
-      "loss": 0.9133,
+      "learning_rate": 0.00029637262984336354,
+      "loss": 3.7607,
       "step": 3917
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.891178895300906e-05,
-      "loss": 0.944,
+      "learning_rate": 0.00029616652926628195,
+      "loss": 3.3526,
       "step": 3918
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.884995877988458e-05,
-      "loss": 1.4799,
+      "learning_rate": 0.0002959604286892003,
+      "loss": 4.6614,
       "step": 3919
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.878812860676008e-05,
-      "loss": 1.0627,
+      "learning_rate": 0.0002957543281121187,
+      "loss": 4.3389,
       "step": 3920
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.872629843363561e-05,
-      "loss": 0.892,
+      "learning_rate": 0.0002955482275350371,
+      "loss": 4.0561,
       "step": 3921
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.866446826051113e-05,
-      "loss": 1.0198,
+      "learning_rate": 0.00029534212695795547,
+      "loss": 4.0836,
       "step": 3922
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.860263808738663e-05,
-      "loss": 0.9239,
+      "learning_rate": 0.0002951360263808739,
+      "loss": 4.4105,
       "step": 3923
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.854080791426216e-05,
-      "loss": 0.8077,
+      "learning_rate": 0.0002949299258037922,
+      "loss": 4.1422,
       "step": 3924
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.847897774113766e-05,
-      "loss": 0.7253,
+      "learning_rate": 0.00029472382522671063,
+      "loss": 4.1495,
       "step": 3925
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.841714756801319e-05,
-      "loss": 0.6386,
+      "learning_rate": 0.000294517724649629,
+      "loss": 4.1073,
       "step": 3926
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.83553173948887e-05,
-      "loss": 0.6247,
+      "learning_rate": 0.0002943116240725474,
+      "loss": 4.1018,
       "step": 3927
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.829348722176421e-05,
-      "loss": 0.6589,
+      "learning_rate": 0.0002941055234954658,
+      "loss": 3.9543,
       "step": 3928
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.823165704863974e-05,
-      "loss": 0.7643,
+      "learning_rate": 0.00029389942291838415,
+      "loss": 4.2391,
       "step": 3929
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.816982687551525e-05,
-      "loss": 0.6978,
+      "learning_rate": 0.00029369332234130256,
+      "loss": 4.7043,
       "step": 3930
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.810799670239076e-05,
-      "loss": 0.668,
+      "learning_rate": 0.0002934872217642209,
+      "loss": 4.084,
       "step": 3931
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.804616652926626e-05,
-      "loss": 0.7327,
+      "learning_rate": 0.00029328112118713937,
+      "loss": 4.185,
       "step": 3932
     },
     {
       "epoch": 2.2,
-      "learning_rate": 8.79843363561418e-05,
-      "loss": 0.781,
+      "learning_rate": 0.0002930750206100577,
+      "loss": 4.1114,
       "step": 3933
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.79225061830173e-05,
-      "loss": 0.6842,
+      "learning_rate": 0.00029286892003297613,
+      "loss": 4.0511,
       "step": 3934
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.786067600989283e-05,
-      "loss": 0.7662,
+      "learning_rate": 0.0002926628194558945,
+      "loss": 4.2042,
       "step": 3935
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.779884583676833e-05,
-      "loss": 0.4751,
+      "learning_rate": 0.0002924567188788129,
+      "loss": 4.0384,
       "step": 3936
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.773701566364385e-05,
-      "loss": 0.9318,
+      "learning_rate": 0.0002922506183017313,
+      "loss": 4.2764,
       "step": 3937
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.767518549051938e-05,
-      "loss": 0.7284,
+      "learning_rate": 0.00029204451772464965,
+      "loss": 4.192,
       "step": 3938
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.761335531739488e-05,
-      "loss": 0.8803,
+      "learning_rate": 0.00029183841714756805,
+      "loss": 4.1381,
       "step": 3939
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.75515251442704e-05,
-      "loss": 0.6332,
+      "learning_rate": 0.0002916323165704864,
+      "loss": 3.9465,
       "step": 3940
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.74896949711459e-05,
-      "loss": 0.7292,
+      "learning_rate": 0.0002914262159934048,
+      "loss": 4.0165,
       "step": 3941
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.742786479802143e-05,
-      "loss": 0.5778,
+      "learning_rate": 0.00029122011541632317,
+      "loss": 3.8843,
       "step": 3942
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.736603462489693e-05,
-      "loss": 0.6657,
+      "learning_rate": 0.0002910140148392416,
+      "loss": 4.3323,
       "step": 3943
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.730420445177246e-05,
-      "loss": 0.7948,
+      "learning_rate": 0.00029080791426216,
+      "loss": 4.069,
       "step": 3944
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.724237427864798e-05,
-      "loss": 0.7021,
+      "learning_rate": 0.00029060181368507833,
+      "loss": 4.1575,
       "step": 3945
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.718054410552348e-05,
-      "loss": 0.774,
+      "learning_rate": 0.00029039571310799674,
+      "loss": 4.3389,
       "step": 3946
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.711871393239901e-05,
-      "loss": 0.5978,
+      "learning_rate": 0.0002901896125309151,
+      "loss": 4.178,
       "step": 3947
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.705688375927452e-05,
-      "loss": 0.8151,
+      "learning_rate": 0.0002899835119538335,
+      "loss": 4.3657,
       "step": 3948
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.699505358615003e-05,
-      "loss": 0.6528,
+      "learning_rate": 0.00028977741137675185,
+      "loss": 4.0921,
       "step": 3949
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.693322341302555e-05,
-      "loss": 0.4512,
+      "learning_rate": 0.00028957131079967026,
+      "loss": 3.72,
       "step": 3950
     },
     {
       "epoch": 2.21,
-      "learning_rate": 8.687139323990107e-05,
-      "loss": 0.5106,
+      "learning_rate": 0.0002893652102225886,
+      "loss": 4.0078,
       "step": 3951
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.680956306677657e-05,
-      "loss": 0.7243,
+      "learning_rate": 0.000289159109645507,
+      "loss": 4.3387,
       "step": 3952
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.67477328936521e-05,
-      "loss": 0.4969,
+      "learning_rate": 0.0002889530090684254,
+      "loss": 3.9529,
       "step": 3953
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.668590272052762e-05,
-      "loss": 0.4942,
+      "learning_rate": 0.0002887469084913438,
+      "loss": 3.7621,
       "step": 3954
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.662407254740312e-05,
-      "loss": 0.4557,
+      "learning_rate": 0.0002885408079142622,
+      "loss": 3.802,
       "step": 3955
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.656224237427865e-05,
-      "loss": 0.6893,
+      "learning_rate": 0.00028833470733718053,
+      "loss": 4.0281,
       "step": 3956
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.650041220115415e-05,
-      "loss": 0.668,
+      "learning_rate": 0.00028812860676009894,
+      "loss": 3.9139,
       "step": 3957
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.643858202802967e-05,
-      "loss": 0.6232,
+      "learning_rate": 0.0002879225061830173,
+      "loss": 3.8896,
       "step": 3958
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.637675185490519e-05,
-      "loss": 0.5959,
+      "learning_rate": 0.0002877164056059357,
+      "loss": 3.8029,
       "step": 3959
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.63149216817807e-05,
-      "loss": 0.6432,
+      "learning_rate": 0.0002875103050288541,
+      "loss": 3.9813,
       "step": 3960
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.625309150865622e-05,
-      "loss": 0.5311,
+      "learning_rate": 0.00028730420445177246,
+      "loss": 3.7209,
       "step": 3961
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.619126133553174e-05,
-      "loss": 0.7527,
+      "learning_rate": 0.00028709810387469087,
+      "loss": 3.9258,
       "step": 3962
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.612943116240725e-05,
-      "loss": 0.8912,
+      "learning_rate": 0.0002868920032976092,
+      "loss": 4.4414,
       "step": 3963
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.606760098928275e-05,
-      "loss": 0.8153,
+      "learning_rate": 0.0002866859027205276,
+      "loss": 3.9385,
       "step": 3964
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.600577081615828e-05,
-      "loss": 0.5079,
+      "learning_rate": 0.000286479802143446,
+      "loss": 3.5593,
       "step": 3965
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.594394064303379e-05,
-      "loss": 0.8524,
+      "learning_rate": 0.0002862737015663644,
+      "loss": 3.8846,
       "step": 3966
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.58821104699093e-05,
-      "loss": 1.0528,
+      "learning_rate": 0.00028606760098928274,
+      "loss": 4.0206,
       "step": 3967
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.582028029678482e-05,
-      "loss": 0.7226,
+      "learning_rate": 0.00028586150041220114,
+      "loss": 3.4617,
       "step": 3968
     },
     {
       "epoch": 2.22,
-      "learning_rate": 8.575845012366034e-05,
-      "loss": 0.9423,
+      "learning_rate": 0.00028565539983511955,
+      "loss": 4.4926,
       "step": 3969
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.569661995053585e-05,
-      "loss": 1.1104,
+      "learning_rate": 0.0002854492992580379,
+      "loss": 4.3168,
       "step": 3970
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.563478977741137e-05,
-      "loss": 1.2915,
+      "learning_rate": 0.0002852431986809563,
+      "loss": 4.428,
       "step": 3971
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.557295960428689e-05,
-      "loss": 0.7356,
+      "learning_rate": 0.00028503709810387466,
+      "loss": 4.0235,
       "step": 3972
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.551112943116239e-05,
-      "loss": 1.2762,
+      "learning_rate": 0.00028483099752679307,
+      "loss": 4.6313,
       "step": 3973
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.544929925803792e-05,
-      "loss": 0.7625,
+      "learning_rate": 0.0002846248969497114,
+      "loss": 4.0336,
       "step": 3974
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.538746908491342e-05,
-      "loss": 0.816,
+      "learning_rate": 0.00028441879637262983,
+      "loss": 4.27,
       "step": 3975
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.532563891178894e-05,
-      "loss": 0.6719,
+      "learning_rate": 0.00028421269579554823,
+      "loss": 4.343,
       "step": 3976
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.526380873866447e-05,
-      "loss": 0.7717,
+      "learning_rate": 0.0002840065952184666,
+      "loss": 4.0177,
       "step": 3977
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.520197856553997e-05,
-      "loss": 0.5259,
+      "learning_rate": 0.000283800494641385,
+      "loss": 4.0206,
       "step": 3978
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.51401483924155e-05,
-      "loss": 1.0245,
+      "learning_rate": 0.0002835943940643034,
+      "loss": 4.4611,
       "step": 3979
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.5078318219291e-05,
-      "loss": 0.6613,
+      "learning_rate": 0.0002833882934872218,
+      "loss": 4.2425,
       "step": 3980
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.501648804616652e-05,
-      "loss": 0.4869,
+      "learning_rate": 0.00028318219291014016,
+      "loss": 3.8993,
       "step": 3981
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.495465787304203e-05,
-      "loss": 0.5906,
+      "learning_rate": 0.00028297609233305857,
+      "loss": 4.0273,
       "step": 3982
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.489282769991756e-05,
-      "loss": 0.7758,
+      "learning_rate": 0.0002827699917559769,
+      "loss": 4.1378,
       "step": 3983
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.483099752679306e-05,
-      "loss": 0.6269,
+      "learning_rate": 0.0002825638911788953,
+      "loss": 4.0979,
       "step": 3984
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.476916735366858e-05,
-      "loss": 0.7002,
+      "learning_rate": 0.00028235779060181373,
+      "loss": 4.19,
       "step": 3985
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.47073371805441e-05,
-      "loss": 0.9105,
+      "learning_rate": 0.0002821516900247321,
+      "loss": 4.0824,
       "step": 3986
     },
     {
       "epoch": 2.23,
-      "learning_rate": 8.464550700741961e-05,
-      "loss": 0.5755,
+      "learning_rate": 0.0002819455894476505,
+      "loss": 4.1026,
       "step": 3987
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.458367683429514e-05,
-      "loss": 0.6278,
+      "learning_rate": 0.00028173948887056884,
+      "loss": 4.1953,
       "step": 3988
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.452184666117064e-05,
-      "loss": 0.7023,
+      "learning_rate": 0.00028153338829348725,
+      "loss": 3.9746,
       "step": 3989
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.446001648804616e-05,
-      "loss": 1.4374,
+      "learning_rate": 0.0002813272877164056,
+      "loss": 4.3866,
       "step": 3990
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.439818631492166e-05,
-      "loss": 0.729,
+      "learning_rate": 0.000281121187139324,
+      "loss": 4.1055,
       "step": 3991
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.433635614179719e-05,
-      "loss": 0.6758,
+      "learning_rate": 0.0002809150865622424,
+      "loss": 4.2328,
       "step": 3992
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.427452596867271e-05,
-      "loss": 0.5556,
+      "learning_rate": 0.00028070898598516077,
+      "loss": 4.0071,
       "step": 3993
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.421269579554821e-05,
-      "loss": 0.829,
+      "learning_rate": 0.0002805028854080792,
+      "loss": 4.1887,
       "step": 3994
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.415086562242374e-05,
-      "loss": 0.7808,
+      "learning_rate": 0.00028029678483099753,
+      "loss": 4.26,
       "step": 3995
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.408903544929924e-05,
-      "loss": 0.6169,
+      "learning_rate": 0.00028009068425391594,
+      "loss": 3.9181,
       "step": 3996
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.402720527617478e-05,
-      "loss": 1.5982,
+      "learning_rate": 0.0002798845836768343,
+      "loss": 4.6314,
       "step": 3997
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.396537510305028e-05,
-      "loss": 0.8222,
+      "learning_rate": 0.0002796784830997527,
+      "loss": 4.187,
       "step": 3998
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.39035449299258e-05,
-      "loss": 0.6408,
+      "learning_rate": 0.00027947238252267105,
+      "loss": 4.0675,
       "step": 3999
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.38417147568013e-05,
-      "loss": 0.8691,
+      "learning_rate": 0.00027926628194558945,
+      "loss": 4.4488,
       "step": 4000
     },
     {
       "epoch": 2.24,
-      "eval_loss": 1.3047311305999756,
-      "eval_runtime": 1254.1128,
-      "eval_samples_per_second": 2.107,
-      "eval_steps_per_second": 0.264,
-      "eval_wer": 1.0996033320111067,
+      "eval_loss": 5.381552219390869,
+      "eval_runtime": 1248.3108,
+      "eval_samples_per_second": 2.116,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.9657080523601744,
       "step": 4000
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.377988458367683e-05,
-      "loss": 0.6498,
+      "learning_rate": 0.00027906018136850786,
+      "loss": 4.1331,
       "step": 4001
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.371805441055234e-05,
-      "loss": 0.723,
+      "learning_rate": 0.0002788540807914262,
+      "loss": 4.2401,
       "step": 4002
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.365622423742786e-05,
-      "loss": 0.7007,
+      "learning_rate": 0.0002786479802143446,
+      "loss": 4.1776,
       "step": 4003
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.359439406430338e-05,
-      "loss": 0.5938,
+      "learning_rate": 0.00027844187963726297,
+      "loss": 3.8531,
       "step": 4004
     },
     {
       "epoch": 2.24,
-      "learning_rate": 8.353256389117888e-05,
-      "loss": 1.0825,
+      "learning_rate": 0.0002782357790601814,
+      "loss": 4.4502,
       "step": 4005
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.347073371805441e-05,
-      "loss": 0.5292,
+      "learning_rate": 0.00027802967848309973,
+      "loss": 4.0371,
       "step": 4006
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.340890354492991e-05,
-      "loss": 0.6996,
+      "learning_rate": 0.00027782357790601814,
+      "loss": 4.0091,
       "step": 4007
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.334707337180543e-05,
-      "loss": 0.5596,
+      "learning_rate": 0.00027761747732893654,
+      "loss": 4.079,
       "step": 4008
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.328524319868096e-05,
-      "loss": 0.5794,
+      "learning_rate": 0.0002774113767518549,
+      "loss": 4.2397,
       "step": 4009
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.322341302555646e-05,
-      "loss": 0.6314,
+      "learning_rate": 0.0002772052761747733,
+      "loss": 4.0205,
       "step": 4010
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.316158285243198e-05,
-      "loss": 0.8064,
+      "learning_rate": 0.00027699917559769166,
+      "loss": 4.4451,
       "step": 4011
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.30997526793075e-05,
-      "loss": 0.7603,
+      "learning_rate": 0.00027679307502061006,
+      "loss": 3.8258,
       "step": 4012
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.303792250618301e-05,
-      "loss": 0.3365,
+      "learning_rate": 0.0002765869744435284,
+      "loss": 3.5809,
       "step": 4013
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.297609233305852e-05,
-      "loss": 0.5149,
+      "learning_rate": 0.0002763808738664468,
+      "loss": 3.6749,
       "step": 4014
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.291426215993405e-05,
-      "loss": 0.4356,
+      "learning_rate": 0.0002761747732893652,
+      "loss": 3.2793,
       "step": 4015
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.285243198680955e-05,
-      "loss": 1.0455,
+      "learning_rate": 0.0002759686727122836,
+      "loss": 3.9043,
       "step": 4016
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.279060181368507e-05,
-      "loss": 0.7449,
+      "learning_rate": 0.000275762572135202,
+      "loss": 3.6897,
       "step": 4017
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.27287716405606e-05,
-      "loss": 0.7342,
+      "learning_rate": 0.00027555647155812034,
+      "loss": 3.6636,
       "step": 4018
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.26669414674361e-05,
-      "loss": 1.0529,
+      "learning_rate": 0.00027535037098103875,
+      "loss": 4.5099,
       "step": 4019
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.260511129431162e-05,
-      "loss": 0.8978,
+      "learning_rate": 0.0002751442704039571,
+      "loss": 4.3459,
       "step": 4020
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.254328112118713e-05,
-      "loss": 0.8821,
+      "learning_rate": 0.0002749381698268755,
+      "loss": 4.1424,
       "step": 4021
     },
     {
       "epoch": 2.25,
-      "learning_rate": 8.248145094806265e-05,
-      "loss": 0.7915,
+      "learning_rate": 0.00027473206924979386,
+      "loss": 4.1823,
       "step": 4022
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.241962077493815e-05,
-      "loss": 0.5727,
+      "learning_rate": 0.00027452596867271227,
+      "loss": 4.1436,
       "step": 4023
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.235779060181368e-05,
-      "loss": 0.8529,
+      "learning_rate": 0.00027431986809563067,
+      "loss": 4.2751,
       "step": 4024
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.22959604286892e-05,
-      "loss": 0.8723,
+      "learning_rate": 0.000274113767518549,
+      "loss": 4.7626,
       "step": 4025
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.22341302555647e-05,
-      "loss": 0.7555,
+      "learning_rate": 0.0002739076669414675,
+      "loss": 4.3093,
       "step": 4026
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.217230008244023e-05,
-      "loss": 0.5952,
+      "learning_rate": 0.00027370156636438584,
+      "loss": 4.3633,
       "step": 4027
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.211046990931574e-05,
-      "loss": 0.8427,
+      "learning_rate": 0.00027349546578730424,
+      "loss": 4.3353,
       "step": 4028
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.204863973619125e-05,
-      "loss": 0.5861,
+      "learning_rate": 0.0002732893652102226,
+      "loss": 4.1317,
       "step": 4029
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.198680956306677e-05,
-      "loss": 0.6408,
+      "learning_rate": 0.000273083264633141,
+      "loss": 3.9832,
       "step": 4030
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.192497938994228e-05,
-      "loss": 0.5399,
+      "learning_rate": 0.00027287716405605936,
+      "loss": 3.9999,
       "step": 4031
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.186314921681779e-05,
-      "loss": 0.6136,
+      "learning_rate": 0.00027267106347897776,
+      "loss": 4.0836,
       "step": 4032
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.180131904369332e-05,
-      "loss": 0.9722,
+      "learning_rate": 0.00027246496290189617,
+      "loss": 4.3445,
       "step": 4033
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.173948887056883e-05,
-      "loss": 0.7486,
+      "learning_rate": 0.0002722588623248145,
+      "loss": 4.0807,
       "step": 4034
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.167765869744434e-05,
-      "loss": 0.8711,
+      "learning_rate": 0.00027205276174773293,
+      "loss": 4.3269,
       "step": 4035
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.161582852431987e-05,
-      "loss": 0.5746,
+      "learning_rate": 0.0002718466611706513,
+      "loss": 3.9357,
       "step": 4036
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.155399835119537e-05,
-      "loss": 0.9245,
+      "learning_rate": 0.0002716405605935697,
+      "loss": 4.1915,
       "step": 4037
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.149216817807089e-05,
-      "loss": 0.6852,
+      "learning_rate": 0.00027143446001648804,
+      "loss": 4.2524,
       "step": 4038
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.14303380049464e-05,
-      "loss": 0.4695,
+      "learning_rate": 0.00027122835943940645,
+      "loss": 4.0003,
       "step": 4039
     },
     {
       "epoch": 2.26,
-      "learning_rate": 8.136850783182192e-05,
-      "loss": 0.8637,
+      "learning_rate": 0.00027102225886232485,
+      "loss": 4.3482,
       "step": 4040
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.130667765869745e-05,
-      "loss": 0.5559,
+      "learning_rate": 0.0002708161582852432,
+      "loss": 3.9382,
       "step": 4041
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.124484748557295e-05,
-      "loss": 0.6245,
+      "learning_rate": 0.0002706100577081616,
+      "loss": 4.1501,
       "step": 4042
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.118301731244847e-05,
-      "loss": 0.7447,
+      "learning_rate": 0.00027040395713107997,
+      "loss": 4.3174,
       "step": 4043
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.112118713932397e-05,
-      "loss": 0.5889,
+      "learning_rate": 0.00027019785655399837,
+      "loss": 4.1654,
       "step": 4044
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.10593569661995e-05,
-      "loss": 0.5829,
+      "learning_rate": 0.0002699917559769167,
+      "loss": 4.0054,
       "step": 4045
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.0997526793075e-05,
-      "loss": 0.676,
+      "learning_rate": 0.00026978565539983513,
+      "loss": 4.3263,
       "step": 4046
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.093569661995052e-05,
-      "loss": 0.5349,
+      "learning_rate": 0.00026957955482275354,
+      "loss": 3.8431,
       "step": 4047
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.087386644682605e-05,
-      "loss": 0.9135,
+      "learning_rate": 0.0002693734542456719,
+      "loss": 4.1205,
       "step": 4048
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.081203627370156e-05,
-      "loss": 0.5713,
+      "learning_rate": 0.0002691673536685903,
+      "loss": 3.9594,
       "step": 4049
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.075020610057709e-05,
-      "loss": 0.5756,
+      "learning_rate": 0.00026896125309150865,
+      "loss": 4.4257,
       "step": 4050
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.068837592745259e-05,
-      "loss": 0.6017,
+      "learning_rate": 0.00026875515251442706,
+      "loss": 4.0251,
       "step": 4051
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.06265457543281e-05,
-      "loss": 0.6203,
+      "learning_rate": 0.0002685490519373454,
+      "loss": 4.0489,
       "step": 4052
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.056471558120361e-05,
-      "loss": 1.0793,
+      "learning_rate": 0.0002683429513602638,
+      "loss": 4.5972,
       "step": 4053
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.050288540807914e-05,
-      "loss": 0.5087,
+      "learning_rate": 0.00026813685078318217,
+      "loss": 4.1315,
       "step": 4054
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.044105523495464e-05,
-      "loss": 0.5439,
+      "learning_rate": 0.0002679307502061006,
+      "loss": 3.922,
       "step": 4055
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.037922506183017e-05,
-      "loss": 0.4802,
+      "learning_rate": 0.000267724649629019,
+      "loss": 3.9014,
       "step": 4056
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.031739488870569e-05,
-      "loss": 0.89,
+      "learning_rate": 0.00026751854905193733,
+      "loss": 4.3813,
       "step": 4057
     },
     {
       "epoch": 2.27,
-      "learning_rate": 8.025556471558119e-05,
-      "loss": 0.7267,
+      "learning_rate": 0.00026731244847485574,
+      "loss": 4.0994,
       "step": 4058
     },
     {
       "epoch": 2.28,
-      "learning_rate": 8.019373454245672e-05,
-      "loss": 0.6521,
+      "learning_rate": 0.0002671063478977741,
+      "loss": 4.0838,
       "step": 4059
     },
     {
       "epoch": 2.28,
-      "learning_rate": 8.013190436933223e-05,
-      "loss": 0.4522,
+      "learning_rate": 0.0002669002473206925,
+      "loss": 4.1131,
       "step": 4060
     },
     {
       "epoch": 2.28,
-      "learning_rate": 8.007007419620774e-05,
-      "loss": 0.857,
+      "learning_rate": 0.00026669414674361085,
+      "loss": 4.0381,
       "step": 4061
     },
     {
       "epoch": 2.28,
-      "learning_rate": 8.000824402308324e-05,
-      "loss": 0.474,
+      "learning_rate": 0.00026648804616652926,
+      "loss": 3.5563,
       "step": 4062
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.994641384995877e-05,
-      "loss": 0.6628,
+      "learning_rate": 0.00026628194558944767,
+      "loss": 4.059,
       "step": 4063
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.988458367683429e-05,
-      "loss": 1.1931,
+      "learning_rate": 0.000266075845012366,
+      "loss": 4.3212,
       "step": 4064
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.982275350370981e-05,
-      "loss": 0.601,
+      "learning_rate": 0.0002658697444352844,
+      "loss": 3.7728,
       "step": 4065
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.976092333058532e-05,
-      "loss": 0.8011,
+      "learning_rate": 0.0002656636438582028,
+      "loss": 3.7103,
       "step": 4066
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.969909315746083e-05,
-      "loss": 0.8401,
+      "learning_rate": 0.0002654575432811212,
+      "loss": 3.4302,
       "step": 4067
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.963726298433636e-05,
-      "loss": 0.665,
+      "learning_rate": 0.00026525144270403954,
+      "loss": 3.9319,
       "step": 4068
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.957543281121186e-05,
-      "loss": 0.9989,
+      "learning_rate": 0.00026504534212695794,
+      "loss": 4.3872,
       "step": 4069
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.951360263808738e-05,
-      "loss": 0.866,
+      "learning_rate": 0.0002648392415498763,
+      "loss": 4.2557,
       "step": 4070
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.945177246496288e-05,
-      "loss": 0.8849,
+      "learning_rate": 0.0002646331409727947,
+      "loss": 4.4009,
       "step": 4071
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.938994229183841e-05,
-      "loss": 0.7853,
+      "learning_rate": 0.00026442704039571316,
+      "loss": 4.1808,
       "step": 4072
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.932811211871393e-05,
-      "loss": 0.7139,
+      "learning_rate": 0.0002642209398186315,
+      "loss": 4.0731,
       "step": 4073
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.926628194558944e-05,
-      "loss": 0.9151,
+      "learning_rate": 0.0002640148392415499,
+      "loss": 4.3833,
       "step": 4074
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.920445177246496e-05,
-      "loss": 0.8789,
+      "learning_rate": 0.0002638087386644683,
+      "loss": 4.3395,
       "step": 4075
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.914262159934046e-05,
-      "loss": 0.8729,
+      "learning_rate": 0.0002636026380873867,
+      "loss": 4.499,
       "step": 4076
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.9080791426216e-05,
-      "loss": 0.9543,
+      "learning_rate": 0.00026339653751030503,
+      "loss": 4.4247,
       "step": 4077
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.90189612530915e-05,
-      "loss": 0.7225,
+      "learning_rate": 0.00026319043693322344,
+      "loss": 4.0909,
       "step": 4078
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.895713107996701e-05,
-      "loss": 1.0453,
+      "learning_rate": 0.00026298433635614185,
+      "loss": 4.6453,
       "step": 4079
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.889530090684254e-05,
-      "loss": 0.6378,
+      "learning_rate": 0.0002627782357790602,
+      "loss": 4.0222,
       "step": 4080
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.883347073371805e-05,
-      "loss": 0.5717,
+      "learning_rate": 0.0002625721352019786,
+      "loss": 4.0939,
       "step": 4081
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.877164056059356e-05,
-      "loss": 0.5046,
+      "learning_rate": 0.00026236603462489696,
+      "loss": 3.9484,
       "step": 4082
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.870981038746908e-05,
-      "loss": 0.6594,
+      "learning_rate": 0.00026215993404781537,
+      "loss": 4.1714,
       "step": 4083
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.86479802143446e-05,
-      "loss": 0.6875,
+      "learning_rate": 0.0002619538334707337,
+      "loss": 3.9365,
       "step": 4084
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.85861500412201e-05,
-      "loss": 0.6762,
+      "learning_rate": 0.0002617477328936521,
+      "loss": 4.2443,
       "step": 4085
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.852431986809563e-05,
-      "loss": 0.598,
+      "learning_rate": 0.0002615416323165705,
+      "loss": 4.2667,
       "step": 4086
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.846248969497113e-05,
-      "loss": 0.6209,
+      "learning_rate": 0.0002613355317394889,
+      "loss": 4.1833,
       "step": 4087
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.840065952184665e-05,
-      "loss": 0.9046,
+      "learning_rate": 0.0002611294311624073,
+      "loss": 4.382,
       "step": 4088
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.833882934872218e-05,
-      "loss": 0.6003,
+      "learning_rate": 0.00026092333058532564,
+      "loss": 4.1521,
       "step": 4089
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.827699917559768e-05,
-      "loss": 0.7829,
+      "learning_rate": 0.00026071723000824405,
+      "loss": 4.0904,
       "step": 4090
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.82151690024732e-05,
-      "loss": 0.6011,
+      "learning_rate": 0.0002605111294311624,
+      "loss": 4.0451,
       "step": 4091
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.815333882934872e-05,
-      "loss": 0.7762,
+      "learning_rate": 0.0002603050288540808,
+      "loss": 4.1446,
       "step": 4092
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.809150865622423e-05,
-      "loss": 0.6469,
+      "learning_rate": 0.00026009892827699916,
+      "loss": 4.1433,
       "step": 4093
     },
     {
       "epoch": 2.29,
-      "learning_rate": 7.802967848309973e-05,
-      "loss": 0.7698,
+      "learning_rate": 0.00025989282769991757,
+      "loss": 4.2448,
       "step": 4094
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.796784830997527e-05,
-      "loss": 0.5493,
+      "learning_rate": 0.000259686727122836,
+      "loss": 4.0876,
       "step": 4095
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.790601813685078e-05,
-      "loss": 0.663,
+      "learning_rate": 0.00025948062654575433,
+      "loss": 4.1934,
       "step": 4096
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.784418796372628e-05,
-      "loss": 0.7201,
+      "learning_rate": 0.00025927452596867273,
+      "loss": 4.3251,
       "step": 4097
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.778235779060181e-05,
-      "loss": 0.7861,
+      "learning_rate": 0.0002590684253915911,
+      "loss": 4.1628,
       "step": 4098
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.772052761747732e-05,
-      "loss": 0.5661,
+      "learning_rate": 0.0002588623248145095,
+      "loss": 4.1732,
       "step": 4099
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.765869744435283e-05,
-      "loss": 0.7298,
+      "learning_rate": 0.00025865622423742785,
+      "loss": 4.3772,
       "step": 4100
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.759686727122835e-05,
-      "loss": 0.6556,
+      "learning_rate": 0.00025845012366034625,
+      "loss": 4.0576,
       "step": 4101
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.753503709810387e-05,
-      "loss": 0.5395,
+      "learning_rate": 0.0002582440230832646,
+      "loss": 4.0362,
       "step": 4102
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.747320692497937e-05,
-      "loss": 0.6532,
+      "learning_rate": 0.000258037922506183,
+      "loss": 3.7747,
       "step": 4103
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.74113767518549e-05,
-      "loss": 0.5568,
+      "learning_rate": 0.0002578318219291014,
+      "loss": 3.9429,
       "step": 4104
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.734954657873042e-05,
-      "loss": 0.5332,
+      "learning_rate": 0.00025762572135201977,
+      "loss": 3.9268,
       "step": 4105
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.728771640560592e-05,
-      "loss": 0.423,
+      "learning_rate": 0.0002574196207749382,
+      "loss": 3.829,
       "step": 4106
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.722588623248145e-05,
-      "loss": 0.6212,
+      "learning_rate": 0.00025721352019785653,
+      "loss": 4.036,
       "step": 4107
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.716405605935695e-05,
-      "loss": 0.5888,
+      "learning_rate": 0.00025700741962077494,
+      "loss": 3.9983,
       "step": 4108
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.710222588623248e-05,
-      "loss": 0.8165,
+      "learning_rate": 0.0002568013190436933,
+      "loss": 4.2549,
       "step": 4109
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.704039571310799e-05,
-      "loss": 0.6413,
+      "learning_rate": 0.0002565952184666117,
+      "loss": 4.0171,
       "step": 4110
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.69785655399835e-05,
-      "loss": 0.5777,
+      "learning_rate": 0.0002563891178895301,
+      "loss": 3.9687,
       "step": 4111
     },
     {
       "epoch": 2.3,
-      "learning_rate": 7.691673536685903e-05,
-      "loss": 0.5962,
+      "learning_rate": 0.00025618301731244845,
+      "loss": 3.6023,
       "step": 4112
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.685490519373454e-05,
-      "loss": 0.3287,
+      "learning_rate": 0.00025597691673536686,
+      "loss": 3.2781,
       "step": 4113
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.679307502061005e-05,
-      "loss": 1.0677,
+      "learning_rate": 0.0002557708161582852,
+      "loss": 4.6506,
       "step": 4114
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.673124484748556e-05,
-      "loss": 0.7265,
+      "learning_rate": 0.0002555647155812036,
+      "loss": 3.5175,
       "step": 4115
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.666941467436109e-05,
-      "loss": 0.6934,
+      "learning_rate": 0.000255358615004122,
+      "loss": 3.6853,
       "step": 4116
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.660758450123659e-05,
-      "loss": 0.6165,
+      "learning_rate": 0.0002551525144270404,
+      "loss": 3.5823,
       "step": 4117
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.654575432811212e-05,
-      "loss": 0.5869,
+      "learning_rate": 0.00025494641384995873,
+      "loss": 3.4434,
       "step": 4118
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.648392415498762e-05,
-      "loss": 0.7948,
+      "learning_rate": 0.0002547403132728772,
+      "loss": 4.2603,
       "step": 4119
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.642209398186314e-05,
-      "loss": 1.1123,
+      "learning_rate": 0.0002545342126957956,
+      "loss": 4.3738,
       "step": 4120
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.636026380873867e-05,
-      "loss": 0.7978,
+      "learning_rate": 0.00025432811211871395,
+      "loss": 4.4777,
       "step": 4121
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.629843363561417e-05,
-      "loss": 0.7496,
+      "learning_rate": 0.00025412201154163236,
+      "loss": 4.0555,
       "step": 4122
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.623660346248969e-05,
-      "loss": 0.6978,
+      "learning_rate": 0.0002539159109645507,
+      "loss": 3.9854,
       "step": 4123
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.617477328936519e-05,
-      "loss": 0.886,
+      "learning_rate": 0.0002537098103874691,
+      "loss": 4.2689,
       "step": 4124
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.611294311624072e-05,
-      "loss": 0.6058,
+      "learning_rate": 0.00025350370981038747,
+      "loss": 4.0145,
       "step": 4125
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.605111294311623e-05,
-      "loss": 0.8336,
+      "learning_rate": 0.0002532976092333059,
+      "loss": 4.0647,
       "step": 4126
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.598928276999176e-05,
-      "loss": 0.6878,
+      "learning_rate": 0.0002530915086562243,
+      "loss": 4.1744,
       "step": 4127
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.592745259686727e-05,
-      "loss": 0.583,
+      "learning_rate": 0.00025288540807914264,
+      "loss": 4.3321,
       "step": 4128
     },
     {
       "epoch": 2.31,
-      "learning_rate": 7.586562242374277e-05,
-      "loss": 0.7582,
+      "learning_rate": 0.00025267930750206104,
+      "loss": 4.1915,
       "step": 4129
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.58037922506183e-05,
-      "loss": 0.8283,
+      "learning_rate": 0.0002524732069249794,
+      "loss": 4.2373,
       "step": 4130
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.574196207749381e-05,
-      "loss": 1.0395,
+      "learning_rate": 0.0002522671063478978,
+      "loss": 4.5036,
       "step": 4131
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.568013190436932e-05,
-      "loss": 0.5309,
+      "learning_rate": 0.00025206100577081616,
+      "loss": 4.1453,
       "step": 4132
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.561830173124484e-05,
-      "loss": 1.1222,
+      "learning_rate": 0.00025185490519373456,
+      "loss": 4.7558,
       "step": 4133
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.555647155812036e-05,
-      "loss": 0.6209,
+      "learning_rate": 0.0002516488046166529,
+      "loss": 4.2711,
       "step": 4134
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.549464138499586e-05,
-      "loss": 0.5924,
+      "learning_rate": 0.0002514427040395713,
+      "loss": 4.1406,
       "step": 4135
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.543281121187139e-05,
-      "loss": 0.6508,
+      "learning_rate": 0.00025123660346248973,
+      "loss": 4.1755,
       "step": 4136
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.537098103874691e-05,
-      "loss": 0.7506,
+      "learning_rate": 0.0002510305028854081,
+      "loss": 4.2562,
       "step": 4137
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.530915086562241e-05,
-      "loss": 0.7206,
+      "learning_rate": 0.0002508244023083265,
+      "loss": 3.9805,
       "step": 4138
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.524732069249794e-05,
-      "loss": 0.6861,
+      "learning_rate": 0.00025061830173124484,
+      "loss": 4.0209,
       "step": 4139
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.518549051937344e-05,
-      "loss": 0.5316,
+      "learning_rate": 0.00025041220115416325,
+      "loss": 4.1379,
       "step": 4140
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.512366034624896e-05,
-      "loss": 0.4754,
+      "learning_rate": 0.0002502061005770816,
+      "loss": 3.8318,
       "step": 4141
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.506183017312448e-05,
-      "loss": 0.6305,
+      "learning_rate": 0.00025,
+      "loss": 4.0634,
       "step": 4142
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.5e-05,
-      "loss": 0.6427,
+      "learning_rate": 0.0002497938994229184,
+      "loss": 3.9837,
       "step": 4143
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.493816982687551e-05,
-      "loss": 0.7704,
+      "learning_rate": 0.00024958779884583676,
+      "loss": 4.3871,
       "step": 4144
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.487633965375103e-05,
-      "loss": 0.5576,
+      "learning_rate": 0.00024938169826875517,
+      "loss": 4.1231,
       "step": 4145
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.481450948062653e-05,
-      "loss": 0.7657,
+      "learning_rate": 0.0002491755976916736,
+      "loss": 4.1318,
       "step": 4146
     },
     {
       "epoch": 2.32,
-      "learning_rate": 7.475267930750206e-05,
-      "loss": 0.833,
+      "learning_rate": 0.00024896949711459193,
+      "loss": 4.1949,
       "step": 4147
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.469084913437758e-05,
-      "loss": 0.5735,
+      "learning_rate": 0.00024876339653751034,
+      "loss": 4.3088,
       "step": 4148
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.462901896125309e-05,
-      "loss": 0.5723,
+      "learning_rate": 0.0002485572959604287,
+      "loss": 4.2365,
       "step": 4149
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.45671887881286e-05,
-      "loss": 0.579,
+      "learning_rate": 0.0002483511953833471,
+      "loss": 3.9402,
       "step": 4150
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.450535861500411e-05,
-      "loss": 0.5688,
+      "learning_rate": 0.00024814509480626545,
+      "loss": 3.9767,
       "step": 4151
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.444352844187963e-05,
-      "loss": 0.6201,
+      "learning_rate": 0.00024793899422918386,
+      "loss": 4.0358,
       "step": 4152
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.438169826875515e-05,
-      "loss": 0.547,
+      "learning_rate": 0.0002477328936521022,
+      "loss": 4.1359,
       "step": 4153
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.431986809563066e-05,
-      "loss": 0.6241,
+      "learning_rate": 0.0002475267930750206,
+      "loss": 4.0114,
       "step": 4154
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.425803792250618e-05,
-      "loss": 0.4889,
+      "learning_rate": 0.000247320692497939,
+      "loss": 3.7886,
       "step": 4155
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.41962077493817e-05,
-      "loss": 0.4921,
+      "learning_rate": 0.0002471145919208574,
+      "loss": 4.1232,
       "step": 4156
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.413437757625721e-05,
-      "loss": 0.3892,
+      "learning_rate": 0.0002469084913437758,
+      "loss": 3.8082,
       "step": 4157
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.407254740313273e-05,
-      "loss": 0.4434,
+      "learning_rate": 0.00024670239076669413,
+      "loss": 3.9314,
       "step": 4158
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.401071723000823e-05,
-      "loss": 0.5716,
+      "learning_rate": 0.00024649629018961254,
+      "loss": 4.0557,
       "step": 4159
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.394888705688375e-05,
-      "loss": 0.7926,
+      "learning_rate": 0.0002462901896125309,
+      "loss": 4.1488,
       "step": 4160
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.388705688375926e-05,
-      "loss": 0.7579,
+      "learning_rate": 0.0002460840890354493,
+      "loss": 4.229,
       "step": 4161
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.382522671063478e-05,
-      "loss": 0.8734,
+      "learning_rate": 0.0002458779884583677,
+      "loss": 3.7991,
       "step": 4162
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.37633965375103e-05,
-      "loss": 0.4651,
+      "learning_rate": 0.00024567188788128606,
+      "loss": 3.8878,
       "step": 4163
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.370156636438581e-05,
-      "loss": 0.9247,
+      "learning_rate": 0.00024546578730420446,
+      "loss": 3.834,
       "step": 4164
     },
     {
       "epoch": 2.33,
-      "learning_rate": 7.363973619126133e-05,
-      "loss": 0.5537,
+      "learning_rate": 0.0002452596867271228,
+      "loss": 3.7531,
       "step": 4165
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.357790601813685e-05,
-      "loss": 0.6934,
+      "learning_rate": 0.0002450535861500412,
+      "loss": 3.6259,
       "step": 4166
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.351607584501236e-05,
-      "loss": 0.8163,
+      "learning_rate": 0.00024484748557295963,
+      "loss": 3.7643,
       "step": 4167
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.345424567188787e-05,
-      "loss": 0.9621,
+      "learning_rate": 0.000244641384995878,
+      "loss": 3.3971,
       "step": 4168
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.339241549876338e-05,
-      "loss": 0.9033,
+      "learning_rate": 0.0002444352844187964,
+      "loss": 4.3359,
       "step": 4169
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.33305853256389e-05,
-      "loss": 0.94,
+      "learning_rate": 0.0002442291838417148,
+      "loss": 4.215,
       "step": 4170
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.326875515251443e-05,
-      "loss": 0.9552,
+      "learning_rate": 0.00024402308326463315,
+      "loss": 4.2213,
       "step": 4171
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.320692497938993e-05,
-      "loss": 0.8346,
+      "learning_rate": 0.00024381698268755153,
+      "loss": 4.1195,
       "step": 4172
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.314509480626545e-05,
-      "loss": 1.0244,
+      "learning_rate": 0.0002436108821104699,
+      "loss": 4.2723,
       "step": 4173
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.308326463314097e-05,
-      "loss": 0.727,
+      "learning_rate": 0.0002434047815333883,
+      "loss": 4.2158,
       "step": 4174
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.302143446001648e-05,
-      "loss": 0.8111,
+      "learning_rate": 0.00024319868095630667,
+      "loss": 4.2159,
       "step": 4175
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.2959604286892e-05,
-      "loss": 0.6138,
+      "learning_rate": 0.00024299258037922505,
+      "loss": 4.2421,
       "step": 4176
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.28977741137675e-05,
-      "loss": 1.1392,
+      "learning_rate": 0.00024278647980214343,
+      "loss": 4.8157,
       "step": 4177
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.283594394064302e-05,
-      "loss": 0.8765,
+      "learning_rate": 0.00024258037922506186,
+      "loss": 4.4018,
       "step": 4178
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.277411376751855e-05,
-      "loss": 0.5707,
+      "learning_rate": 0.00024237427864798024,
+      "loss": 3.9332,
       "step": 4179
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.271228359439407e-05,
-      "loss": 0.9211,
+      "learning_rate": 0.00024216817807089862,
+      "loss": 4.3388,
       "step": 4180
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.265045342126957e-05,
-      "loss": 1.1647,
+      "learning_rate": 0.000241962077493817,
+      "loss": 4.4778,
       "step": 4181
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.258862324814509e-05,
-      "loss": 0.6839,
+      "learning_rate": 0.00024175597691673538,
+      "loss": 4.2407,
       "step": 4182
     },
     {
       "epoch": 2.34,
-      "learning_rate": 7.25267930750206e-05,
-      "loss": 0.7736,
+      "learning_rate": 0.00024154987633965376,
+      "loss": 4.2373,
       "step": 4183
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.246496290189612e-05,
-      "loss": 0.5913,
+      "learning_rate": 0.00024134377576257214,
+      "loss": 4.2351,
       "step": 4184
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.240313272877164e-05,
-      "loss": 0.5022,
+      "learning_rate": 0.00024113767518549052,
+      "loss": 3.8392,
       "step": 4185
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.234130255564715e-05,
-      "loss": 0.5623,
+      "learning_rate": 0.00024093157460840892,
+      "loss": 4.1394,
       "step": 4186
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.227947238252267e-05,
-      "loss": 0.8312,
+      "learning_rate": 0.0002407254740313273,
+      "loss": 3.9883,
       "step": 4187
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.221764220939819e-05,
-      "loss": 0.7901,
+      "learning_rate": 0.00024051937345424568,
+      "loss": 3.9671,
       "step": 4188
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.21558120362737e-05,
-      "loss": 0.8905,
+      "learning_rate": 0.00024031327287716406,
+      "loss": 4.5236,
       "step": 4189
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.20939818631492e-05,
-      "loss": 0.4917,
+      "learning_rate": 0.00024010717230008244,
+      "loss": 4.0749,
       "step": 4190
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.203215169002472e-05,
-      "loss": 0.4809,
+      "learning_rate": 0.00023990107172300082,
+      "loss": 4.0904,
       "step": 4191
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.197032151690024e-05,
-      "loss": 0.5333,
+      "learning_rate": 0.0002396949711459192,
+      "loss": 3.903,
       "step": 4192
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.190849134377575e-05,
-      "loss": 0.6644,
+      "learning_rate": 0.00023948887056883758,
+      "loss": 4.2202,
       "step": 4193
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.184666117065127e-05,
-      "loss": 0.8391,
+      "learning_rate": 0.000239282769991756,
+      "loss": 4.2079,
       "step": 4194
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.178483099752679e-05,
-      "loss": 0.6095,
+      "learning_rate": 0.00023907666941467437,
+      "loss": 4.1324,
       "step": 4195
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.17230008244023e-05,
-      "loss": 0.5264,
+      "learning_rate": 0.00023887056883759275,
+      "loss": 4.038,
       "step": 4196
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.166117065127782e-05,
-      "loss": 0.6987,
+      "learning_rate": 0.00023866446826051113,
+      "loss": 4.0589,
       "step": 4197
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.159934047815334e-05,
-      "loss": 0.914,
+      "learning_rate": 0.0002384583676834295,
+      "loss": 4.2291,
       "step": 4198
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.153751030502884e-05,
-      "loss": 0.6654,
+      "learning_rate": 0.00023825226710634789,
+      "loss": 3.9721,
       "step": 4199
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.147568013190436e-05,
-      "loss": 0.5368,
+      "learning_rate": 0.00023804616652926627,
+      "loss": 3.859,
       "step": 4200
     },
     {
       "epoch": 2.35,
-      "learning_rate": 7.141384995877987e-05,
-      "loss": 0.5058,
+      "learning_rate": 0.00023784006595218467,
+      "loss": 3.9392,
       "step": 4201
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.135201978565539e-05,
-      "loss": 0.6045,
+      "learning_rate": 0.00023763396537510308,
+      "loss": 4.3039,
       "step": 4202
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.129018961253091e-05,
-      "loss": 0.743,
+      "learning_rate": 0.00023742786479802146,
+      "loss": 4.1007,
       "step": 4203
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.122835943940642e-05,
-      "loss": 0.6918,
+      "learning_rate": 0.00023722176422093984,
+      "loss": 4.0879,
       "step": 4204
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.116652926628194e-05,
-      "loss": 0.5073,
+      "learning_rate": 0.00023701566364385822,
+      "loss": 4.2285,
       "step": 4205
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.110469909315746e-05,
-      "loss": 0.5016,
+      "learning_rate": 0.0002368095630667766,
+      "loss": 4.1627,
       "step": 4206
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.104286892003297e-05,
-      "loss": 0.5668,
+      "learning_rate": 0.00023660346248969498,
+      "loss": 4.1993,
       "step": 4207
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.098103874690849e-05,
-      "loss": 0.566,
+      "learning_rate": 0.00023639736191261336,
+      "loss": 4.0148,
       "step": 4208
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.0919208573784e-05,
-      "loss": 0.6318,
+      "learning_rate": 0.00023619126133553174,
+      "loss": 4.1544,
       "step": 4209
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.085737840065951e-05,
-      "loss": 0.694,
+      "learning_rate": 0.00023598516075845014,
+      "loss": 4.0744,
       "step": 4210
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.079554822753504e-05,
-      "loss": 0.5433,
+      "learning_rate": 0.00023577906018136852,
+      "loss": 4.105,
       "step": 4211
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.073371805441054e-05,
-      "loss": 0.4485,
+      "learning_rate": 0.0002355729596042869,
+      "loss": 3.5753,
       "step": 4212
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.067188788128606e-05,
-      "loss": 0.4746,
+      "learning_rate": 0.00023536685902720528,
+      "loss": 3.9634,
       "step": 4213
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.061005770816158e-05,
-      "loss": 0.5738,
+      "learning_rate": 0.00023516075845012366,
+      "loss": 3.4128,
       "step": 4214
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.054822753503709e-05,
-      "loss": 0.6298,
+      "learning_rate": 0.00023495465787304204,
+      "loss": 3.8764,
       "step": 4215
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.048639736191261e-05,
-      "loss": 0.6964,
+      "learning_rate": 0.00023474855729596042,
+      "loss": 3.7209,
       "step": 4216
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.042456718878813e-05,
-      "loss": 0.6293,
+      "learning_rate": 0.0002345424567188788,
+      "loss": 3.7795,
       "step": 4217
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.036273701566363e-05,
-      "loss": 0.6546,
+      "learning_rate": 0.0002343363561417972,
+      "loss": 3.546,
       "step": 4218
     },
     {
       "epoch": 2.36,
-      "learning_rate": 7.030090684253916e-05,
-      "loss": 0.8654,
+      "learning_rate": 0.00023413025556471559,
+      "loss": 4.1219,
       "step": 4219
     },
     {
       "epoch": 2.37,
-      "learning_rate": 7.023907666941468e-05,
-      "loss": 1.0744,
+      "learning_rate": 0.00023392415498763397,
+      "loss": 4.6844,
       "step": 4220
     },
     {
       "epoch": 2.37,
-      "learning_rate": 7.017724649629018e-05,
-      "loss": 0.8325,
+      "learning_rate": 0.00023371805441055234,
+      "loss": 4.4344,
       "step": 4221
     },
     {
       "epoch": 2.37,
-      "learning_rate": 7.01154163231657e-05,
-      "loss": 1.0698,
+      "learning_rate": 0.00023351195383347072,
+      "loss": 4.2514,
       "step": 4222
     },
     {
       "epoch": 2.37,
-      "learning_rate": 7.005358615004121e-05,
-      "loss": 0.6021,
+      "learning_rate": 0.0002333058532563891,
+      "loss": 3.8876,
       "step": 4223
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.999175597691673e-05,
-      "loss": 0.6508,
+      "learning_rate": 0.0002330997526793075,
+      "loss": 3.9969,
       "step": 4224
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.992992580379225e-05,
-      "loss": 0.5948,
+      "learning_rate": 0.0002328936521022259,
+      "loss": 3.9539,
       "step": 4225
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.986809563066776e-05,
-      "loss": 0.7577,
+      "learning_rate": 0.0002326875515251443,
+      "loss": 4.2602,
       "step": 4226
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.980626545754328e-05,
-      "loss": 0.69,
+      "learning_rate": 0.00023248145094806268,
+      "loss": 4.1161,
       "step": 4227
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.97444352844188e-05,
-      "loss": 0.77,
+      "learning_rate": 0.00023227535037098106,
+      "loss": 4.4657,
       "step": 4228
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.968260511129431e-05,
-      "loss": 0.7909,
+      "learning_rate": 0.00023206924979389944,
+      "loss": 4.1667,
       "step": 4229
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.962077493816981e-05,
-      "loss": 0.6395,
+      "learning_rate": 0.00023186314921681782,
+      "loss": 4.3854,
       "step": 4230
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.955894476504533e-05,
-      "loss": 0.8912,
+      "learning_rate": 0.0002316570486397362,
+      "loss": 4.3664,
       "step": 4231
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.949711459192085e-05,
-      "loss": 0.5428,
+      "learning_rate": 0.00023145094806265457,
+      "loss": 4.0751,
       "step": 4232
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.943528441879636e-05,
-      "loss": 0.5158,
+      "learning_rate": 0.00023124484748557295,
+      "loss": 3.9203,
       "step": 4233
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.937345424567188e-05,
-      "loss": 0.6782,
+      "learning_rate": 0.00023103874690849136,
+      "loss": 4.2466,
       "step": 4234
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.93116240725474e-05,
-      "loss": 0.7005,
+      "learning_rate": 0.00023083264633140974,
+      "loss": 4.1337,
       "step": 4235
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.924979389942291e-05,
-      "loss": 0.6676,
+      "learning_rate": 0.00023062654575432812,
+      "loss": 4.1854,
       "step": 4236
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.918796372629843e-05,
-      "loss": 0.7141,
+      "learning_rate": 0.0002304204451772465,
+      "loss": 4.2772,
       "step": 4237
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.912613355317395e-05,
-      "loss": 0.6354,
+      "learning_rate": 0.00023021434460016488,
+      "loss": 4.1271,
       "step": 4238
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.906430338004946e-05,
-      "loss": 0.683,
+      "learning_rate": 0.00023000824402308326,
+      "loss": 4.0891,
       "step": 4239
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.900247320692497e-05,
-      "loss": 0.7404,
+      "learning_rate": 0.00022980214344600164,
+      "loss": 4.1034,
       "step": 4240
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.894064303380048e-05,
-      "loss": 0.7103,
+      "learning_rate": 0.00022959604286892002,
+      "loss": 3.9906,
       "step": 4241
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.8878812860676e-05,
-      "loss": 0.6838,
+      "learning_rate": 0.00022938994229183842,
+      "loss": 4.3241,
       "step": 4242
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.881698268755152e-05,
-      "loss": 0.732,
+      "learning_rate": 0.0002291838417147568,
+      "loss": 4.0666,
       "step": 4243
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.875515251442703e-05,
-      "loss": 0.7482,
+      "learning_rate": 0.00022897774113767518,
+      "loss": 4.1986,
       "step": 4244
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.869332234130255e-05,
-      "loss": 0.5641,
+      "learning_rate": 0.00022877164056059356,
+      "loss": 4.1445,
       "step": 4245
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.863149216817807e-05,
-      "loss": 0.5604,
+      "learning_rate": 0.00022856553998351194,
+      "loss": 3.9616,
       "step": 4246
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.856966199505358e-05,
-      "loss": 1.1417,
+      "learning_rate": 0.00022835943940643032,
+      "loss": 4.2208,
       "step": 4247
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.85078318219291e-05,
-      "loss": 0.5763,
+      "learning_rate": 0.00022815333882934873,
+      "loss": 3.8156,
       "step": 4248
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.84460016488046e-05,
-      "loss": 0.7965,
+      "learning_rate": 0.0002279472382522671,
+      "loss": 4.0649,
       "step": 4249
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.838417147568012e-05,
-      "loss": 0.4285,
+      "learning_rate": 0.00022774113767518552,
+      "loss": 4.0061,
       "step": 4250
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.832234130255565e-05,
-      "loss": 0.3939,
+      "learning_rate": 0.0002275350370981039,
+      "loss": 3.9628,
       "step": 4251
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.826051112943115e-05,
-      "loss": 0.4407,
+      "learning_rate": 0.00022732893652102227,
+      "loss": 3.8302,
       "step": 4252
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.819868095630667e-05,
-      "loss": 0.5608,
+      "learning_rate": 0.00022712283594394065,
+      "loss": 3.9246,
       "step": 4253
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.813685078318219e-05,
-      "loss": 0.6879,
+      "learning_rate": 0.00022691673536685903,
+      "loss": 4.0111,
       "step": 4254
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.80750206100577e-05,
-      "loss": 0.7033,
+      "learning_rate": 0.0002267106347897774,
+      "loss": 4.0364,
       "step": 4255
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.801319043693322e-05,
-      "loss": 0.4709,
+      "learning_rate": 0.0002265045342126958,
+      "loss": 3.6394,
       "step": 4256
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.795136026380874e-05,
-      "loss": 0.4585,
+      "learning_rate": 0.0002262984336356142,
+      "loss": 3.7725,
       "step": 4257
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.788953009068425e-05,
-      "loss": 0.5118,
+      "learning_rate": 0.00022609233305853258,
+      "loss": 4.0937,
       "step": 4258
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.782769991755977e-05,
-      "loss": 0.8163,
+      "learning_rate": 0.00022588623248145096,
+      "loss": 4.0166,
       "step": 4259
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.776586974443528e-05,
-      "loss": 0.5101,
+      "learning_rate": 0.00022568013190436934,
+      "loss": 3.659,
       "step": 4260
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.770403957131079e-05,
-      "loss": 0.548,
+      "learning_rate": 0.00022547403132728772,
+      "loss": 3.853,
       "step": 4261
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.76422093981863e-05,
-      "loss": 0.8203,
+      "learning_rate": 0.0002252679307502061,
+      "loss": 4.0426,
       "step": 4262
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.758037922506182e-05,
-      "loss": 0.5455,
+      "learning_rate": 0.00022506183017312448,
+      "loss": 3.7204,
       "step": 4263
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.751854905193734e-05,
-      "loss": 0.4144,
+      "learning_rate": 0.00022485572959604286,
+      "loss": 3.6215,
       "step": 4264
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.745671887881285e-05,
-      "loss": 0.3804,
+      "learning_rate": 0.00022464962901896126,
+      "loss": 3.4008,
       "step": 4265
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.739488870568837e-05,
-      "loss": 0.4357,
+      "learning_rate": 0.00022444352844187964,
+      "loss": 3.4636,
       "step": 4266
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.733305853256389e-05,
-      "loss": 0.3713,
+      "learning_rate": 0.00022423742786479802,
+      "loss": 3.3569,
       "step": 4267
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.72712283594394e-05,
-      "loss": 0.7835,
+      "learning_rate": 0.0002240313272877164,
+      "loss": 3.873,
       "step": 4268
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.720939818631492e-05,
-      "loss": 0.8733,
+      "learning_rate": 0.00022382522671063478,
+      "loss": 4.3543,
       "step": 4269
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.714756801319044e-05,
-      "loss": 0.8224,
+      "learning_rate": 0.00022361912613355316,
+      "loss": 4.4158,
       "step": 4270
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.708573784006594e-05,
-      "loss": 0.8192,
+      "learning_rate": 0.00022341302555647157,
+      "loss": 4.2406,
       "step": 4271
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.702390766694146e-05,
-      "loss": 0.5682,
+      "learning_rate": 0.00022320692497938995,
+      "loss": 3.8841,
       "step": 4272
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.696207749381697e-05,
-      "loss": 0.6789,
+      "learning_rate": 0.00022300082440230835,
+      "loss": 4.4527,
       "step": 4273
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.690024732069249e-05,
-      "loss": 0.6833,
+      "learning_rate": 0.00022279472382522673,
+      "loss": 4.1433,
       "step": 4274
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.6838417147568e-05,
-      "loss": 0.6008,
+      "learning_rate": 0.00022258862324814511,
+      "loss": 4.0397,
       "step": 4275
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.677658697444352e-05,
-      "loss": 0.7112,
+      "learning_rate": 0.0002223825226710635,
+      "loss": 4.4524,
       "step": 4276
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.671475680131904e-05,
-      "loss": 0.7327,
+      "learning_rate": 0.00022217642209398187,
+      "loss": 4.1937,
       "step": 4277
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.665292662819456e-05,
-      "loss": 0.6956,
+      "learning_rate": 0.00022197032151690025,
+      "loss": 4.2321,
       "step": 4278
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.659109645507007e-05,
-      "loss": 0.7422,
+      "learning_rate": 0.00022176422093981863,
+      "loss": 4.1546,
       "step": 4279
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.652926628194558e-05,
-      "loss": 0.6569,
+      "learning_rate": 0.000221558120362737,
+      "loss": 4.2091,
       "step": 4280
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.646743610882109e-05,
-      "loss": 0.6469,
+      "learning_rate": 0.00022135201978565542,
+      "loss": 4.2479,
       "step": 4281
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.640560593569662e-05,
-      "loss": 0.4449,
+      "learning_rate": 0.0002211459192085738,
+      "loss": 4.1404,
       "step": 4282
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.634377576257213e-05,
-      "loss": 0.6842,
+      "learning_rate": 0.00022093981863149218,
+      "loss": 4.2063,
       "step": 4283
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.628194558944764e-05,
-      "loss": 0.6611,
+      "learning_rate": 0.00022073371805441056,
+      "loss": 4.2816,
       "step": 4284
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.622011541632316e-05,
-      "loss": 0.7678,
+      "learning_rate": 0.00022052761747732894,
+      "loss": 4.4483,
       "step": 4285
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.615828524319868e-05,
-      "loss": 0.6302,
+      "learning_rate": 0.00022032151690024732,
+      "loss": 3.9143,
       "step": 4286
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.609645507007419e-05,
-      "loss": 0.6883,
+      "learning_rate": 0.0002201154163231657,
+      "loss": 4.0612,
       "step": 4287
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.603462489694971e-05,
-      "loss": 0.5194,
+      "learning_rate": 0.00021990931574608408,
+      "loss": 4.0047,
       "step": 4288
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.597279472382521e-05,
-      "loss": 0.6342,
+      "learning_rate": 0.00021970321516900248,
+      "loss": 4.3881,
       "step": 4289
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.591096455070074e-05,
-      "loss": 0.5488,
+      "learning_rate": 0.00021949711459192086,
+      "loss": 4.1476,
       "step": 4290
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.584913437757626e-05,
-      "loss": 0.7895,
+      "learning_rate": 0.00021929101401483924,
+      "loss": 4.2333,
       "step": 4291
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.578730420445178e-05,
-      "loss": 0.7249,
+      "learning_rate": 0.00021908491343775762,
+      "loss": 3.8476,
       "step": 4292
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.572547403132728e-05,
-      "loss": 0.7935,
+      "learning_rate": 0.000218878812860676,
+      "loss": 3.9771,
       "step": 4293
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.56636438582028e-05,
-      "loss": 0.613,
+      "learning_rate": 0.0002186727122835944,
+      "loss": 4.364,
       "step": 4294
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.560181368507831e-05,
-      "loss": 0.5721,
+      "learning_rate": 0.0002184666117065128,
+      "loss": 4.0412,
       "step": 4295
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.553998351195383e-05,
-      "loss": 0.6103,
+      "learning_rate": 0.00021826051112943117,
+      "loss": 4.0513,
       "step": 4296
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.547815333882934e-05,
-      "loss": 0.4708,
+      "learning_rate": 0.00021805441055234957,
+      "loss": 3.9818,
       "step": 4297
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.541632316570486e-05,
-      "loss": 0.592,
+      "learning_rate": 0.00021784830997526795,
+      "loss": 4.434,
       "step": 4298
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.535449299258038e-05,
-      "loss": 0.4149,
+      "learning_rate": 0.00021764220939818633,
+      "loss": 3.962,
       "step": 4299
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.52926628194559e-05,
-      "loss": 0.5844,
+      "learning_rate": 0.0002174361088211047,
+      "loss": 4.1632,
       "step": 4300
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.523083264633141e-05,
-      "loss": 0.7714,
+      "learning_rate": 0.0002172300082440231,
+      "loss": 4.0821,
       "step": 4301
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.516900247320691e-05,
-      "loss": 0.5254,
+      "learning_rate": 0.00021702390766694147,
+      "loss": 4.0697,
       "step": 4302
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.510717230008243e-05,
-      "loss": 0.6662,
+      "learning_rate": 0.00021681780708985985,
+      "loss": 4.4036,
       "step": 4303
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.504534212695795e-05,
-      "loss": 0.6351,
+      "learning_rate": 0.00021661170651277823,
+      "loss": 3.9768,
       "step": 4304
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.498351195383346e-05,
-      "loss": 0.7572,
+      "learning_rate": 0.00021640560593569664,
+      "loss": 4.1046,
       "step": 4305
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.492168178070898e-05,
-      "loss": 0.6154,
+      "learning_rate": 0.00021619950535861502,
+      "loss": 3.835,
       "step": 4306
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.48598516075845e-05,
-      "loss": 0.4586,
+      "learning_rate": 0.0002159934047815334,
+      "loss": 3.9192,
       "step": 4307
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.479802143446001e-05,
-      "loss": 0.4433,
+      "learning_rate": 0.00021578730420445178,
+      "loss": 4.0563,
       "step": 4308
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.473619126133553e-05,
-      "loss": 0.8088,
+      "learning_rate": 0.00021558120362737016,
+      "loss": 4.0672,
       "step": 4309
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.467436108821105e-05,
-      "loss": 0.636,
+      "learning_rate": 0.00021537510305028853,
+      "loss": 4.1773,
       "step": 4310
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.461253091508655e-05,
-      "loss": 0.8955,
+      "learning_rate": 0.00021516900247320691,
+      "loss": 4.252,
       "step": 4311
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.455070074196207e-05,
-      "loss": 0.9427,
+      "learning_rate": 0.0002149629018961253,
+      "loss": 4.163,
       "step": 4312
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.448887056883758e-05,
-      "loss": 1.8502,
+      "learning_rate": 0.0002147568013190437,
+      "loss": 4.0419,
       "step": 4313
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.44270403957131e-05,
-      "loss": 0.861,
+      "learning_rate": 0.00021455070074196208,
+      "loss": 3.884,
       "step": 4314
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.436521022258862e-05,
-      "loss": 0.6008,
+      "learning_rate": 0.00021434460016488046,
+      "loss": 4.0691,
       "step": 4315
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.430338004946413e-05,
-      "loss": 0.8569,
+      "learning_rate": 0.00021413849958779884,
+      "loss": 3.6671,
       "step": 4316
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.424154987633965e-05,
-      "loss": 0.6407,
+      "learning_rate": 0.00021393239901071725,
+      "loss": 3.4863,
       "step": 4317
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.417971970321517e-05,
-      "loss": 1.4529,
+      "learning_rate": 0.00021372629843363563,
+      "loss": 3.971,
       "step": 4318
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.411788953009068e-05,
-      "loss": 0.6873,
+      "learning_rate": 0.000213520197856554,
+      "loss": 4.4743,
       "step": 4319
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.405605935696619e-05,
-      "loss": 1.0775,
+      "learning_rate": 0.00021331409727947238,
+      "loss": 4.534,
       "step": 4320
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.39942291838417e-05,
-      "loss": 0.5443,
+      "learning_rate": 0.0002131079967023908,
+      "loss": 4.0873,
       "step": 4321
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.393239901071723e-05,
-      "loss": 0.587,
+      "learning_rate": 0.00021290189612530917,
+      "loss": 4.4119,
       "step": 4322
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.387056883759275e-05,
-      "loss": 0.6451,
+      "learning_rate": 0.00021269579554822755,
+      "loss": 4.189,
       "step": 4323
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.380873866446825e-05,
-      "loss": 0.6211,
+      "learning_rate": 0.00021248969497114593,
+      "loss": 4.1494,
       "step": 4324
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.374690849134377e-05,
-      "loss": 0.5693,
+      "learning_rate": 0.0002122835943940643,
+      "loss": 3.9685,
       "step": 4325
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.368507831821928e-05,
-      "loss": 0.8025,
+      "learning_rate": 0.0002120774938169827,
+      "loss": 4.4685,
       "step": 4326
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.36232481450948e-05,
-      "loss": 0.7969,
+      "learning_rate": 0.00021187139323990107,
+      "loss": 4.4001,
       "step": 4327
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.356141797197032e-05,
-      "loss": 0.5909,
+      "learning_rate": 0.00021166529266281945,
+      "loss": 3.8996,
       "step": 4328
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.349958779884582e-05,
-      "loss": 0.7125,
+      "learning_rate": 0.00021145919208573786,
+      "loss": 3.9122,
       "step": 4329
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.343775762572135e-05,
-      "loss": 0.5376,
+      "learning_rate": 0.00021125309150865623,
+      "loss": 3.8971,
       "step": 4330
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.337592745259687e-05,
-      "loss": 0.7419,
+      "learning_rate": 0.00021104699093157461,
+      "loss": 4.2987,
       "step": 4331
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.331409727947238e-05,
-      "loss": 0.8415,
+      "learning_rate": 0.000210840890354493,
+      "loss": 4.3142,
       "step": 4332
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.325226710634789e-05,
-      "loss": 0.7286,
+      "learning_rate": 0.00021063478977741137,
+      "loss": 4.117,
       "step": 4333
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.31904369332234e-05,
-      "loss": 0.689,
+      "learning_rate": 0.00021042868920032975,
+      "loss": 4.1344,
       "step": 4334
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.312860676009892e-05,
-      "loss": 0.6194,
+      "learning_rate": 0.00021022258862324813,
+      "loss": 4.1553,
       "step": 4335
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.306677658697444e-05,
-      "loss": 0.8395,
+      "learning_rate": 0.0002100164880461665,
+      "loss": 4.3156,
       "step": 4336
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.300494641384995e-05,
-      "loss": 0.505,
+      "learning_rate": 0.00020981038746908492,
+      "loss": 3.9727,
       "step": 4337
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.294311624072547e-05,
-      "loss": 0.7368,
+      "learning_rate": 0.0002096042868920033,
+      "loss": 4.224,
       "step": 4338
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.288128606760099e-05,
-      "loss": 0.7676,
+      "learning_rate": 0.00020939818631492168,
+      "loss": 4.668,
       "step": 4339
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.28194558944765e-05,
-      "loss": 0.4953,
+      "learning_rate": 0.00020919208573784006,
+      "loss": 4.1099,
       "step": 4340
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.275762572135202e-05,
-      "loss": 0.6061,
+      "learning_rate": 0.00020898598516075846,
+      "loss": 4.0461,
       "step": 4341
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.269579554822752e-05,
-      "loss": 0.5589,
+      "learning_rate": 0.00020877988458367684,
+      "loss": 4.0291,
       "step": 4342
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.263396537510304e-05,
-      "loss": 0.5059,
+      "learning_rate": 0.00020857378400659522,
+      "loss": 4.0483,
       "step": 4343
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.257213520197856e-05,
-      "loss": 0.5793,
+      "learning_rate": 0.0002083676834295136,
+      "loss": 4.3188,
       "step": 4344
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.251030502885407e-05,
-      "loss": 0.6119,
+      "learning_rate": 0.000208161582852432,
+      "loss": 4.2807,
       "step": 4345
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.244847485572959e-05,
-      "loss": 0.4308,
+      "learning_rate": 0.0002079554822753504,
+      "loss": 4.0702,
       "step": 4346
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.23866446826051e-05,
-      "loss": 0.5375,
+      "learning_rate": 0.00020774938169826877,
+      "loss": 4.2158,
       "step": 4347
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.232481450948062e-05,
-      "loss": 0.5339,
+      "learning_rate": 0.00020754328112118715,
+      "loss": 3.8323,
       "step": 4348
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.226298433635614e-05,
-      "loss": 0.7879,
+      "learning_rate": 0.00020733718054410553,
+      "loss": 4.178,
       "step": 4349
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.220115416323166e-05,
-      "loss": 0.5375,
+      "learning_rate": 0.0002071310799670239,
+      "loss": 3.981,
       "step": 4350
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.213932399010716e-05,
-      "loss": 0.4801,
+      "learning_rate": 0.0002069249793899423,
+      "loss": 3.7945,
       "step": 4351
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.207749381698268e-05,
-      "loss": 0.5647,
+      "learning_rate": 0.00020671887881286067,
+      "loss": 3.8524,
       "step": 4352
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.201566364385819e-05,
-      "loss": 0.7182,
+      "learning_rate": 0.00020651277823577907,
+      "loss": 4.0307,
       "step": 4353
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.195383347073372e-05,
-      "loss": 0.7379,
+      "learning_rate": 0.00020630667765869745,
+      "loss": 3.9921,
       "step": 4354
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.189200329760923e-05,
-      "loss": 0.4785,
+      "learning_rate": 0.00020610057708161583,
+      "loss": 3.9507,
       "step": 4355
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.183017312448474e-05,
-      "loss": 0.5898,
+      "learning_rate": 0.0002058944765045342,
+      "loss": 4.1041,
       "step": 4356
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.176834295136026e-05,
-      "loss": 0.5954,
+      "learning_rate": 0.0002056883759274526,
+      "loss": 4.1293,
       "step": 4357
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.170651277823577e-05,
-      "loss": 0.607,
+      "learning_rate": 0.00020548227535037097,
+      "loss": 3.8735,
       "step": 4358
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.164468260511129e-05,
-      "loss": 0.5469,
+      "learning_rate": 0.00020527617477328935,
+      "loss": 4.2201,
       "step": 4359
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.15828524319868e-05,
-      "loss": 0.6523,
+      "learning_rate": 0.00020507007419620773,
+      "loss": 4.1878,
       "step": 4360
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.152102225886231e-05,
-      "loss": 0.5033,
+      "learning_rate": 0.00020486397361912614,
+      "loss": 4.0328,
       "step": 4361
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.145919208573784e-05,
-      "loss": 0.7534,
+      "learning_rate": 0.00020465787304204452,
+      "loss": 4.0569,
       "step": 4362
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.139736191261336e-05,
-      "loss": 0.5159,
+      "learning_rate": 0.0002044517724649629,
+      "loss": 3.8381,
       "step": 4363
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.133553173948886e-05,
-      "loss": 0.6199,
+      "learning_rate": 0.0002042456718878813,
+      "loss": 3.9157,
       "step": 4364
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.127370156636438e-05,
-      "loss": 0.668,
+      "learning_rate": 0.00020403957131079968,
+      "loss": 3.924,
       "step": 4365
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.12118713932399e-05,
-      "loss": 0.4411,
+      "learning_rate": 0.00020383347073371806,
+      "loss": 3.4105,
       "step": 4366
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.115004122011541e-05,
-      "loss": 0.5445,
+      "learning_rate": 0.00020362737015663644,
+      "loss": 3.723,
       "step": 4367
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.108821104699093e-05,
-      "loss": 0.7114,
+      "learning_rate": 0.00020342126957955485,
+      "loss": 2.9416,
       "step": 4368
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.1026380873866444e-05,
-      "loss": 1.1683,
+      "learning_rate": 0.00020321516900247323,
+      "loss": 4.4254,
       "step": 4369
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.096455070074196e-05,
-      "loss": 0.7845,
+      "learning_rate": 0.0002030090684253916,
+      "loss": 4.253,
       "step": 4370
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.090272052761748e-05,
-      "loss": 0.7969,
+      "learning_rate": 0.00020280296784831,
+      "loss": 4.1225,
       "step": 4371
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.084089035449299e-05,
-      "loss": 0.95,
+      "learning_rate": 0.00020259686727122837,
+      "loss": 4.199,
       "step": 4372
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.07790601813685e-05,
-      "loss": 0.5985,
+      "learning_rate": 0.00020239076669414675,
+      "loss": 3.975,
       "step": 4373
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.071723000824402e-05,
-      "loss": 0.7091,
+      "learning_rate": 0.00020218466611706513,
+      "loss": 4.3805,
       "step": 4374
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.065539983511953e-05,
-      "loss": 0.6615,
+      "learning_rate": 0.0002019785655399835,
+      "loss": 4.2384,
       "step": 4375
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.0593569661995046e-05,
-      "loss": 0.5962,
+      "learning_rate": 0.0002017724649629019,
+      "loss": 4.4688,
       "step": 4376
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.053173948887057e-05,
-      "loss": 0.5157,
+      "learning_rate": 0.0002015663643858203,
+      "loss": 4.1422,
       "step": 4377
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.046990931574608e-05,
-      "loss": 0.8782,
+      "learning_rate": 0.00020136026380873867,
+      "loss": 4.2854,
       "step": 4378
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.0408079142621596e-05,
-      "loss": 0.8945,
+      "learning_rate": 0.00020115416323165705,
+      "loss": 4.4665,
       "step": 4379
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.034624896949711e-05,
-      "loss": 0.5431,
+      "learning_rate": 0.00020094806265457543,
+      "loss": 4.1536,
       "step": 4380
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.028441879637262e-05,
-      "loss": 0.493,
+      "learning_rate": 0.0002007419620774938,
+      "loss": 3.9008,
       "step": 4381
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.022258862324814e-05,
-      "loss": 0.6601,
+      "learning_rate": 0.0002005358615004122,
+      "loss": 3.98,
       "step": 4382
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.0160758450123656e-05,
-      "loss": 0.6354,
+      "learning_rate": 0.00020032976092333057,
+      "loss": 4.0527,
       "step": 4383
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.0098928276999165e-05,
-      "loss": 0.6059,
+      "learning_rate": 0.00020012366034624898,
+      "loss": 4.4258,
       "step": 4384
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.003709810387469e-05,
-      "loss": 0.7857,
+      "learning_rate": 0.00019991755976916736,
+      "loss": 4.3031,
       "step": 4385
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.9975267930750205e-05,
-      "loss": 0.5738,
+      "learning_rate": 0.00019971145919208574,
+      "loss": 3.924,
       "step": 4386
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.9913437757625715e-05,
-      "loss": 0.5845,
+      "learning_rate": 0.00019950535861500414,
+      "loss": 4.0057,
       "step": 4387
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.985160758450123e-05,
-      "loss": 0.596,
+      "learning_rate": 0.00019929925803792252,
+      "loss": 4.1019,
       "step": 4388
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.978977741137675e-05,
-      "loss": 0.717,
+      "learning_rate": 0.0001990931574608409,
+      "loss": 4.1623,
       "step": 4389
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.972794723825226e-05,
-      "loss": 0.7027,
+      "learning_rate": 0.00019888705688375928,
+      "loss": 4.0478,
       "step": 4390
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.9666117065127775e-05,
-      "loss": 0.5131,
+      "learning_rate": 0.00019868095630667766,
+      "loss": 3.8388,
       "step": 4391
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.960428689200329e-05,
-      "loss": 0.6801,
+      "learning_rate": 0.00019847485572959607,
+      "loss": 4.2833,
       "step": 4392
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.9542456718878815e-05,
-      "loss": 0.4856,
+      "learning_rate": 0.00019826875515251445,
+      "loss": 4.0378,
       "step": 4393
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.9480626545754325e-05,
-      "loss": 0.5228,
+      "learning_rate": 0.00019806265457543283,
+      "loss": 4.0154,
       "step": 4394
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.941879637262984e-05,
-      "loss": 0.7635,
+      "learning_rate": 0.0001978565539983512,
+      "loss": 4.1751,
       "step": 4395
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.935696619950535e-05,
-      "loss": 0.6106,
+      "learning_rate": 0.00019765045342126959,
+      "loss": 3.7957,
       "step": 4396
     },
     {
       "epoch": 2.46,
-      "learning_rate": 5.929513602638087e-05,
-      "loss": 0.6899,
+      "learning_rate": 0.00019744435284418797,
+      "loss": 4.3307,
       "step": 4397
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.9233305853256384e-05,
-      "loss": 0.6318,
+      "learning_rate": 0.00019723825226710634,
+      "loss": 4.1971,
       "step": 4398
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.9171475680131894e-05,
-      "loss": 0.5392,
+      "learning_rate": 0.00019703215169002472,
+      "loss": 4.0775,
       "step": 4399
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.910964550700741e-05,
-      "loss": 0.4787,
+      "learning_rate": 0.00019682605111294313,
+      "loss": 3.9886,
       "step": 4400
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.9047815333882934e-05,
-      "loss": 0.4874,
+      "learning_rate": 0.0001966199505358615,
+      "loss": 3.9144,
       "step": 4401
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.898598516075845e-05,
-      "loss": 0.4457,
+      "learning_rate": 0.0001964138499587799,
+      "loss": 3.9152,
       "step": 4402
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.892415498763396e-05,
-      "loss": 0.5146,
+      "learning_rate": 0.00019620774938169827,
+      "loss": 3.8627,
       "step": 4403
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.886232481450948e-05,
-      "loss": 0.6994,
+      "learning_rate": 0.00019600164880461665,
+      "loss": 4.3605,
       "step": 4404
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.8800494641384993e-05,
-      "loss": 0.5641,
+      "learning_rate": 0.00019579554822753503,
+      "loss": 3.942,
       "step": 4405
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.87386644682605e-05,
-      "loss": 0.4032,
+      "learning_rate": 0.0001955894476504534,
+      "loss": 3.8692,
       "step": 4406
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.867683429513602e-05,
-      "loss": 0.6829,
+      "learning_rate": 0.0001953833470733718,
+      "loss": 4.1607,
       "step": 4407
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.861500412201153e-05,
-      "loss": 0.4751,
+      "learning_rate": 0.0001951772464962902,
+      "loss": 3.7447,
       "step": 4408
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.855317394888705e-05,
-      "loss": 0.3901,
+      "learning_rate": 0.00019497114591920857,
+      "loss": 3.8476,
       "step": 4409
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.849134377576257e-05,
-      "loss": 0.8171,
+      "learning_rate": 0.00019476504534212695,
+      "loss": 4.0081,
       "step": 4410
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.8429513602638086e-05,
-      "loss": 0.6643,
+      "learning_rate": 0.00019455894476504536,
+      "loss": 4.1192,
       "step": 4411
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.8367683429513596e-05,
-      "loss": 0.4921,
+      "learning_rate": 0.00019435284418796374,
+      "loss": 3.8437,
       "step": 4412
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.830585325638911e-05,
-      "loss": 0.574,
+      "learning_rate": 0.00019414674361088212,
+      "loss": 3.9487,
       "step": 4413
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.824402308326463e-05,
-      "loss": 0.9775,
+      "learning_rate": 0.0001939406430338005,
+      "loss": 3.8404,
       "step": 4414
     },
     {
       "epoch": 2.47,
-      "learning_rate": 5.818219291014014e-05,
-      "loss": 0.637,
+      "learning_rate": 0.00019373454245671888,
+      "loss": 3.7573,
       "step": 4415
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.8120362737015656e-05,
-      "loss": 0.56,
+      "learning_rate": 0.00019352844187963729,
+      "loss": 3.5396,
       "step": 4416
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.805853256389118e-05,
-      "loss": 1.1831,
+      "learning_rate": 0.00019332234130255567,
+      "loss": 3.7796,
       "step": 4417
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.799670239076669e-05,
-      "loss": 1.4152,
+      "learning_rate": 0.00019311624072547404,
+      "loss": 3.6335,
       "step": 4418
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.7934872217642205e-05,
-      "loss": 0.6738,
+      "learning_rate": 0.00019291014014839242,
+      "loss": 4.1384,
       "step": 4419
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.787304204451772e-05,
-      "loss": 0.7504,
+      "learning_rate": 0.0001927040395713108,
+      "loss": 4.0137,
       "step": 4420
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.781121187139323e-05,
-      "loss": 0.6394,
+      "learning_rate": 0.00019249793899422918,
+      "loss": 4.3262,
       "step": 4421
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.774938169826875e-05,
-      "loss": 0.7092,
+      "learning_rate": 0.00019229183841714756,
+      "loss": 4.0526,
       "step": 4422
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.7687551525144265e-05,
-      "loss": 0.7196,
+      "learning_rate": 0.00019208573784006594,
+      "loss": 4.165,
       "step": 4423
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.7625721352019775e-05,
-      "loss": 0.4925,
+      "learning_rate": 0.00019187963726298435,
+      "loss": 3.9193,
       "step": 4424
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.75638911788953e-05,
-      "loss": 0.6831,
+      "learning_rate": 0.00019167353668590273,
+      "loss": 4.3552,
       "step": 4425
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.7502061005770815e-05,
-      "loss": 0.774,
+      "learning_rate": 0.0001914674361088211,
+      "loss": 4.1973,
       "step": 4426
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.7440230832646325e-05,
-      "loss": 0.781,
+      "learning_rate": 0.0001912613355317395,
+      "loss": 4.2136,
       "step": 4427
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.737840065952184e-05,
-      "loss": 0.7338,
+      "learning_rate": 0.00019105523495465787,
+      "loss": 4.1574,
       "step": 4428
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.731657048639736e-05,
-      "loss": 0.6138,
+      "learning_rate": 0.00019084913437757625,
+      "loss": 4.375,
       "step": 4429
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.725474031327287e-05,
-      "loss": 0.6329,
+      "learning_rate": 0.00019064303380049463,
+      "loss": 4.0145,
       "step": 4430
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.7192910140148384e-05,
-      "loss": 0.4225,
+      "learning_rate": 0.000190436933223413,
+      "loss": 4.0281,
       "step": 4431
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.71310799670239e-05,
-      "loss": 0.5809,
+      "learning_rate": 0.0001902308326463314,
+      "loss": 3.9624,
       "step": 4432
     },
     {
       "epoch": 2.48,
-      "learning_rate": 5.7069249793899424e-05,
-      "loss": 0.573,
+      "learning_rate": 0.0001900247320692498,
+      "loss": 4.1391,
       "step": 4433
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.7007419620774934e-05,
-      "loss": 0.5846,
+      "learning_rate": 0.0001898186314921682,
+      "loss": 4.3172,
       "step": 4434
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.694558944765045e-05,
-      "loss": 0.5749,
+      "learning_rate": 0.00018961253091508658,
+      "loss": 4.17,
       "step": 4435
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.688375927452597e-05,
-      "loss": 0.4551,
+      "learning_rate": 0.00018940643033800496,
+      "loss": 4.0972,
       "step": 4436
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.682192910140148e-05,
-      "loss": 0.4542,
+      "learning_rate": 0.00018920032976092334,
+      "loss": 3.8981,
       "step": 4437
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.6760098928276993e-05,
-      "loss": 0.6524,
+      "learning_rate": 0.00018899422918384172,
+      "loss": 4.2449,
       "step": 4438
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.669826875515251e-05,
-      "loss": 0.7845,
+      "learning_rate": 0.0001887881286067601,
+      "loss": 4.3083,
       "step": 4439
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.663643858202802e-05,
-      "loss": 0.7949,
+      "learning_rate": 0.0001885820280296785,
+      "loss": 4.2294,
       "step": 4440
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.657460840890354e-05,
-      "loss": 0.5958,
+      "learning_rate": 0.00018837592745259688,
+      "loss": 4.1225,
       "step": 4441
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.651277823577906e-05,
-      "loss": 0.4844,
+      "learning_rate": 0.00018816982687551526,
+      "loss": 3.9648,
       "step": 4442
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.645094806265457e-05,
-      "loss": 0.6933,
+      "learning_rate": 0.00018796372629843364,
+      "loss": 4.2298,
       "step": 4443
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.6389117889530086e-05,
-      "loss": 0.6436,
+      "learning_rate": 0.00018775762572135202,
+      "loss": 4.1792,
       "step": 4444
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.63272877164056e-05,
-      "loss": 0.5814,
+      "learning_rate": 0.0001875515251442704,
+      "loss": 4.2069,
       "step": 4445
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.626545754328111e-05,
-      "loss": 0.5575,
+      "learning_rate": 0.00018734542456718878,
+      "loss": 3.9598,
       "step": 4446
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.620362737015663e-05,
-      "loss": 0.5125,
+      "learning_rate": 0.00018713932399010716,
+      "loss": 3.9503,
       "step": 4447
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.6141797197032146e-05,
-      "loss": 0.6898,
+      "learning_rate": 0.00018693322341302557,
+      "loss": 4.3953,
       "step": 4448
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.607996702390766e-05,
-      "loss": 0.7557,
+      "learning_rate": 0.00018672712283594395,
+      "loss": 4.2098,
       "step": 4449
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.601813685078318e-05,
-      "loss": 0.5462,
+      "learning_rate": 0.00018652102225886233,
+      "loss": 3.9062,
       "step": 4450
     },
     {
       "epoch": 2.49,
-      "learning_rate": 5.5956306677658695e-05,
-      "loss": 0.4893,
+      "learning_rate": 0.0001863149216817807,
+      "loss": 4.1454,
       "step": 4451
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.5894476504534205e-05,
-      "loss": 0.5474,
+      "learning_rate": 0.00018610882110469909,
+      "loss": 4.0466,
       "step": 4452
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.583264633140972e-05,
-      "loss": 0.6475,
+      "learning_rate": 0.00018590272052761747,
+      "loss": 3.9341,
       "step": 4453
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.577081615828524e-05,
-      "loss": 0.5394,
+      "learning_rate": 0.00018569661995053585,
+      "loss": 4.1494,
       "step": 4454
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.570898598516075e-05,
-      "loss": 0.7626,
+      "learning_rate": 0.00018549051937345423,
+      "loss": 4.1228,
       "step": 4455
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.5647155812036265e-05,
-      "loss": 0.4851,
+      "learning_rate": 0.00018528441879637263,
+      "loss": 3.9948,
       "step": 4456
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.558532563891179e-05,
-      "loss": 0.6686,
+      "learning_rate": 0.00018507831821929104,
+      "loss": 3.9963,
       "step": 4457
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.5523495465787305e-05,
-      "loss": 0.5683,
+      "learning_rate": 0.00018487221764220942,
+      "loss": 4.0268,
       "step": 4458
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.5461665292662815e-05,
-      "loss": 0.657,
+      "learning_rate": 0.0001846661170651278,
+      "loss": 4.4171,
       "step": 4459
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.539983511953833e-05,
-      "loss": 0.7796,
+      "learning_rate": 0.00018446001648804618,
+      "loss": 4.2467,
       "step": 4460
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.533800494641384e-05,
-      "loss": 0.371,
+      "learning_rate": 0.00018425391591096456,
+      "loss": 3.8718,
       "step": 4461
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.527617477328936e-05,
-      "loss": 0.4394,
+      "learning_rate": 0.00018404781533388294,
+      "loss": 3.6873,
       "step": 4462
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.5214344600164874e-05,
-      "loss": 0.5926,
+      "learning_rate": 0.00018384171475680132,
+      "loss": 3.7128,
       "step": 4463
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.5152514427040384e-05,
-      "loss": 0.518,
+      "learning_rate": 0.00018363561417971972,
+      "loss": 3.8094,
       "step": 4464
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.509068425391591e-05,
-      "loss": 1.0273,
+      "learning_rate": 0.0001834295136026381,
+      "loss": 3.9145,
       "step": 4465
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.5028854080791424e-05,
-      "loss": 0.6148,
+      "learning_rate": 0.00018322341302555648,
+      "loss": 3.9553,
       "step": 4466
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.496702390766694e-05,
-      "loss": 0.567,
+      "learning_rate": 0.00018301731244847486,
+      "loss": 3.7035,
       "step": 4467
     },
     {
       "epoch": 2.5,
-      "learning_rate": 5.490519373454245e-05,
-      "loss": 0.9855,
+      "learning_rate": 0.00018281121187139324,
+      "loss": 3.6201,
       "step": 4468
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.484336356141797e-05,
-      "loss": 0.6797,
+      "learning_rate": 0.00018260511129431162,
+      "loss": 4.1343,
       "step": 4469
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.4781533388293484e-05,
-      "loss": 0.6813,
+      "learning_rate": 0.00018239901071723,
+      "loss": 4.4102,
       "step": 4470
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.471970321516899e-05,
-      "loss": 0.7993,
+      "learning_rate": 0.00018219291014014838,
+      "loss": 4.172,
       "step": 4471
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.465787304204451e-05,
-      "loss": 1.0215,
+      "learning_rate": 0.0001819868095630668,
+      "loss": 4.4094,
       "step": 4472
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.459604286892003e-05,
-      "loss": 0.5475,
+      "learning_rate": 0.00018178070898598517,
+      "loss": 3.9854,
       "step": 4473
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.453421269579554e-05,
-      "loss": 0.4585,
+      "learning_rate": 0.00018157460840890355,
+      "loss": 4.1397,
       "step": 4474
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.447238252267106e-05,
-      "loss": 0.5766,
+      "learning_rate": 0.00018136850783182193,
+      "loss": 3.8177,
       "step": 4475
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.4410552349546576e-05,
-      "loss": 0.7794,
+      "learning_rate": 0.0001811624072547403,
+      "loss": 3.941,
       "step": 4476
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.4348722176422086e-05,
-      "loss": 0.6804,
+      "learning_rate": 0.00018095630667765868,
+      "loss": 4.1557,
       "step": 4477
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.42868920032976e-05,
-      "loss": 0.5286,
+      "learning_rate": 0.00018075020610057706,
+      "loss": 4.1745,
       "step": 4478
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.422506183017312e-05,
-      "loss": 0.6293,
+      "learning_rate": 0.00018054410552349547,
+      "loss": 4.4811,
       "step": 4479
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.4163231657048636e-05,
-      "loss": 0.6875,
+      "learning_rate": 0.00018033800494641385,
+      "loss": 4.0649,
       "step": 4480
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.410140148392415e-05,
-      "loss": 0.7421,
+      "learning_rate": 0.00018013190436933226,
+      "loss": 4.1412,
       "step": 4481
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.403957131079967e-05,
-      "loss": 0.6206,
+      "learning_rate": 0.00017992580379225064,
+      "loss": 4.0778,
       "step": 4482
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.397774113767518e-05,
-      "loss": 0.398,
+      "learning_rate": 0.00017971970321516902,
+      "loss": 3.8618,
       "step": 4483
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.3915910964550695e-05,
-      "loss": 0.5672,
+      "learning_rate": 0.0001795136026380874,
+      "loss": 3.8971,
       "step": 4484
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.385408079142621e-05,
-      "loss": 0.5163,
+      "learning_rate": 0.00017930750206100578,
+      "loss": 3.9643,
       "step": 4485
     },
     {
       "epoch": 2.51,
-      "learning_rate": 5.379225061830172e-05,
-      "loss": 0.6961,
+      "learning_rate": 0.00017910140148392416,
+      "loss": 4.0749,
       "step": 4486
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.373042044517724e-05,
-      "loss": 0.4097,
+      "learning_rate": 0.00017889530090684256,
+      "loss": 3.9006,
       "step": 4487
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.366859027205276e-05,
-      "loss": 0.5382,
+      "learning_rate": 0.00017868920032976094,
+      "loss": 4.1171,
       "step": 4488
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.360676009892828e-05,
-      "loss": 0.5708,
+      "learning_rate": 0.00017848309975267932,
+      "loss": 4.0268,
       "step": 4489
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.354492992580379e-05,
-      "loss": 0.6177,
+      "learning_rate": 0.0001782769991755977,
+      "loss": 4.1329,
       "step": 4490
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.3483099752679305e-05,
-      "loss": 0.6666,
+      "learning_rate": 0.00017807089859851608,
+      "loss": 4.3143,
       "step": 4491
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.3421269579554815e-05,
-      "loss": 0.5758,
+      "learning_rate": 0.00017786479802143446,
+      "loss": 4.2383,
       "step": 4492
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.335943940643033e-05,
-      "loss": 0.5833,
+      "learning_rate": 0.00017765869744435284,
+      "loss": 4.0743,
       "step": 4493
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.329760923330585e-05,
-      "loss": 0.7307,
+      "learning_rate": 0.00017745259686727122,
+      "loss": 4.1571,
       "step": 4494
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.323577906018136e-05,
-      "loss": 0.3851,
+      "learning_rate": 0.00017724649629018963,
+      "loss": 3.9069,
       "step": 4495
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.317394888705688e-05,
-      "loss": 0.5413,
+      "learning_rate": 0.000177040395713108,
+      "loss": 4.1684,
       "step": 4496
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.31121187139324e-05,
-      "loss": 0.505,
+      "learning_rate": 0.00017683429513602638,
+      "loss": 4.0489,
       "step": 4497
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.3050288540807914e-05,
-      "loss": 0.4924,
+      "learning_rate": 0.00017662819455894476,
+      "loss": 3.9246,
       "step": 4498
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.2988458367683424e-05,
-      "loss": 0.5359,
+      "learning_rate": 0.00017642209398186314,
+      "loss": 3.8547,
       "step": 4499
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.292662819455894e-05,
-      "loss": 0.4536,
+      "learning_rate": 0.00017621599340478152,
+      "loss": 3.9933,
       "step": 4500
     },
     {
       "epoch": 2.52,
-      "eval_loss": 1.1965099573135376,
-      "eval_runtime": 1250.2219,
-      "eval_samples_per_second": 2.113,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.069873066243554,
+      "eval_loss": 5.2800092697143555,
+      "eval_runtime": 1256.1512,
+      "eval_samples_per_second": 2.103,
+      "eval_steps_per_second": 0.264,
+      "eval_wer": 1.9741372471241572,
       "step": 4500
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.286479802143446e-05,
-      "loss": 0.7909,
+      "learning_rate": 0.0001760098928276999,
+      "loss": 4.2388,
       "step": 4501
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.280296784830997e-05,
-      "loss": 0.705,
+      "learning_rate": 0.00017580379225061828,
+      "loss": 4.0934,
       "step": 4502
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.2741137675185483e-05,
-      "loss": 0.8439,
+      "learning_rate": 0.0001755976916735367,
+      "loss": 4.4195,
       "step": 4503
     },
     {
       "epoch": 2.52,
-      "learning_rate": 5.267930750206101e-05,
-      "loss": 0.6159,
+      "learning_rate": 0.0001753915910964551,
+      "loss": 3.9851,
       "step": 4504
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.2617477328936517e-05,
-      "loss": 0.5645,
+      "learning_rate": 0.00017518549051937348,
+      "loss": 4.2827,
       "step": 4505
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.255564715581203e-05,
-      "loss": 0.4492,
+      "learning_rate": 0.00017497938994229186,
+      "loss": 3.7112,
       "step": 4506
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.249381698268755e-05,
-      "loss": 0.4406,
+      "learning_rate": 0.00017477328936521023,
+      "loss": 3.8002,
       "step": 4507
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.243198680956306e-05,
-      "loss": 0.4415,
+      "learning_rate": 0.00017456718878812861,
+      "loss": 4.237,
       "step": 4508
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.2370156636438576e-05,
-      "loss": 0.7071,
+      "learning_rate": 0.000174361088211047,
+      "loss": 4.2165,
       "step": 4509
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.230832646331409e-05,
-      "loss": 0.4849,
+      "learning_rate": 0.00017415498763396537,
+      "loss": 3.9822,
       "step": 4510
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.22464962901896e-05,
-      "loss": 0.4733,
+      "learning_rate": 0.00017394888705688378,
+      "loss": 3.8946,
       "step": 4511
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.2184666117065126e-05,
-      "loss": 0.539,
+      "learning_rate": 0.00017374278647980216,
+      "loss": 4.0408,
       "step": 4512
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.212283594394064e-05,
-      "loss": 0.8359,
+      "learning_rate": 0.00017353668590272054,
+      "loss": 4.0769,
       "step": 4513
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.206100577081615e-05,
-      "loss": 1.023,
+      "learning_rate": 0.00017333058532563892,
+      "loss": 4.1706,
       "step": 4514
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.199917559769167e-05,
-      "loss": 0.5929,
+      "learning_rate": 0.0001731244847485573,
+      "loss": 3.707,
       "step": 4515
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.1937345424567185e-05,
-      "loss": 0.8036,
+      "learning_rate": 0.00017291838417147568,
+      "loss": 3.446,
       "step": 4516
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.1875515251442695e-05,
-      "loss": 0.1783,
+      "learning_rate": 0.00017271228359439406,
+      "loss": 3.3464,
       "step": 4517
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.181368507831821e-05,
-      "loss": 0.2514,
+      "learning_rate": 0.00017250618301731244,
+      "loss": 3.0425,
       "step": 4518
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.175185490519373e-05,
-      "loss": 0.9932,
+      "learning_rate": 0.00017230008244023084,
+      "loss": 4.6893,
       "step": 4519
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.169002473206925e-05,
-      "loss": 0.9177,
+      "learning_rate": 0.00017209398186314922,
+      "loss": 4.1628,
       "step": 4520
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.162819455894476e-05,
-      "loss": 0.9789,
+      "learning_rate": 0.0001718878812860676,
+      "loss": 4.3683,
       "step": 4521
     },
     {
       "epoch": 2.53,
-      "learning_rate": 5.156636438582028e-05,
-      "loss": 0.854,
+      "learning_rate": 0.00017168178070898598,
+      "loss": 4.0764,
       "step": 4522
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.1504534212695795e-05,
-      "loss": 0.6063,
+      "learning_rate": 0.00017147568013190436,
+      "loss": 3.9563,
       "step": 4523
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.1442704039571305e-05,
-      "loss": 0.8088,
+      "learning_rate": 0.00017126957955482274,
+      "loss": 4.2778,
       "step": 4524
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.138087386644682e-05,
-      "loss": 0.9998,
+      "learning_rate": 0.00017106347897774112,
+      "loss": 4.1989,
       "step": 4525
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.131904369332233e-05,
-      "loss": 0.6275,
+      "learning_rate": 0.0001708573784006595,
+      "loss": 4.2573,
       "step": 4526
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.125721352019785e-05,
-      "loss": 0.6751,
+      "learning_rate": 0.00017065127782357793,
+      "loss": 4.2192,
       "step": 4527
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.119538334707337e-05,
-      "loss": 0.7182,
+      "learning_rate": 0.00017044517724649631,
+      "loss": 4.3957,
       "step": 4528
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.113355317394889e-05,
-      "loss": 0.5832,
+      "learning_rate": 0.0001702390766694147,
+      "loss": 3.9751,
       "step": 4529
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.10717230008244e-05,
-      "loss": 0.6226,
+      "learning_rate": 0.00017003297609233307,
+      "loss": 4.1727,
       "step": 4530
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.1009892827699914e-05,
-      "loss": 0.7702,
+      "learning_rate": 0.00016982687551525145,
+      "loss": 4.1672,
       "step": 4531
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.094806265457543e-05,
-      "loss": 0.7063,
+      "learning_rate": 0.00016962077493816983,
+      "loss": 4.1682,
       "step": 4532
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.088623248145094e-05,
-      "loss": 0.4582,
+      "learning_rate": 0.0001694146743610882,
+      "loss": 3.8833,
       "step": 4533
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.082440230832646e-05,
-      "loss": 0.593,
+      "learning_rate": 0.0001692085737840066,
+      "loss": 4.1081,
       "step": 4534
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.0762572135201974e-05,
-      "loss": 0.5642,
+      "learning_rate": 0.000169002473206925,
+      "loss": 3.7789,
       "step": 4535
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.070074196207749e-05,
-      "loss": 0.7173,
+      "learning_rate": 0.00016879637262984338,
+      "loss": 4.2123,
       "step": 4536
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.063891178895301e-05,
-      "loss": 0.654,
+      "learning_rate": 0.00016859027205276176,
+      "loss": 4.4283,
       "step": 4537
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.057708161582852e-05,
-      "loss": 0.675,
+      "learning_rate": 0.00016838417147568014,
+      "loss": 4.5046,
       "step": 4538
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.051525144270403e-05,
-      "loss": 0.5771,
+      "learning_rate": 0.00016817807089859852,
+      "loss": 4.1229,
       "step": 4539
     },
     {
       "epoch": 2.54,
-      "learning_rate": 5.045342126957955e-05,
-      "loss": 0.535,
+      "learning_rate": 0.0001679719703215169,
+      "loss": 4.0794,
       "step": 4540
     },
     {
       "epoch": 2.55,
-      "learning_rate": 5.0391591096455066e-05,
-      "loss": 0.603,
+      "learning_rate": 0.00016776586974443528,
+      "loss": 4.0489,
       "step": 4541
     },
     {
       "epoch": 2.55,
-      "learning_rate": 5.0329760923330576e-05,
-      "loss": 0.3752,
+      "learning_rate": 0.00016755976916735366,
+      "loss": 3.844,
       "step": 4542
     },
     {
       "epoch": 2.55,
-      "learning_rate": 5.026793075020609e-05,
-      "loss": 0.6568,
+      "learning_rate": 0.00016735366859027206,
+      "loss": 4.0522,
       "step": 4543
     },
     {
       "epoch": 2.55,
-      "learning_rate": 5.0206100577081616e-05,
-      "loss": 0.5144,
+      "learning_rate": 0.00016714756801319044,
+      "loss": 4.0933,
       "step": 4544
     },
     {
       "epoch": 2.55,
-      "learning_rate": 5.0144270403957126e-05,
-      "loss": 0.5009,
+      "learning_rate": 0.00016694146743610882,
+      "loss": 3.9495,
       "step": 4545
     },
     {
       "epoch": 2.55,
-      "learning_rate": 5.008244023083264e-05,
-      "loss": 0.5208,
+      "learning_rate": 0.0001667353668590272,
+      "loss": 3.7867,
       "step": 4546
     },
     {
       "epoch": 2.55,
-      "learning_rate": 5.002061005770816e-05,
-      "loss": 0.5435,
+      "learning_rate": 0.00016652926628194558,
+      "loss": 4.242,
       "step": 4547
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.995877988458367e-05,
-      "loss": 0.7616,
+      "learning_rate": 0.00016632316570486396,
+      "loss": 4.3095,
       "step": 4548
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.9896949711459185e-05,
-      "loss": 0.4241,
+      "learning_rate": 0.00016611706512778234,
+      "loss": 3.9004,
       "step": 4549
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.98351195383347e-05,
-      "loss": 0.5358,
+      "learning_rate": 0.00016591096455070072,
+      "loss": 4.032,
       "step": 4550
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.977328936521021e-05,
-      "loss": 0.5214,
+      "learning_rate": 0.00016570486397361915,
+      "loss": 4.0965,
       "step": 4551
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.9711459192085735e-05,
-      "loss": 0.5162,
+      "learning_rate": 0.00016549876339653753,
+      "loss": 4.0481,
       "step": 4552
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.964962901896125e-05,
-      "loss": 0.5749,
+      "learning_rate": 0.0001652926628194559,
+      "loss": 3.9588,
       "step": 4553
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.958779884583677e-05,
-      "loss": 0.6166,
+      "learning_rate": 0.0001650865622423743,
+      "loss": 4.219,
       "step": 4554
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.952596867271228e-05,
-      "loss": 0.4056,
+      "learning_rate": 0.00016488046166529267,
+      "loss": 3.8771,
       "step": 4555
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.9464138499587795e-05,
-      "loss": 0.3992,
+      "learning_rate": 0.00016467436108821105,
+      "loss": 3.8846,
       "step": 4556
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.940230832646331e-05,
-      "loss": 0.4989,
+      "learning_rate": 0.00016446826051112943,
+      "loss": 3.9496,
       "step": 4557
     },
     {
       "epoch": 2.55,
-      "learning_rate": 4.934047815333882e-05,
-      "loss": 0.5398,
+      "learning_rate": 0.0001642621599340478,
+      "loss": 3.8547,
       "step": 4558
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.927864798021434e-05,
-      "loss": 0.4999,
+      "learning_rate": 0.00016405605935696622,
+      "loss": 3.7413,
       "step": 4559
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.921681780708986e-05,
-      "loss": 0.4794,
+      "learning_rate": 0.0001638499587798846,
+      "loss": 3.8909,
       "step": 4560
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.915498763396537e-05,
-      "loss": 0.4085,
+      "learning_rate": 0.00016364385820280298,
+      "loss": 3.9593,
       "step": 4561
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.909315746084089e-05,
-      "loss": 0.5051,
+      "learning_rate": 0.00016343775762572136,
+      "loss": 3.7484,
       "step": 4562
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.9031327287716404e-05,
-      "loss": 0.4207,
+      "learning_rate": 0.00016323165704863974,
+      "loss": 3.6006,
       "step": 4563
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.8969497114591914e-05,
-      "loss": 0.4783,
+      "learning_rate": 0.00016302555647155812,
+      "loss": 3.73,
       "step": 4564
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.890766694146743e-05,
-      "loss": 0.6026,
+      "learning_rate": 0.0001628194558944765,
+      "loss": 3.7694,
       "step": 4565
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.884583676834295e-05,
-      "loss": 0.4547,
+      "learning_rate": 0.00016261335531739487,
+      "loss": 3.6685,
       "step": 4566
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.878400659521846e-05,
-      "loss": 0.2995,
+      "learning_rate": 0.00016240725474031328,
+      "loss": 3.2225,
       "step": 4567
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.872217642209398e-05,
-      "loss": 1.66,
+      "learning_rate": 0.00016220115416323166,
+      "loss": 3.7021,
       "step": 4568
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.86603462489695e-05,
-      "loss": 0.734,
+      "learning_rate": 0.00016199505358615004,
+      "loss": 3.9801,
       "step": 4569
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.859851607584501e-05,
-      "loss": 0.55,
+      "learning_rate": 0.00016178895300906842,
+      "loss": 4.0441,
       "step": 4570
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.853668590272052e-05,
-      "loss": 0.7989,
+      "learning_rate": 0.0001615828524319868,
+      "loss": 4.1885,
       "step": 4571
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.847485572959604e-05,
-      "loss": 0.6823,
+      "learning_rate": 0.00016137675185490518,
+      "loss": 3.9689,
       "step": 4572
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.841302555647155e-05,
-      "loss": 0.6688,
+      "learning_rate": 0.00016117065127782356,
+      "loss": 4.1468,
       "step": 4573
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.8351195383347066e-05,
-      "loss": 0.625,
+      "learning_rate": 0.00016096455070074197,
+      "loss": 4.0761,
       "step": 4574
     },
     {
       "epoch": 2.56,
-      "learning_rate": 4.828936521022258e-05,
-      "loss": 0.5521,
+      "learning_rate": 0.00016075845012366037,
+      "loss": 3.9796,
       "step": 4575
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.8227535037098106e-05,
-      "loss": 0.6015,
+      "learning_rate": 0.00016055234954657875,
+      "loss": 4.192,
       "step": 4576
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.8165704863973616e-05,
-      "loss": 0.6116,
+      "learning_rate": 0.00016034624896949713,
+      "loss": 4.1491,
       "step": 4577
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.810387469084913e-05,
-      "loss": 0.5949,
+      "learning_rate": 0.0001601401483924155,
+      "loss": 4.1111,
       "step": 4578
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.804204451772464e-05,
-      "loss": 0.7906,
+      "learning_rate": 0.0001599340478153339,
+      "loss": 4.1536,
       "step": 4579
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.798021434460016e-05,
-      "loss": 0.4375,
+      "learning_rate": 0.00015972794723825227,
+      "loss": 3.8145,
       "step": 4580
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.7918384171475676e-05,
-      "loss": 0.9382,
+      "learning_rate": 0.00015952184666117065,
+      "loss": 4.2785,
       "step": 4581
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.7856553998351185e-05,
-      "loss": 0.4806,
+      "learning_rate": 0.00015931574608408903,
+      "loss": 3.8792,
       "step": 4582
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.77947238252267e-05,
-      "loss": 0.5971,
+      "learning_rate": 0.00015910964550700744,
+      "loss": 4.0433,
       "step": 4583
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.7732893652102225e-05,
-      "loss": 0.4752,
+      "learning_rate": 0.00015890354492992582,
+      "loss": 4.0371,
       "step": 4584
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.767106347897774e-05,
-      "loss": 0.56,
+      "learning_rate": 0.0001586974443528442,
+      "loss": 4.3614,
       "step": 4585
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.760923330585325e-05,
-      "loss": 0.5115,
+      "learning_rate": 0.00015849134377576257,
+      "loss": 4.0529,
       "step": 4586
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.754740313272877e-05,
-      "loss": 0.816,
+      "learning_rate": 0.00015828524319868095,
+      "loss": 4.5024,
       "step": 4587
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.7485572959604285e-05,
-      "loss": 0.612,
+      "learning_rate": 0.00015807914262159933,
+      "loss": 3.8636,
       "step": 4588
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.7423742786479795e-05,
-      "loss": 0.7054,
+      "learning_rate": 0.0001578730420445177,
+      "loss": 4.177,
       "step": 4589
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.736191261335531e-05,
-      "loss": 0.4716,
+      "learning_rate": 0.00015766694146743612,
+      "loss": 4.0446,
       "step": 4590
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.7300082440230835e-05,
-      "loss": 0.4488,
+      "learning_rate": 0.0001574608408903545,
+      "loss": 4.1794,
       "step": 4591
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.7238252267106344e-05,
-      "loss": 0.74,
+      "learning_rate": 0.00015725474031327288,
+      "loss": 4.2114,
       "step": 4592
     },
     {
       "epoch": 2.57,
-      "learning_rate": 4.717642209398186e-05,
-      "loss": 0.4962,
+      "learning_rate": 0.00015704863973619126,
+      "loss": 3.9537,
       "step": 4593
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.711459192085738e-05,
-      "loss": 0.4861,
+      "learning_rate": 0.00015684253915910964,
+      "loss": 4.1271,
       "step": 4594
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.705276174773289e-05,
-      "loss": 0.5495,
+      "learning_rate": 0.00015663643858202802,
+      "loss": 4.0257,
       "step": 4595
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.6990931574608404e-05,
-      "loss": 0.4405,
+      "learning_rate": 0.0001564303380049464,
+      "loss": 3.9439,
       "step": 4596
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.692910140148392e-05,
-      "loss": 0.4499,
+      "learning_rate": 0.00015622423742786478,
+      "loss": 3.8106,
       "step": 4597
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.686727122835943e-05,
-      "loss": 0.6707,
+      "learning_rate": 0.0001560181368507832,
+      "loss": 4.206,
       "step": 4598
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.6805441055234954e-05,
-      "loss": 0.4395,
+      "learning_rate": 0.0001558120362737016,
+      "loss": 3.8263,
       "step": 4599
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.674361088211047e-05,
-      "loss": 0.5179,
+      "learning_rate": 0.00015560593569661997,
+      "loss": 4.1825,
       "step": 4600
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.668178070898598e-05,
-      "loss": 0.4496,
+      "learning_rate": 0.00015539983511953835,
+      "loss": 4.086,
       "step": 4601
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.66199505358615e-05,
-      "loss": 0.5322,
+      "learning_rate": 0.00015519373454245673,
+      "loss": 3.6768,
       "step": 4602
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.655812036273701e-05,
-      "loss": 0.5304,
+      "learning_rate": 0.0001549876339653751,
+      "loss": 4.1132,
       "step": 4603
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.649629018961252e-05,
-      "loss": 0.3887,
+      "learning_rate": 0.0001547815333882935,
+      "loss": 3.7393,
       "step": 4604
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.643446001648804e-05,
-      "loss": 0.7016,
+      "learning_rate": 0.00015457543281121187,
+      "loss": 4.2305,
       "step": 4605
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.6372629843363556e-05,
-      "loss": 1.4081,
+      "learning_rate": 0.00015436933223413027,
+      "loss": 4.2687,
       "step": 4606
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.631079967023908e-05,
-      "loss": 0.5309,
+      "learning_rate": 0.00015416323165704865,
+      "loss": 4.051,
       "step": 4607
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.624896949711459e-05,
-      "loss": 0.3325,
+      "learning_rate": 0.00015395713107996703,
+      "loss": 3.8112,
       "step": 4608
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.6187139323990106e-05,
-      "loss": 0.5146,
+      "learning_rate": 0.0001537510305028854,
+      "loss": 4.1069,
       "step": 4609
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.6125309150865616e-05,
-      "loss": 0.5458,
+      "learning_rate": 0.0001535449299258038,
+      "loss": 4.158,
       "step": 4610
     },
     {
       "epoch": 2.58,
-      "learning_rate": 4.606347897774113e-05,
-      "loss": 0.6403,
+      "learning_rate": 0.00015333882934872217,
+      "loss": 3.9009,
       "step": 4611
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.600164880461665e-05,
-      "loss": 0.3939,
+      "learning_rate": 0.00015313272877164055,
+      "loss": 3.7662,
       "step": 4612
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.593981863149216e-05,
-      "loss": 0.6947,
+      "learning_rate": 0.00015292662819455893,
+      "loss": 4.091,
       "step": 4613
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.5877988458367675e-05,
-      "loss": 0.2514,
+      "learning_rate": 0.00015272052761747734,
+      "loss": 3.328,
       "step": 4614
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.58161582852432e-05,
-      "loss": 0.3742,
+      "learning_rate": 0.00015251442704039572,
+      "loss": 3.6052,
       "step": 4615
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.5754328112118715e-05,
-      "loss": 0.3304,
+      "learning_rate": 0.0001523083264633141,
+      "loss": 3.556,
       "step": 4616
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.5692497938994225e-05,
-      "loss": 0.5438,
+      "learning_rate": 0.00015210222588623248,
+      "loss": 3.6166,
       "step": 4617
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.563066776586974e-05,
-      "loss": 0.2844,
+      "learning_rate": 0.00015189612530915086,
+      "loss": 2.9811,
       "step": 4618
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.556883759274526e-05,
-      "loss": 0.9369,
+      "learning_rate": 0.00015169002473206924,
+      "loss": 4.8348,
       "step": 4619
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.550700741962077e-05,
-      "loss": 1.034,
+      "learning_rate": 0.00015148392415498762,
+      "loss": 4.3396,
       "step": 4620
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.5445177246496285e-05,
-      "loss": 0.7153,
+      "learning_rate": 0.00015127782357790602,
+      "loss": 4.0664,
       "step": 4621
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.53833470733718e-05,
-      "loss": 0.6719,
+      "learning_rate": 0.00015107172300082443,
+      "loss": 4.1203,
       "step": 4622
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.532151690024732e-05,
-      "loss": 0.6055,
+      "learning_rate": 0.0001508656224237428,
+      "loss": 4.2238,
       "step": 4623
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.5259686727122835e-05,
-      "loss": 0.6199,
+      "learning_rate": 0.0001506595218466612,
+      "loss": 4.2009,
       "step": 4624
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.519785655399835e-05,
-      "loss": 0.7435,
+      "learning_rate": 0.00015045342126957957,
+      "loss": 4.2993,
       "step": 4625
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.513602638087386e-05,
-      "loss": 0.6889,
+      "learning_rate": 0.00015024732069249795,
+      "loss": 4.2291,
       "step": 4626
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.507419620774938e-05,
-      "loss": 0.6426,
+      "learning_rate": 0.00015004122011541633,
+      "loss": 4.2534,
       "step": 4627
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.5012366034624894e-05,
-      "loss": 0.4679,
+      "learning_rate": 0.0001498351195383347,
+      "loss": 4.2264,
       "step": 4628
     },
     {
       "epoch": 2.59,
-      "learning_rate": 4.4950535861500404e-05,
-      "loss": 0.7043,
+      "learning_rate": 0.00014962901896125309,
+      "loss": 4.4302,
       "step": 4629
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.488870568837592e-05,
-      "loss": 0.8308,
+      "learning_rate": 0.0001494229183841715,
+      "loss": 4.1627,
       "step": 4630
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.4826875515251444e-05,
-      "loss": 0.7393,
+      "learning_rate": 0.00014921681780708987,
+      "loss": 4.0143,
       "step": 4631
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.4765045342126954e-05,
-      "loss": 0.7376,
+      "learning_rate": 0.00014901071723000825,
+      "loss": 4.201,
       "step": 4632
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.470321516900247e-05,
-      "loss": 0.5199,
+      "learning_rate": 0.00014880461665292663,
+      "loss": 4.0288,
       "step": 4633
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.464138499587799e-05,
-      "loss": 0.4418,
+      "learning_rate": 0.000148598516075845,
+      "loss": 4.0679,
       "step": 4634
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.45795548227535e-05,
-      "loss": 0.5535,
+      "learning_rate": 0.0001483924154987634,
+      "loss": 3.9816,
       "step": 4635
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.451772464962901e-05,
-      "loss": 0.4004,
+      "learning_rate": 0.00014818631492168177,
+      "loss": 4.0635,
       "step": 4636
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.445589447650453e-05,
-      "loss": 0.4087,
+      "learning_rate": 0.00014798021434460015,
+      "loss": 4.0209,
       "step": 4637
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.439406430338004e-05,
-      "loss": 0.442,
+      "learning_rate": 0.00014777411376751856,
+      "loss": 4.093,
       "step": 4638
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.433223413025556e-05,
-      "loss": 0.6812,
+      "learning_rate": 0.00014756801319043694,
+      "loss": 4.1456,
       "step": 4639
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.427040395713108e-05,
-      "loss": 0.5398,
+      "learning_rate": 0.00014736191261335532,
+      "loss": 4.0307,
       "step": 4640
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.4208573784006596e-05,
-      "loss": 0.4008,
+      "learning_rate": 0.0001471558120362737,
+      "loss": 3.783,
       "step": 4641
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.4146743610882106e-05,
-      "loss": 0.6952,
+      "learning_rate": 0.00014694971145919208,
+      "loss": 4.1906,
       "step": 4642
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.408491343775762e-05,
-      "loss": 0.5337,
+      "learning_rate": 0.00014674361088211045,
+      "loss": 4.3647,
       "step": 4643
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.402308326463313e-05,
-      "loss": 0.5081,
+      "learning_rate": 0.00014653751030502886,
+      "loss": 3.9119,
       "step": 4644
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.396125309150865e-05,
-      "loss": 0.5091,
+      "learning_rate": 0.00014633140972794724,
+      "loss": 4.0297,
       "step": 4645
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.3899422918384166e-05,
-      "loss": 0.481,
+      "learning_rate": 0.00014612530915086565,
+      "loss": 4.3018,
       "step": 4646
     },
     {
       "epoch": 2.6,
-      "learning_rate": 4.383759274525969e-05,
-      "loss": 0.7246,
+      "learning_rate": 0.00014591920857378403,
+      "loss": 4.0806,
       "step": 4647
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.37757625721352e-05,
-      "loss": 0.4759,
+      "learning_rate": 0.0001457131079967024,
+      "loss": 3.9516,
       "step": 4648
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.3713932399010715e-05,
-      "loss": 0.5937,
+      "learning_rate": 0.0001455070074196208,
+      "loss": 4.1027,
       "step": 4649
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.365210222588623e-05,
-      "loss": 0.2856,
+      "learning_rate": 0.00014530090684253917,
+      "loss": 3.7111,
       "step": 4650
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.359027205276174e-05,
-      "loss": 0.7138,
+      "learning_rate": 0.00014509480626545755,
+      "loss": 4.0773,
       "step": 4651
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.352844187963726e-05,
-      "loss": 0.3981,
+      "learning_rate": 0.00014488870568837593,
+      "loss": 3.7453,
       "step": 4652
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.3466611706512775e-05,
-      "loss": 0.6704,
+      "learning_rate": 0.0001446826051112943,
+      "loss": 4.2575,
       "step": 4653
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.3404781533388285e-05,
-      "loss": 0.6644,
+      "learning_rate": 0.0001444765045342127,
+      "loss": 4.1788,
       "step": 4654
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.334295136026381e-05,
-      "loss": 0.3829,
+      "learning_rate": 0.0001442704039571311,
+      "loss": 3.8356,
       "step": 4655
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.3281121187139325e-05,
-      "loss": 0.5882,
+      "learning_rate": 0.00014406430338004947,
+      "loss": 4.2063,
       "step": 4656
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.3219291014014834e-05,
-      "loss": 0.4101,
+      "learning_rate": 0.00014385820280296785,
+      "loss": 3.9853,
       "step": 4657
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.315746084089035e-05,
-      "loss": 0.441,
+      "learning_rate": 0.00014365210222588623,
+      "loss": 3.9686,
       "step": 4658
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.309563066776587e-05,
-      "loss": 0.4486,
+      "learning_rate": 0.0001434460016488046,
+      "loss": 3.9942,
       "step": 4659
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.303380049464138e-05,
-      "loss": 0.4643,
+      "learning_rate": 0.000143239901071723,
+      "loss": 3.837,
       "step": 4660
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.2971970321516894e-05,
-      "loss": 0.4193,
+      "learning_rate": 0.00014303380049464137,
+      "loss": 3.8776,
       "step": 4661
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.291014014839241e-05,
-      "loss": 0.3979,
+      "learning_rate": 0.00014282769991755978,
+      "loss": 3.7688,
       "step": 4662
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.284830997526793e-05,
-      "loss": 0.4213,
+      "learning_rate": 0.00014262159934047815,
+      "loss": 3.4921,
       "step": 4663
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.2786479802143444e-05,
-      "loss": 0.7279,
+      "learning_rate": 0.00014241549876339653,
+      "loss": 4.0959,
       "step": 4664
     },
     {
       "epoch": 2.61,
-      "learning_rate": 4.272464962901896e-05,
-      "loss": 0.5858,
+      "learning_rate": 0.00014220939818631491,
+      "loss": 3.5117,
       "step": 4665
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.266281945589447e-05,
-      "loss": 0.4847,
+      "learning_rate": 0.0001420032976092333,
+      "loss": 3.4592,
       "step": 4666
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.260098928276999e-05,
-      "loss": 0.7092,
+      "learning_rate": 0.0001417971970321517,
+      "loss": 3.7219,
       "step": 4667
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.25391591096455e-05,
-      "loss": 0.6781,
+      "learning_rate": 0.00014159109645507008,
+      "loss": 3.4239,
       "step": 4668
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.247732893652101e-05,
-      "loss": 0.6843,
+      "learning_rate": 0.00014138499587798846,
+      "loss": 4.2929,
       "step": 4669
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.241549876339653e-05,
-      "loss": 0.677,
+      "learning_rate": 0.00014117889530090687,
+      "loss": 4.2338,
       "step": 4670
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.235366859027205e-05,
-      "loss": 0.7157,
+      "learning_rate": 0.00014097279472382525,
+      "loss": 4.3697,
       "step": 4671
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.229183841714757e-05,
-      "loss": 1.0034,
+      "learning_rate": 0.00014076669414674363,
+      "loss": 4.1693,
       "step": 4672
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.223000824402308e-05,
-      "loss": 0.4742,
+      "learning_rate": 0.000140560593569662,
+      "loss": 3.8348,
       "step": 4673
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.2168178070898596e-05,
-      "loss": 0.6314,
+      "learning_rate": 0.00014035449299258038,
+      "loss": 4.2477,
       "step": 4674
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.2106347897774106e-05,
-      "loss": 0.6084,
+      "learning_rate": 0.00014014839241549876,
+      "loss": 4.0977,
       "step": 4675
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.204451772464962e-05,
-      "loss": 0.6281,
+      "learning_rate": 0.00013994229183841714,
+      "loss": 4.1638,
       "step": 4676
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.198268755152514e-05,
-      "loss": 0.5683,
+      "learning_rate": 0.00013973619126133552,
+      "loss": 3.9306,
       "step": 4677
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.192085737840065e-05,
-      "loss": 0.5202,
+      "learning_rate": 0.00013953009068425393,
+      "loss": 4.0746,
       "step": 4678
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.185902720527617e-05,
-      "loss": 0.5176,
+      "learning_rate": 0.0001393239901071723,
+      "loss": 4.264,
       "step": 4679
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.179719703215169e-05,
-      "loss": 0.4758,
+      "learning_rate": 0.0001391178895300907,
+      "loss": 4.2246,
       "step": 4680
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.1735366859027205e-05,
-      "loss": 0.581,
+      "learning_rate": 0.00013891178895300907,
+      "loss": 4.1687,
       "step": 4681
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.1673536685902715e-05,
-      "loss": 0.4478,
+      "learning_rate": 0.00013870568837592745,
+      "loss": 3.8738,
       "step": 4682
     },
     {
       "epoch": 2.62,
-      "learning_rate": 4.161170651277823e-05,
-      "loss": 0.6742,
+      "learning_rate": 0.00013849958779884583,
+      "loss": 4.1893,
       "step": 4683
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.154987633965375e-05,
-      "loss": 0.5164,
+      "learning_rate": 0.0001382934872217642,
+      "loss": 3.9788,
       "step": 4684
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.148804616652926e-05,
-      "loss": 0.6075,
+      "learning_rate": 0.0001380873866446826,
+      "loss": 4.1328,
       "step": 4685
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.1426215993404775e-05,
-      "loss": 0.4357,
+      "learning_rate": 0.000137881286067601,
+      "loss": 4.0293,
       "step": 4686
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.13643858202803e-05,
-      "loss": 0.5011,
+      "learning_rate": 0.00013767518549051937,
+      "loss": 4.1305,
       "step": 4687
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.130255564715581e-05,
-      "loss": 0.3677,
+      "learning_rate": 0.00013746908491343775,
+      "loss": 3.8114,
       "step": 4688
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.1240725474031325e-05,
-      "loss": 0.5488,
+      "learning_rate": 0.00013726298433635613,
+      "loss": 4.2089,
       "step": 4689
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.117889530090684e-05,
-      "loss": 0.4642,
+      "learning_rate": 0.0001370568837592745,
+      "loss": 3.8248,
       "step": 4690
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.111706512778235e-05,
-      "loss": 0.8383,
+      "learning_rate": 0.00013685078318219292,
+      "loss": 4.4089,
       "step": 4691
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.105523495465787e-05,
-      "loss": 0.71,
+      "learning_rate": 0.0001366446826051113,
+      "loss": 4.1576,
       "step": 4692
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.0993404781533384e-05,
-      "loss": 0.4997,
+      "learning_rate": 0.00013643858202802968,
+      "loss": 4.0801,
       "step": 4693
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.0931574608408894e-05,
-      "loss": 0.6218,
+      "learning_rate": 0.00013623248145094808,
+      "loss": 4.061,
       "step": 4694
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.086974443528442e-05,
-      "loss": 0.6437,
+      "learning_rate": 0.00013602638087386646,
+      "loss": 3.9781,
       "step": 4695
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.0807914262159934e-05,
-      "loss": 0.5211,
+      "learning_rate": 0.00013582028029678484,
+      "loss": 4.1386,
       "step": 4696
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.0746084089035444e-05,
-      "loss": 0.4493,
+      "learning_rate": 0.00013561417971970322,
+      "loss": 3.952,
       "step": 4697
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.068425391591096e-05,
-      "loss": 0.5784,
+      "learning_rate": 0.0001354080791426216,
+      "loss": 4.3411,
       "step": 4698
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.062242374278648e-05,
-      "loss": 0.6784,
+      "learning_rate": 0.00013520197856553998,
+      "loss": 4.1418,
       "step": 4699
     },
     {
       "epoch": 2.63,
-      "learning_rate": 4.056059356966199e-05,
-      "loss": 0.4709,
+      "learning_rate": 0.00013499587798845836,
+      "loss": 3.9478,
       "step": 4700
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.04987633965375e-05,
-      "loss": 0.59,
+      "learning_rate": 0.00013478977741137677,
+      "loss": 4.02,
       "step": 4701
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.043693322341303e-05,
-      "loss": 0.5372,
+      "learning_rate": 0.00013458367683429515,
+      "loss": 3.8414,
       "step": 4702
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.037510305028854e-05,
-      "loss": 0.4513,
+      "learning_rate": 0.00013437757625721353,
+      "loss": 4.0611,
       "step": 4703
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.031327287716405e-05,
-      "loss": 0.4807,
+      "learning_rate": 0.0001341714756801319,
+      "loss": 4.0525,
       "step": 4704
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.025144270403957e-05,
-      "loss": 0.5638,
+      "learning_rate": 0.0001339653751030503,
+      "loss": 4.1811,
       "step": 4705
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.0189612530915086e-05,
-      "loss": 0.5615,
+      "learning_rate": 0.00013375927452596867,
+      "loss": 4.0968,
       "step": 4706
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.0127782357790596e-05,
-      "loss": 0.5072,
+      "learning_rate": 0.00013355317394888705,
+      "loss": 4.0181,
       "step": 4707
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.006595218466611e-05,
-      "loss": 0.4904,
+      "learning_rate": 0.00013334707337180543,
+      "loss": 4.1776,
       "step": 4708
     },
     {
       "epoch": 2.64,
-      "learning_rate": 4.000412201154162e-05,
-      "loss": 0.5015,
+      "learning_rate": 0.00013314097279472383,
+      "loss": 3.9946,
       "step": 4709
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.9942291838417146e-05,
-      "loss": 0.5458,
+      "learning_rate": 0.0001329348722176422,
+      "loss": 4.0251,
       "step": 4710
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.988046166529266e-05,
-      "loss": 0.5534,
+      "learning_rate": 0.0001327287716405606,
+      "loss": 3.7346,
       "step": 4711
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.981863149216818e-05,
-      "loss": 0.6546,
+      "learning_rate": 0.00013252267106347897,
+      "loss": 3.7563,
       "step": 4712
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.975680131904369e-05,
-      "loss": 0.3969,
+      "learning_rate": 0.00013231657048639735,
+      "loss": 3.5387,
       "step": 4713
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.9694971145919205e-05,
-      "loss": 0.3925,
+      "learning_rate": 0.00013211046990931576,
+      "loss": 3.6172,
       "step": 4714
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.963314097279472e-05,
-      "loss": 0.3512,
+      "learning_rate": 0.00013190436933223414,
+      "loss": 3.3348,
       "step": 4715
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.957131079967023e-05,
-      "loss": 0.6999,
+      "learning_rate": 0.00013169826875515252,
+      "loss": 3.7602,
       "step": 4716
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.950948062654575e-05,
-      "loss": 1.728,
+      "learning_rate": 0.00013149216817807092,
+      "loss": 4.1846,
       "step": 4717
     },
     {
       "epoch": 2.64,
-      "learning_rate": 3.944765045342127e-05,
-      "loss": 0.8844,
+      "learning_rate": 0.0001312860676009893,
+      "loss": 3.5609,
       "step": 4718
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.938582028029678e-05,
-      "loss": 0.9684,
+      "learning_rate": 0.00013107996702390768,
+      "loss": 4.5563,
       "step": 4719
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.93239901071723e-05,
-      "loss": 0.763,
+      "learning_rate": 0.00013087386644682606,
+      "loss": 4.1744,
       "step": 4720
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.9262159934047815e-05,
-      "loss": 0.5065,
+      "learning_rate": 0.00013066776586974444,
+      "loss": 4.0958,
       "step": 4721
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.9200329760923324e-05,
-      "loss": 0.7004,
+      "learning_rate": 0.00013046166529266282,
+      "loss": 4.094,
       "step": 4722
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.913849958779884e-05,
-      "loss": 0.5447,
+      "learning_rate": 0.0001302555647155812,
+      "loss": 4.0284,
       "step": 4723
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.907666941467436e-05,
-      "loss": 0.806,
+      "learning_rate": 0.00013004946413849958,
+      "loss": 4.5333,
       "step": 4724
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.901483924154987e-05,
-      "loss": 0.8093,
+      "learning_rate": 0.000129843363561418,
+      "loss": 4.3174,
       "step": 4725
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.895300906842539e-05,
-      "loss": 0.6254,
+      "learning_rate": 0.00012963726298433637,
+      "loss": 4.1769,
       "step": 4726
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.889117889530091e-05,
-      "loss": 0.6857,
+      "learning_rate": 0.00012943116240725475,
+      "loss": 3.9789,
       "step": 4727
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.882934872217642e-05,
-      "loss": 0.4884,
+      "learning_rate": 0.00012922506183017313,
+      "loss": 3.9255,
       "step": 4728
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.8767518549051934e-05,
-      "loss": 0.57,
+      "learning_rate": 0.0001290189612530915,
+      "loss": 4.257,
       "step": 4729
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.870568837592745e-05,
-      "loss": 0.6559,
+      "learning_rate": 0.00012881286067600989,
+      "loss": 4.3208,
       "step": 4730
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.864385820280296e-05,
-      "loss": 0.6019,
+      "learning_rate": 0.00012860676009892826,
+      "loss": 3.8751,
       "step": 4731
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.858202802967848e-05,
-      "loss": 0.4852,
+      "learning_rate": 0.00012840065952184664,
+      "loss": 4.2571,
       "step": 4732
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.852019785655399e-05,
-      "loss": 0.6484,
+      "learning_rate": 0.00012819455894476505,
+      "loss": 4.1136,
       "step": 4733
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.845836768342952e-05,
-      "loss": 0.641,
+      "learning_rate": 0.00012798845836768343,
+      "loss": 4.1469,
       "step": 4734
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.8396537510305027e-05,
-      "loss": 0.5425,
+      "learning_rate": 0.0001277823577906018,
+      "loss": 4.1639,
       "step": 4735
     },
     {
       "epoch": 2.65,
-      "learning_rate": 3.833470733718054e-05,
-      "loss": 0.6252,
+      "learning_rate": 0.0001275762572135202,
+      "loss": 4.3048,
       "step": 4736
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.827287716405606e-05,
-      "loss": 0.4637,
+      "learning_rate": 0.0001273701566364386,
+      "loss": 3.7976,
       "step": 4737
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.821104699093157e-05,
-      "loss": 0.5082,
+      "learning_rate": 0.00012716405605935698,
+      "loss": 4.0234,
       "step": 4738
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.8149216817807086e-05,
-      "loss": 0.9391,
+      "learning_rate": 0.00012695795548227536,
+      "loss": 4.1319,
       "step": 4739
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.8087386644682596e-05,
-      "loss": 0.4528,
+      "learning_rate": 0.00012675185490519374,
+      "loss": 3.864,
       "step": 4740
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.802555647155811e-05,
-      "loss": 0.5318,
+      "learning_rate": 0.00012654575432811214,
+      "loss": 4.2154,
       "step": 4741
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.7963726298433636e-05,
-      "loss": 0.4438,
+      "learning_rate": 0.00012633965375103052,
+      "loss": 3.7397,
       "step": 4742
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.790189612530915e-05,
-      "loss": 0.5789,
+      "learning_rate": 0.0001261335531739489,
+      "loss": 3.9282,
       "step": 4743
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.784006595218466e-05,
-      "loss": 0.4387,
+      "learning_rate": 0.00012592745259686728,
+      "loss": 4.1275,
       "step": 4744
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.777823577906018e-05,
-      "loss": 0.4595,
+      "learning_rate": 0.00012572135201978566,
+      "loss": 3.9722,
       "step": 4745
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.7716405605935695e-05,
-      "loss": 0.5646,
+      "learning_rate": 0.00012551525144270404,
+      "loss": 4.0371,
       "step": 4746
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.7654575432811205e-05,
-      "loss": 0.6426,
+      "learning_rate": 0.00012530915086562242,
+      "loss": 4.1809,
       "step": 4747
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.759274525968672e-05,
-      "loss": 0.467,
+      "learning_rate": 0.0001251030502885408,
+      "loss": 3.9882,
       "step": 4748
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.753091508656224e-05,
-      "loss": 0.4941,
+      "learning_rate": 0.0001248969497114592,
+      "loss": 4.129,
       "step": 4749
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.7469084913437755e-05,
-      "loss": 0.7508,
+      "learning_rate": 0.00012469084913437759,
+      "loss": 4.5172,
       "step": 4750
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.7407254740313265e-05,
-      "loss": 0.5004,
+      "learning_rate": 0.00012448474855729597,
+      "loss": 4.0818,
       "step": 4751
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.734542456718879e-05,
-      "loss": 0.4991,
+      "learning_rate": 0.00012427864798021434,
+      "loss": 3.8811,
       "step": 4752
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.72835943940643e-05,
-      "loss": 0.6378,
+      "learning_rate": 0.00012407254740313272,
+      "loss": 4.1954,
       "step": 4753
     },
     {
       "epoch": 2.66,
-      "learning_rate": 3.7221764220939815e-05,
-      "loss": 0.4404,
+      "learning_rate": 0.0001238664468260511,
+      "loss": 3.9921,
       "step": 4754
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.715993404781533e-05,
-      "loss": 0.5497,
+      "learning_rate": 0.0001236603462489695,
+      "loss": 4.1134,
       "step": 4755
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.709810387469085e-05,
-      "loss": 0.4543,
+      "learning_rate": 0.0001234542456718879,
+      "loss": 4.2043,
       "step": 4756
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.7036273701566364e-05,
-      "loss": 0.4183,
+      "learning_rate": 0.00012324814509480627,
+      "loss": 4.2192,
       "step": 4757
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.6974443528441874e-05,
-      "loss": 0.476,
+      "learning_rate": 0.00012304204451772465,
+      "loss": 3.771,
       "step": 4758
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.691261335531739e-05,
-      "loss": 0.4058,
+      "learning_rate": 0.00012283594394064303,
+      "loss": 3.8709,
       "step": 4759
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.685078318219291e-05,
-      "loss": 0.5702,
+      "learning_rate": 0.0001226298433635614,
+      "loss": 4.2509,
       "step": 4760
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.6788953009068424e-05,
-      "loss": 0.4485,
+      "learning_rate": 0.00012242374278647982,
+      "loss": 3.8805,
       "step": 4761
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.6727122835943934e-05,
-      "loss": 0.4788,
+      "learning_rate": 0.0001222176422093982,
+      "loss": 3.7339,
       "step": 4762
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.666529266281945e-05,
-      "loss": 0.4693,
+      "learning_rate": 0.00012201154163231657,
+      "loss": 3.4137,
       "step": 4763
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.660346248969497e-05,
-      "loss": 0.3801,
+      "learning_rate": 0.00012180544105523495,
+      "loss": 3.6801,
       "step": 4764
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.6541632316570483e-05,
-      "loss": 0.742,
+      "learning_rate": 0.00012159934047815333,
+      "loss": 3.9871,
       "step": 4765
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.6479802143446e-05,
-      "loss": 0.7167,
+      "learning_rate": 0.00012139323990107171,
+      "loss": 3.5351,
       "step": 4766
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.641797197032151e-05,
-      "loss": 0.8209,
+      "learning_rate": 0.00012118713932399012,
+      "loss": 3.8323,
       "step": 4767
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.635614179719703e-05,
-      "loss": 0.4051,
+      "learning_rate": 0.0001209810387469085,
+      "loss": 3.0118,
       "step": 4768
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.629431162407254e-05,
-      "loss": 0.9904,
+      "learning_rate": 0.00012077493816982688,
+      "loss": 4.4507,
       "step": 4769
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.623248145094806e-05,
-      "loss": 0.7051,
+      "learning_rate": 0.00012056883759274526,
+      "loss": 4.258,
       "step": 4770
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.6170651277823576e-05,
-      "loss": 0.5431,
+      "learning_rate": 0.00012036273701566365,
+      "loss": 4.1163,
       "step": 4771
     },
     {
       "epoch": 2.67,
-      "learning_rate": 3.610882110469909e-05,
-      "loss": 0.853,
+      "learning_rate": 0.00012015663643858203,
+      "loss": 4.1499,
       "step": 4772
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.60469909315746e-05,
-      "loss": 0.6249,
+      "learning_rate": 0.00011995053586150041,
+      "loss": 3.9667,
       "step": 4773
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.598516075845012e-05,
-      "loss": 0.6426,
+      "learning_rate": 0.00011974443528441879,
+      "loss": 3.9557,
       "step": 4774
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.5923330585325636e-05,
-      "loss": 0.6051,
+      "learning_rate": 0.00011953833470733718,
+      "loss": 4.0949,
       "step": 4775
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.586150041220115e-05,
-      "loss": 0.5908,
+      "learning_rate": 0.00011933223413025556,
+      "loss": 4.099,
       "step": 4776
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.579967023907667e-05,
-      "loss": 0.5661,
+      "learning_rate": 0.00011912613355317394,
+      "loss": 4.1776,
       "step": 4777
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.573784006595218e-05,
-      "loss": 0.3947,
+      "learning_rate": 0.00011892003297609234,
+      "loss": 3.852,
       "step": 4778
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.5676009892827695e-05,
-      "loss": 0.3968,
+      "learning_rate": 0.00011871393239901073,
+      "loss": 3.9028,
       "step": 4779
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.561417971970321e-05,
-      "loss": 0.5109,
+      "learning_rate": 0.00011850783182192911,
+      "loss": 4.0701,
       "step": 4780
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.555234954657873e-05,
-      "loss": 0.3957,
+      "learning_rate": 0.00011830173124484749,
+      "loss": 3.9176,
       "step": 4781
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.5490519373454245e-05,
-      "loss": 0.5369,
+      "learning_rate": 0.00011809563066776587,
+      "loss": 3.9677,
       "step": 4782
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.5428689200329755e-05,
-      "loss": 0.326,
+      "learning_rate": 0.00011788953009068426,
+      "loss": 3.6118,
       "step": 4783
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.536685902720527e-05,
-      "loss": 0.4215,
+      "learning_rate": 0.00011768342951360264,
+      "loss": 3.8207,
       "step": 4784
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.530502885408079e-05,
-      "loss": 0.6245,
+      "learning_rate": 0.00011747732893652102,
+      "loss": 4.1745,
       "step": 4785
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.5243198680956305e-05,
-      "loss": 0.5653,
+      "learning_rate": 0.0001172712283594394,
+      "loss": 3.9395,
       "step": 4786
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.5181368507831814e-05,
-      "loss": 0.5996,
+      "learning_rate": 0.00011706512778235779,
+      "loss": 4.3215,
       "step": 4787
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.511953833470734e-05,
-      "loss": 0.6091,
+      "learning_rate": 0.00011685902720527617,
+      "loss": 4.0716,
       "step": 4788
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.505770816158285e-05,
-      "loss": 0.6385,
+      "learning_rate": 0.00011665292662819455,
+      "loss": 3.9891,
       "step": 4789
     },
     {
       "epoch": 2.68,
-      "learning_rate": 3.4995877988458364e-05,
-      "loss": 0.4593,
+      "learning_rate": 0.00011644682605111295,
+      "loss": 4.063,
       "step": 4790
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.493404781533388e-05,
-      "loss": 0.4391,
+      "learning_rate": 0.00011624072547403134,
+      "loss": 3.9047,
       "step": 4791
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.48722176422094e-05,
-      "loss": 0.5572,
+      "learning_rate": 0.00011603462489694972,
+      "loss": 4.2279,
       "step": 4792
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.481038746908491e-05,
-      "loss": 0.5204,
+      "learning_rate": 0.0001158285243198681,
+      "loss": 3.9869,
       "step": 4793
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.4748557295960424e-05,
-      "loss": 0.4922,
+      "learning_rate": 0.00011562242374278648,
+      "loss": 3.9489,
       "step": 4794
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.468672712283594e-05,
-      "loss": 0.5238,
+      "learning_rate": 0.00011541632316570487,
+      "loss": 4.068,
       "step": 4795
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.462489694971146e-05,
-      "loss": 0.4227,
+      "learning_rate": 0.00011521022258862325,
+      "loss": 3.9465,
       "step": 4796
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.4563066776586974e-05,
-      "loss": 0.4704,
+      "learning_rate": 0.00011500412201154163,
+      "loss": 4.1374,
       "step": 4797
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.4501236603462483e-05,
-      "loss": 0.6245,
+      "learning_rate": 0.00011479802143446001,
+      "loss": 4.2393,
       "step": 4798
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.4439406430338e-05,
-      "loss": 0.5644,
+      "learning_rate": 0.0001145919208573784,
+      "loss": 4.0904,
       "step": 4799
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.4377576257213517e-05,
-      "loss": 0.5432,
+      "learning_rate": 0.00011438582028029678,
+      "loss": 4.013,
       "step": 4800
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.431574608408903e-05,
-      "loss": 0.4134,
+      "learning_rate": 0.00011417971970321516,
+      "loss": 3.8992,
       "step": 4801
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.425391591096455e-05,
-      "loss": 0.4899,
+      "learning_rate": 0.00011397361912613355,
+      "loss": 4.3932,
       "step": 4802
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.419208573784006e-05,
-      "loss": 0.3474,
+      "learning_rate": 0.00011376751854905195,
+      "loss": 3.8581,
       "step": 4803
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.4130255564715576e-05,
-      "loss": 0.3749,
+      "learning_rate": 0.00011356141797197033,
+      "loss": 3.8381,
       "step": 4804
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.406842539159109e-05,
-      "loss": 1.1411,
+      "learning_rate": 0.0001133553173948887,
+      "loss": 4.3234,
       "step": 4805
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.400659521846661e-05,
-      "loss": 0.4049,
+      "learning_rate": 0.0001131492168178071,
+      "loss": 3.7634,
       "step": 4806
     },
     {
       "epoch": 2.69,
-      "learning_rate": 3.3944765045342126e-05,
-      "loss": 0.4807,
+      "learning_rate": 0.00011294311624072548,
+      "loss": 3.8043,
       "step": 4807
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.388293487221764e-05,
-      "loss": 0.3015,
+      "learning_rate": 0.00011273701566364386,
+      "loss": 3.6171,
       "step": 4808
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.382110469909315e-05,
-      "loss": 0.4987,
+      "learning_rate": 0.00011253091508656224,
+      "loss": 3.9244,
       "step": 4809
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.375927452596867e-05,
-      "loss": 0.4407,
+      "learning_rate": 0.00011232481450948063,
+      "loss": 4.1457,
       "step": 4810
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.3697444352844185e-05,
-      "loss": 0.5519,
+      "learning_rate": 0.00011211871393239901,
+      "loss": 3.9543,
       "step": 4811
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.36356141797197e-05,
-      "loss": 0.6438,
+      "learning_rate": 0.00011191261335531739,
+      "loss": 4.0763,
       "step": 4812
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.357378400659522e-05,
-      "loss": 0.6513,
+      "learning_rate": 0.00011170651277823578,
+      "loss": 3.8289,
       "step": 4813
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.351195383347073e-05,
-      "loss": 0.5624,
+      "learning_rate": 0.00011150041220115418,
+      "loss": 3.7572,
       "step": 4814
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.3450123660346245e-05,
-      "loss": 0.8613,
+      "learning_rate": 0.00011129431162407256,
+      "loss": 3.3532,
       "step": 4815
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.338829348722176e-05,
-      "loss": 0.6313,
+      "learning_rate": 0.00011108821104699094,
+      "loss": 3.3987,
       "step": 4816
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.332646331409728e-05,
-      "loss": 0.5786,
+      "learning_rate": 0.00011088211046990932,
+      "loss": 4.1033,
       "step": 4817
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.326463314097279e-05,
-      "loss": 0.7007,
+      "learning_rate": 0.00011067600989282771,
+      "loss": 3.4816,
       "step": 4818
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.320280296784831e-05,
-      "loss": 0.7154,
+      "learning_rate": 0.00011046990931574609,
+      "loss": 4.1807,
       "step": 4819
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.314097279472382e-05,
-      "loss": 0.6676,
+      "learning_rate": 0.00011026380873866447,
+      "loss": 4.3756,
       "step": 4820
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.307914262159934e-05,
-      "loss": 0.8081,
+      "learning_rate": 0.00011005770816158285,
+      "loss": 4.0864,
       "step": 4821
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.3017312448474854e-05,
-      "loss": 0.6516,
+      "learning_rate": 0.00010985160758450124,
+      "loss": 4.0796,
       "step": 4822
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.295548227535037e-05,
-      "loss": 0.4241,
+      "learning_rate": 0.00010964550700741962,
+      "loss": 3.9667,
       "step": 4823
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.289365210222589e-05,
-      "loss": 0.7149,
+      "learning_rate": 0.000109439406430338,
+      "loss": 4.1096,
       "step": 4824
     },
     {
       "epoch": 2.7,
-      "learning_rate": 3.28318219291014e-05,
-      "loss": 0.6767,
+      "learning_rate": 0.0001092333058532564,
+      "loss": 4.1128,
       "step": 4825
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.2769991755976914e-05,
-      "loss": 0.61,
+      "learning_rate": 0.00010902720527617479,
+      "loss": 3.9939,
       "step": 4826
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.270816158285243e-05,
-      "loss": 0.7411,
+      "learning_rate": 0.00010882110469909317,
+      "loss": 4.4454,
       "step": 4827
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.264633140972795e-05,
-      "loss": 0.4044,
+      "learning_rate": 0.00010861500412201155,
+      "loss": 3.9747,
       "step": 4828
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.258450123660346e-05,
-      "loss": 0.4533,
+      "learning_rate": 0.00010840890354492993,
+      "loss": 4.0162,
       "step": 4829
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.2522671063478974e-05,
-      "loss": 0.5793,
+      "learning_rate": 0.00010820280296784832,
+      "loss": 4.1327,
       "step": 4830
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.246084089035449e-05,
-      "loss": 0.4457,
+      "learning_rate": 0.0001079967023907667,
+      "loss": 3.8298,
       "step": 4831
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.239901071723001e-05,
-      "loss": 0.4696,
+      "learning_rate": 0.00010779060181368508,
+      "loss": 3.9623,
       "step": 4832
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.233718054410552e-05,
-      "loss": 0.5721,
+      "learning_rate": 0.00010758450123660346,
+      "loss": 4.23,
       "step": 4833
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.227535037098103e-05,
-      "loss": 0.4985,
+      "learning_rate": 0.00010737840065952185,
+      "loss": 4.1793,
       "step": 4834
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.221352019785655e-05,
-      "loss": 0.4786,
+      "learning_rate": 0.00010717230008244023,
+      "loss": 4.0039,
       "step": 4835
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.2151690024732066e-05,
-      "loss": 0.5676,
+      "learning_rate": 0.00010696619950535862,
+      "loss": 3.9986,
       "step": 4836
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.208985985160758e-05,
-      "loss": 0.6459,
+      "learning_rate": 0.000106760098928277,
+      "loss": 4.1194,
       "step": 4837
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.202802967848309e-05,
-      "loss": 0.5694,
+      "learning_rate": 0.0001065539983511954,
+      "loss": 3.829,
       "step": 4838
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.1966199505358616e-05,
-      "loss": 0.5583,
+      "learning_rate": 0.00010634789777411378,
+      "loss": 4.0765,
       "step": 4839
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.1904369332234126e-05,
-      "loss": 0.4841,
+      "learning_rate": 0.00010614179719703215,
+      "loss": 4.1087,
       "step": 4840
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.184253915910964e-05,
-      "loss": 0.5871,
+      "learning_rate": 0.00010593569661995053,
+      "loss": 4.2709,
       "step": 4841
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.178070898598516e-05,
-      "loss": 0.5412,
+      "learning_rate": 0.00010572959604286893,
+      "loss": 4.093,
       "step": 4842
     },
     {
       "epoch": 2.71,
-      "learning_rate": 3.1718878812860676e-05,
-      "loss": 0.6543,
+      "learning_rate": 0.00010552349546578731,
+      "loss": 3.8756,
       "step": 4843
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.165704863973619e-05,
-      "loss": 0.3619,
+      "learning_rate": 0.00010531739488870569,
+      "loss": 3.8849,
       "step": 4844
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.15952184666117e-05,
-      "loss": 0.5835,
+      "learning_rate": 0.00010511129431162407,
+      "loss": 4.0173,
       "step": 4845
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.153338829348722e-05,
-      "loss": 0.4275,
+      "learning_rate": 0.00010490519373454246,
+      "loss": 4.33,
       "step": 4846
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.1471558120362735e-05,
-      "loss": 0.3692,
+      "learning_rate": 0.00010469909315746084,
+      "loss": 3.6238,
       "step": 4847
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.140972794723825e-05,
-      "loss": 0.5072,
+      "learning_rate": 0.00010449299258037923,
+      "loss": 3.8127,
       "step": 4848
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.134789777411376e-05,
-      "loss": 0.5042,
+      "learning_rate": 0.00010428689200329761,
+      "loss": 4.2908,
       "step": 4849
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.128606760098928e-05,
-      "loss": 0.6934,
+      "learning_rate": 0.000104080791426216,
+      "loss": 4.1978,
       "step": 4850
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.1224237427864795e-05,
-      "loss": 0.4285,
+      "learning_rate": 0.00010387469084913438,
+      "loss": 3.7541,
       "step": 4851
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.116240725474031e-05,
-      "loss": 0.5427,
+      "learning_rate": 0.00010366859027205276,
+      "loss": 4.2765,
       "step": 4852
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.110057708161583e-05,
-      "loss": 0.4144,
+      "learning_rate": 0.00010346248969497114,
+      "loss": 3.8571,
       "step": 4853
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.103874690849134e-05,
-      "loss": 1.2919,
+      "learning_rate": 0.00010325638911788954,
+      "loss": 4.3368,
       "step": 4854
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.097691673536686e-05,
-      "loss": 0.552,
+      "learning_rate": 0.00010305028854080792,
+      "loss": 4.0531,
       "step": 4855
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.091508656224237e-05,
-      "loss": 0.4579,
+      "learning_rate": 0.0001028441879637263,
+      "loss": 3.9592,
       "step": 4856
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.085325638911789e-05,
-      "loss": 0.38,
+      "learning_rate": 0.00010263808738664468,
+      "loss": 3.9429,
       "step": 4857
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.07914262159934e-05,
-      "loss": 0.4802,
+      "learning_rate": 0.00010243198680956307,
+      "loss": 3.8624,
       "step": 4858
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.072959604286892e-05,
-      "loss": 0.58,
+      "learning_rate": 0.00010222588623248145,
+      "loss": 3.9816,
       "step": 4859
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.066776586974443e-05,
-      "loss": 0.7091,
+      "learning_rate": 0.00010201978565539984,
+      "loss": 4.476,
       "step": 4860
     },
     {
       "epoch": 2.72,
-      "learning_rate": 3.060593569661995e-05,
-      "loss": 0.4126,
+      "learning_rate": 0.00010181368507831822,
+      "loss": 3.5891,
       "step": 4861
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.0544105523495464e-05,
-      "loss": 0.9858,
+      "learning_rate": 0.00010160758450123661,
+      "loss": 4.1855,
       "step": 4862
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.048227535037098e-05,
-      "loss": 0.2725,
+      "learning_rate": 0.000101401483924155,
+      "loss": 3.6596,
       "step": 4863
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.0420445177246493e-05,
-      "loss": 0.4263,
+      "learning_rate": 0.00010119538334707337,
+      "loss": 4.0405,
       "step": 4864
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.035861500412201e-05,
-      "loss": 0.3962,
+      "learning_rate": 0.00010098928276999175,
+      "loss": 3.3664,
       "step": 4865
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.0296784830997523e-05,
-      "loss": 0.7613,
+      "learning_rate": 0.00010078318219291015,
+      "loss": 3.8735,
       "step": 4866
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.023495465787304e-05,
-      "loss": 1.3348,
+      "learning_rate": 0.00010057708161582853,
+      "loss": 3.7382,
       "step": 4867
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.0173124484748556e-05,
-      "loss": 0.4485,
+      "learning_rate": 0.0001003709810387469,
+      "loss": 3.3152,
       "step": 4868
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.011129431162407e-05,
-      "loss": 0.7761,
+      "learning_rate": 0.00010016488046166528,
+      "loss": 4.3013,
       "step": 4869
     },
     {
       "epoch": 2.73,
-      "learning_rate": 3.0049464138499583e-05,
-      "loss": 0.5934,
+      "learning_rate": 9.995877988458368e-05,
+      "loss": 4.1549,
       "step": 4870
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9987633965375103e-05,
-      "loss": 0.6929,
+      "learning_rate": 9.975267930750207e-05,
+      "loss": 4.1639,
       "step": 4871
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9925803792250616e-05,
-      "loss": 0.4865,
+      "learning_rate": 9.954657873042045e-05,
+      "loss": 4.2283,
       "step": 4872
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.986397361912613e-05,
-      "loss": 0.6834,
+      "learning_rate": 9.934047815333883e-05,
+      "loss": 4.1777,
       "step": 4873
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9802143446001646e-05,
-      "loss": 0.5141,
+      "learning_rate": 9.913437757625722e-05,
+      "loss": 3.9231,
       "step": 4874
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9740313272877162e-05,
-      "loss": 0.6189,
+      "learning_rate": 9.89282769991756e-05,
+      "loss": 4.2395,
       "step": 4875
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9678483099752675e-05,
-      "loss": 0.524,
+      "learning_rate": 9.872217642209398e-05,
+      "loss": 4.016,
       "step": 4876
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9616652926628192e-05,
-      "loss": 0.6475,
+      "learning_rate": 9.851607584501236e-05,
+      "loss": 4.1604,
       "step": 4877
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9554822753503705e-05,
-      "loss": 0.4665,
+      "learning_rate": 9.830997526793076e-05,
+      "loss": 3.9011,
       "step": 4878
     },
     {
       "epoch": 2.73,
-      "learning_rate": 2.9492992580379225e-05,
-      "loss": 0.4691,
+      "learning_rate": 9.810387469084913e-05,
+      "loss": 4.1842,
       "step": 4879
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.943116240725474e-05,
-      "loss": 0.527,
+      "learning_rate": 9.789777411376751e-05,
+      "loss": 4.2167,
       "step": 4880
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.936933223413025e-05,
-      "loss": 0.3687,
+      "learning_rate": 9.76916735366859e-05,
+      "loss": 3.9214,
       "step": 4881
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.9307502061005765e-05,
-      "loss": 0.5641,
+      "learning_rate": 9.748557295960429e-05,
+      "loss": 4.1699,
       "step": 4882
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.9245671887881285e-05,
-      "loss": 0.6738,
+      "learning_rate": 9.727947238252268e-05,
+      "loss": 4.0379,
       "step": 4883
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.9183841714756798e-05,
-      "loss": 0.4017,
+      "learning_rate": 9.707337180544106e-05,
+      "loss": 3.9369,
       "step": 4884
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.9122011541632315e-05,
-      "loss": 0.8762,
+      "learning_rate": 9.686727122835944e-05,
+      "loss": 4.3384,
       "step": 4885
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.9060181368507828e-05,
-      "loss": 0.51,
+      "learning_rate": 9.666117065127783e-05,
+      "loss": 3.9574,
       "step": 4886
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8998351195383344e-05,
-      "loss": 0.5628,
+      "learning_rate": 9.645507007419621e-05,
+      "loss": 4.1505,
       "step": 4887
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.893652102225886e-05,
-      "loss": 0.5182,
+      "learning_rate": 9.624896949711459e-05,
+      "loss": 3.9677,
       "step": 4888
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8874690849134374e-05,
-      "loss": 0.7842,
+      "learning_rate": 9.604286892003297e-05,
+      "loss": 4.4408,
       "step": 4889
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8812860676009887e-05,
-      "loss": 0.387,
+      "learning_rate": 9.583676834295136e-05,
+      "loss": 4.0143,
       "step": 4890
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8751030502885407e-05,
-      "loss": 0.5456,
+      "learning_rate": 9.563066776586974e-05,
+      "loss": 4.0088,
       "step": 4891
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.868920032976092e-05,
-      "loss": 0.5504,
+      "learning_rate": 9.542456718878812e-05,
+      "loss": 4.0876,
       "step": 4892
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8627370156636434e-05,
-      "loss": 0.5195,
+      "learning_rate": 9.52184666117065e-05,
+      "loss": 4.0073,
       "step": 4893
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.856553998351195e-05,
-      "loss": 0.5706,
+      "learning_rate": 9.50123660346249e-05,
+      "loss": 3.9534,
       "step": 4894
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8503709810387467e-05,
-      "loss": 0.5748,
+      "learning_rate": 9.480626545754329e-05,
+      "loss": 4.0666,
       "step": 4895
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8441879637262984e-05,
-      "loss": 0.4479,
+      "learning_rate": 9.460016488046167e-05,
+      "loss": 3.8993,
       "step": 4896
     },
     {
       "epoch": 2.74,
-      "learning_rate": 2.8380049464138497e-05,
-      "loss": 0.4109,
+      "learning_rate": 9.439406430338005e-05,
+      "loss": 4.0755,
       "step": 4897
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.831821929101401e-05,
-      "loss": 0.604,
+      "learning_rate": 9.418796372629844e-05,
+      "loss": 4.0357,
       "step": 4898
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.825638911788953e-05,
-      "loss": 0.3763,
+      "learning_rate": 9.398186314921682e-05,
+      "loss": 3.9521,
       "step": 4899
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.8194558944765043e-05,
-      "loss": 0.4544,
+      "learning_rate": 9.37757625721352e-05,
+      "loss": 4.2231,
       "step": 4900
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.8132728771640556e-05,
-      "loss": 0.4915,
+      "learning_rate": 9.356966199505358e-05,
+      "loss": 3.9994,
       "step": 4901
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.8070898598516073e-05,
-      "loss": 0.6209,
+      "learning_rate": 9.336356141797197e-05,
+      "loss": 4.2229,
       "step": 4902
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.800906842539159e-05,
-      "loss": 0.3965,
+      "learning_rate": 9.315746084089035e-05,
+      "loss": 4.002,
       "step": 4903
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7947238252267103e-05,
-      "loss": 0.2971,
+      "learning_rate": 9.295136026380873e-05,
+      "loss": 3.7344,
       "step": 4904
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.788540807914262e-05,
-      "loss": 0.4348,
+      "learning_rate": 9.274525968672711e-05,
+      "loss": 3.9874,
       "step": 4905
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7823577906018132e-05,
-      "loss": 0.4457,
+      "learning_rate": 9.253915910964552e-05,
+      "loss": 3.7257,
       "step": 4906
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7761747732893652e-05,
-      "loss": 0.4031,
+      "learning_rate": 9.23330585325639e-05,
+      "loss": 3.8268,
       "step": 4907
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7699917559769166e-05,
-      "loss": 0.4824,
+      "learning_rate": 9.212695795548228e-05,
+      "loss": 4.0252,
       "step": 4908
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.763808738664468e-05,
-      "loss": 0.5082,
+      "learning_rate": 9.192085737840066e-05,
+      "loss": 3.6188,
       "step": 4909
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7576257213520192e-05,
-      "loss": 0.5497,
+      "learning_rate": 9.171475680131905e-05,
+      "loss": 4.1624,
       "step": 4910
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7514427040395712e-05,
-      "loss": 0.4542,
+      "learning_rate": 9.150865622423743e-05,
+      "loss": 4.0217,
       "step": 4911
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7452596867271225e-05,
-      "loss": 0.4587,
+      "learning_rate": 9.130255564715581e-05,
+      "loss": 3.7265,
       "step": 4912
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7390766694146742e-05,
-      "loss": 0.364,
+      "learning_rate": 9.109645507007419e-05,
+      "loss": 3.6108,
       "step": 4913
     },
     {
       "epoch": 2.75,
-      "learning_rate": 2.7328936521022255e-05,
-      "loss": 0.4705,
+      "learning_rate": 9.089035449299258e-05,
+      "loss": 3.9056,
       "step": 4914
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.726710634789777e-05,
-      "loss": 0.4843,
+      "learning_rate": 9.068425391591096e-05,
+      "loss": 3.5867,
       "step": 4915
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.7205276174773288e-05,
-      "loss": 0.5084,
+      "learning_rate": 9.047815333882934e-05,
+      "loss": 3.5735,
       "step": 4916
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.71434460016488e-05,
-      "loss": 0.5814,
+      "learning_rate": 9.027205276174774e-05,
+      "loss": 3.8849,
       "step": 4917
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.7081615828524318e-05,
-      "loss": 0.4058,
+      "learning_rate": 9.006595218466613e-05,
+      "loss": 3.794,
       "step": 4918
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.7019785655399834e-05,
-      "loss": 0.533,
+      "learning_rate": 8.985985160758451e-05,
+      "loss": 4.0421,
       "step": 4919
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6957955482275348e-05,
-      "loss": 0.6041,
+      "learning_rate": 8.965375103050289e-05,
+      "loss": 4.0963,
       "step": 4920
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.689612530915086e-05,
-      "loss": 0.7437,
+      "learning_rate": 8.944765045342128e-05,
+      "loss": 4.1222,
       "step": 4921
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.683429513602638e-05,
-      "loss": 1.0156,
+      "learning_rate": 8.924154987633966e-05,
+      "loss": 4.3553,
       "step": 4922
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6772464962901894e-05,
-      "loss": 0.6331,
+      "learning_rate": 8.903544929925804e-05,
+      "loss": 3.9195,
       "step": 4923
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6710634789777407e-05,
-      "loss": 0.723,
+      "learning_rate": 8.882934872217642e-05,
+      "loss": 4.3621,
       "step": 4924
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6648804616652924e-05,
-      "loss": 0.646,
+      "learning_rate": 8.862324814509481e-05,
+      "loss": 4.0662,
       "step": 4925
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.658697444352844e-05,
-      "loss": 0.5587,
+      "learning_rate": 8.841714756801319e-05,
+      "loss": 4.0381,
       "step": 4926
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6525144270403957e-05,
-      "loss": 0.5021,
+      "learning_rate": 8.821104699093157e-05,
+      "loss": 4.1145,
       "step": 4927
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.646331409727947e-05,
-      "loss": 0.7141,
+      "learning_rate": 8.800494641384995e-05,
+      "loss": 4.3163,
       "step": 4928
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6401483924154983e-05,
-      "loss": 0.4792,
+      "learning_rate": 8.779884583676834e-05,
+      "loss": 3.9003,
       "step": 4929
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6339653751030503e-05,
-      "loss": 0.4483,
+      "learning_rate": 8.759274525968674e-05,
+      "loss": 3.9313,
       "step": 4930
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.6277823577906017e-05,
-      "loss": 0.4778,
+      "learning_rate": 8.738664468260512e-05,
+      "loss": 4.1562,
       "step": 4931
     },
     {
       "epoch": 2.76,
-      "learning_rate": 2.621599340478153e-05,
-      "loss": 0.7086,
+      "learning_rate": 8.71805441055235e-05,
+      "loss": 4.5369,
       "step": 4932
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.6154163231657046e-05,
-      "loss": 0.4909,
+      "learning_rate": 8.697444352844189e-05,
+      "loss": 4.2737,
       "step": 4933
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.6092333058532563e-05,
-      "loss": 1.0768,
+      "learning_rate": 8.676834295136027e-05,
+      "loss": 4.0935,
       "step": 4934
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.6030502885408076e-05,
-      "loss": 0.6224,
+      "learning_rate": 8.656224237427865e-05,
+      "loss": 3.9734,
       "step": 4935
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5968672712283593e-05,
-      "loss": 0.6131,
+      "learning_rate": 8.635614179719703e-05,
+      "loss": 4.4055,
       "step": 4936
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5906842539159106e-05,
-      "loss": 0.6396,
+      "learning_rate": 8.615004122011542e-05,
+      "loss": 4.1519,
       "step": 4937
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5845012366034626e-05,
-      "loss": 0.5124,
+      "learning_rate": 8.59439406430338e-05,
+      "loss": 4.1157,
       "step": 4938
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.578318219291014e-05,
-      "loss": 0.4799,
+      "learning_rate": 8.573784006595218e-05,
+      "loss": 4.1236,
       "step": 4939
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5721352019785652e-05,
-      "loss": 0.5859,
+      "learning_rate": 8.553173948887056e-05,
+      "loss": 4.0669,
       "step": 4940
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5659521846661166e-05,
-      "loss": 0.3815,
+      "learning_rate": 8.532563891178897e-05,
+      "loss": 3.9359,
       "step": 4941
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5597691673536685e-05,
-      "loss": 0.6162,
+      "learning_rate": 8.511953833470735e-05,
+      "loss": 3.9318,
       "step": 4942
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.55358615004122e-05,
-      "loss": 0.4978,
+      "learning_rate": 8.491343775762573e-05,
+      "loss": 3.985,
       "step": 4943
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5474031327287715e-05,
-      "loss": 0.6989,
+      "learning_rate": 8.47073371805441e-05,
+      "loss": 4.2771,
       "step": 4944
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.541220115416323e-05,
-      "loss": 0.4154,
+      "learning_rate": 8.45012366034625e-05,
+      "loss": 3.9242,
       "step": 4945
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5350370981038745e-05,
-      "loss": 0.5382,
+      "learning_rate": 8.429513602638088e-05,
+      "loss": 4.0654,
       "step": 4946
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.528854080791426e-05,
-      "loss": 0.4898,
+      "learning_rate": 8.408903544929926e-05,
+      "loss": 3.9158,
       "step": 4947
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5226710634789775e-05,
-      "loss": 0.8497,
+      "learning_rate": 8.388293487221764e-05,
+      "loss": 4.3161,
       "step": 4948
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5164880461665288e-05,
-      "loss": 0.596,
+      "learning_rate": 8.367683429513603e-05,
+      "loss": 4.1337,
       "step": 4949
     },
     {
       "epoch": 2.77,
-      "learning_rate": 2.5103050288540808e-05,
-      "loss": 0.4983,
+      "learning_rate": 8.347073371805441e-05,
+      "loss": 3.8717,
       "step": 4950
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.504122011541632e-05,
-      "loss": 0.4661,
+      "learning_rate": 8.326463314097279e-05,
+      "loss": 4.1009,
       "step": 4951
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4979389942291834e-05,
-      "loss": 0.6458,
+      "learning_rate": 8.305853256389117e-05,
+      "loss": 4.0677,
       "step": 4952
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.491755976916735e-05,
-      "loss": 0.5585,
+      "learning_rate": 8.285243198680958e-05,
+      "loss": 4.2589,
       "step": 4953
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4855729596042868e-05,
-      "loss": 0.4435,
+      "learning_rate": 8.264633140972796e-05,
+      "loss": 3.8468,
       "step": 4954
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4793899422918384e-05,
-      "loss": 0.6545,
+      "learning_rate": 8.244023083264634e-05,
+      "loss": 3.9947,
       "step": 4955
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4732069249793897e-05,
-      "loss": 0.4578,
+      "learning_rate": 8.223413025556472e-05,
+      "loss": 3.7464,
       "step": 4956
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.467023907666941e-05,
-      "loss": 0.5838,
+      "learning_rate": 8.202802967848311e-05,
+      "loss": 4.03,
       "step": 4957
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.460840890354493e-05,
-      "loss": 0.519,
+      "learning_rate": 8.182192910140149e-05,
+      "loss": 4.2251,
       "step": 4958
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4546578730420444e-05,
-      "loss": 0.4154,
+      "learning_rate": 8.161582852431987e-05,
+      "loss": 4.0451,
       "step": 4959
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4484748557295957e-05,
-      "loss": 0.7069,
+      "learning_rate": 8.140972794723825e-05,
+      "loss": 4.0621,
       "step": 4960
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4422918384171474e-05,
-      "loss": 0.2884,
+      "learning_rate": 8.120362737015664e-05,
+      "loss": 3.6358,
       "step": 4961
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.436108821104699e-05,
-      "loss": 0.4631,
+      "learning_rate": 8.099752679307502e-05,
+      "loss": 3.6623,
       "step": 4962
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4299258037922503e-05,
-      "loss": 0.5266,
+      "learning_rate": 8.07914262159934e-05,
+      "loss": 3.7436,
       "step": 4963
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.423742786479802e-05,
-      "loss": 0.6228,
+      "learning_rate": 8.058532563891178e-05,
+      "loss": 4.141,
       "step": 4964
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4175597691673533e-05,
-      "loss": 0.2278,
+      "learning_rate": 8.037922506183019e-05,
+      "loss": 3.5503,
       "step": 4965
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4113767518549053e-05,
-      "loss": 0.7495,
+      "learning_rate": 8.017312448474857e-05,
+      "loss": 3.8188,
       "step": 4966
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.4051937345424566e-05,
-      "loss": 0.5075,
+      "learning_rate": 7.996702390766695e-05,
+      "loss": 3.7271,
       "step": 4967
     },
     {
       "epoch": 2.78,
-      "learning_rate": 2.399010717230008e-05,
-      "loss": 0.7033,
+      "learning_rate": 7.976092333058532e-05,
+      "loss": 3.376,
       "step": 4968
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3928276999175593e-05,
-      "loss": 0.7029,
+      "learning_rate": 7.955482275350372e-05,
+      "loss": 4.1115,
       "step": 4969
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3866446826051113e-05,
-      "loss": 0.5457,
+      "learning_rate": 7.93487221764221e-05,
+      "loss": 4.1032,
       "step": 4970
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3804616652926626e-05,
-      "loss": 0.7123,
+      "learning_rate": 7.914262159934048e-05,
+      "loss": 4.1524,
       "step": 4971
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3742786479802142e-05,
-      "loss": 0.5904,
+      "learning_rate": 7.893652102225886e-05,
+      "loss": 3.9931,
       "step": 4972
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3680956306677656e-05,
-      "loss": 0.5814,
+      "learning_rate": 7.873042044517725e-05,
+      "loss": 4.0175,
       "step": 4973
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3619126133553172e-05,
-      "loss": 0.5126,
+      "learning_rate": 7.852431986809563e-05,
+      "loss": 4.0084,
       "step": 4974
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.355729596042869e-05,
-      "loss": 0.4418,
+      "learning_rate": 7.831821929101401e-05,
+      "loss": 3.8754,
       "step": 4975
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3495465787304202e-05,
-      "loss": 0.6303,
+      "learning_rate": 7.811211871393239e-05,
+      "loss": 4.2908,
       "step": 4976
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3433635614179715e-05,
-      "loss": 0.4401,
+      "learning_rate": 7.79060181368508e-05,
+      "loss": 3.9824,
       "step": 4977
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3371805441055235e-05,
-      "loss": 0.4902,
+      "learning_rate": 7.769991755976917e-05,
+      "loss": 4.1766,
       "step": 4978
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.330997526793075e-05,
-      "loss": 0.5498,
+      "learning_rate": 7.749381698268755e-05,
+      "loss": 4.1298,
       "step": 4979
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.324814509480626e-05,
-      "loss": 0.5129,
+      "learning_rate": 7.728771640560593e-05,
+      "loss": 3.9726,
       "step": 4980
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3186314921681778e-05,
-      "loss": 0.6535,
+      "learning_rate": 7.708161582852433e-05,
+      "loss": 4.0816,
       "step": 4981
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3124484748557295e-05,
-      "loss": 0.4256,
+      "learning_rate": 7.68755152514427e-05,
+      "loss": 3.7708,
       "step": 4982
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3062654575432808e-05,
-      "loss": 0.5084,
+      "learning_rate": 7.666941467436109e-05,
+      "loss": 4.2322,
       "step": 4983
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.3000824402308325e-05,
-      "loss": 0.3738,
+      "learning_rate": 7.646331409727947e-05,
+      "loss": 3.9224,
       "step": 4984
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.2938994229183838e-05,
-      "loss": 0.5167,
+      "learning_rate": 7.625721352019786e-05,
+      "loss": 3.9643,
       "step": 4985
     },
     {
       "epoch": 2.79,
-      "learning_rate": 2.2877164056059358e-05,
-      "loss": 0.5382,
+      "learning_rate": 7.605111294311624e-05,
+      "loss": 3.905,
       "step": 4986
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.281533388293487e-05,
-      "loss": 0.3562,
+      "learning_rate": 7.584501236603462e-05,
+      "loss": 3.803,
       "step": 4987
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2753503709810384e-05,
-      "loss": 0.443,
+      "learning_rate": 7.563891178895301e-05,
+      "loss": 4.1213,
       "step": 4988
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.26916735366859e-05,
-      "loss": 0.4065,
+      "learning_rate": 7.54328112118714e-05,
+      "loss": 4.0797,
       "step": 4989
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2629843363561417e-05,
-      "loss": 0.481,
+      "learning_rate": 7.522671063478978e-05,
+      "loss": 4.1437,
       "step": 4990
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.256801319043693e-05,
-      "loss": 0.6236,
+      "learning_rate": 7.502061005770816e-05,
+      "loss": 3.9164,
       "step": 4991
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2506183017312447e-05,
-      "loss": 0.5323,
+      "learning_rate": 7.481450948062654e-05,
+      "loss": 3.8516,
       "step": 4992
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.244435284418796e-05,
-      "loss": 0.5176,
+      "learning_rate": 7.460840890354494e-05,
+      "loss": 4.1073,
       "step": 4993
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2382522671063477e-05,
-      "loss": 0.4302,
+      "learning_rate": 7.440230832646332e-05,
+      "loss": 4.1057,
       "step": 4994
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2320692497938993e-05,
-      "loss": 0.6067,
+      "learning_rate": 7.41962077493817e-05,
+      "loss": 4.1811,
       "step": 4995
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2258862324814507e-05,
-      "loss": 0.3255,
+      "learning_rate": 7.399010717230008e-05,
+      "loss": 3.6598,
       "step": 4996
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.219703215169002e-05,
-      "loss": 0.463,
+      "learning_rate": 7.378400659521847e-05,
+      "loss": 4.0507,
       "step": 4997
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.213520197856554e-05,
-      "loss": 0.4728,
+      "learning_rate": 7.357790601813685e-05,
+      "loss": 4.1363,
       "step": 4998
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2073371805441053e-05,
-      "loss": 0.6834,
+      "learning_rate": 7.337180544105523e-05,
+      "loss": 4.2833,
       "step": 4999
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.2011541632316566e-05,
-      "loss": 0.4566,
+      "learning_rate": 7.316570486397362e-05,
+      "loss": 3.8714,
       "step": 5000
     },
     {
       "epoch": 2.8,
-      "eval_loss": 1.086005687713623,
-      "eval_runtime": 1253.3524,
-      "eval_samples_per_second": 2.108,
-      "eval_steps_per_second": 0.264,
-      "eval_wer": 1.055850852836176,
+      "eval_loss": 5.234638214111328,
+      "eval_runtime": 1248.2731,
+      "eval_samples_per_second": 2.117,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.9970646568821897,
       "step": 5000
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.1949711459192083e-05,
-      "loss": 0.445,
+      "learning_rate": 7.295960428689201e-05,
+      "loss": 3.925,
       "step": 5001
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.18878812860676e-05,
-      "loss": 0.5702,
+      "learning_rate": 7.27535037098104e-05,
+      "loss": 4.1954,
       "step": 5002
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.1826051112943116e-05,
-      "loss": 0.5254,
+      "learning_rate": 7.254740313272877e-05,
+      "loss": 3.8644,
       "step": 5003
     },
     {
       "epoch": 2.8,
-      "learning_rate": 2.176422093981863e-05,
-      "loss": 0.5405,
+      "learning_rate": 7.234130255564715e-05,
+      "loss": 4.1649,
       "step": 5004
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1702390766694142e-05,
-      "loss": 0.728,
+      "learning_rate": 7.213520197856555e-05,
+      "loss": 3.9817,
       "step": 5005
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1640560593569662e-05,
-      "loss": 0.5176,
+      "learning_rate": 7.192910140148393e-05,
+      "loss": 4.116,
       "step": 5006
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1578730420445176e-05,
-      "loss": 0.3405,
+      "learning_rate": 7.17230008244023e-05,
+      "loss": 3.9249,
       "step": 5007
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.151690024732069e-05,
-      "loss": 0.6375,
+      "learning_rate": 7.151690024732068e-05,
+      "loss": 4.2712,
       "step": 5008
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1455070074196205e-05,
-      "loss": 0.3338,
+      "learning_rate": 7.131079967023908e-05,
+      "loss": 3.7553,
       "step": 5009
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1393239901071722e-05,
-      "loss": 0.4282,
+      "learning_rate": 7.110469909315746e-05,
+      "loss": 3.9739,
       "step": 5010
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1331409727947235e-05,
-      "loss": 0.7284,
+      "learning_rate": 7.089859851607585e-05,
+      "loss": 4.2576,
       "step": 5011
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.126957955482275e-05,
-      "loss": 0.5127,
+      "learning_rate": 7.069249793899423e-05,
+      "loss": 3.9414,
       "step": 5012
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1207749381698265e-05,
-      "loss": 0.5661,
+      "learning_rate": 7.048639736191262e-05,
+      "loss": 3.9823,
       "step": 5013
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1145919208573785e-05,
-      "loss": 0.6669,
+      "learning_rate": 7.0280296784831e-05,
+      "loss": 3.7228,
       "step": 5014
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.1084089035449298e-05,
-      "loss": 0.5827,
+      "learning_rate": 7.007419620774938e-05,
+      "loss": 3.5145,
       "step": 5015
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.102225886232481e-05,
-      "loss": 0.8898,
+      "learning_rate": 6.986809563066776e-05,
+      "loss": 3.7612,
       "step": 5016
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.0960428689200324e-05,
-      "loss": 0.5003,
+      "learning_rate": 6.966199505358615e-05,
+      "loss": 3.6675,
       "step": 5017
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.0898598516075844e-05,
-      "loss": 0.5983,
+      "learning_rate": 6.945589447650453e-05,
+      "loss": 3.4239,
       "step": 5018
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.0836768342951358e-05,
-      "loss": 0.6977,
+      "learning_rate": 6.924979389942291e-05,
+      "loss": 3.9699,
       "step": 5019
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.0774938169826874e-05,
-      "loss": 0.6952,
+      "learning_rate": 6.90436933223413e-05,
+      "loss": 4.1043,
       "step": 5020
     },
     {
       "epoch": 2.81,
-      "learning_rate": 2.0713107996702387e-05,
-      "loss": 0.7674,
+      "learning_rate": 6.883759274525969e-05,
+      "loss": 4.2392,
       "step": 5021
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0651277823577904e-05,
-      "loss": 0.752,
+      "learning_rate": 6.863149216817807e-05,
+      "loss": 4.3094,
       "step": 5022
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.058944765045342e-05,
-      "loss": 0.7364,
+      "learning_rate": 6.842539159109646e-05,
+      "loss": 4.3282,
       "step": 5023
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0527617477328934e-05,
-      "loss": 0.5069,
+      "learning_rate": 6.821929101401484e-05,
+      "loss": 3.9143,
       "step": 5024
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0465787304204447e-05,
-      "loss": 0.4253,
+      "learning_rate": 6.801319043693323e-05,
+      "loss": 3.969,
       "step": 5025
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0403957131079967e-05,
-      "loss": 0.7113,
+      "learning_rate": 6.780708985985161e-05,
+      "loss": 4.253,
       "step": 5026
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.034212695795548e-05,
-      "loss": 0.4718,
+      "learning_rate": 6.760098928276999e-05,
+      "loss": 3.8793,
       "step": 5027
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0280296784830993e-05,
-      "loss": 0.6638,
+      "learning_rate": 6.739488870568838e-05,
+      "loss": 4.1997,
       "step": 5028
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0218466611706513e-05,
-      "loss": 0.7708,
+      "learning_rate": 6.718878812860676e-05,
+      "loss": 4.1319,
       "step": 5029
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0156636438582027e-05,
-      "loss": 0.571,
+      "learning_rate": 6.698268755152514e-05,
+      "loss": 3.9122,
       "step": 5030
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0094806265457543e-05,
-      "loss": 0.5411,
+      "learning_rate": 6.677658697444352e-05,
+      "loss": 4.3918,
       "step": 5031
     },
     {
       "epoch": 2.82,
-      "learning_rate": 2.0032976092333056e-05,
-      "loss": 0.332,
+      "learning_rate": 6.657048639736192e-05,
+      "loss": 3.8107,
       "step": 5032
     },
     {
       "epoch": 2.82,
-      "learning_rate": 1.9971145919208573e-05,
-      "loss": 0.6233,
+      "learning_rate": 6.63643858202803e-05,
+      "loss": 4.026,
       "step": 5033
     },
     {
       "epoch": 2.82,
-      "learning_rate": 1.990931574608409e-05,
-      "loss": 0.335,
+      "learning_rate": 6.615828524319868e-05,
+      "loss": 3.7601,
       "step": 5034
     },
     {
       "epoch": 2.82,
-      "learning_rate": 1.9847485572959603e-05,
-      "loss": 0.5157,
+      "learning_rate": 6.595218466611707e-05,
+      "loss": 3.9656,
       "step": 5035
     },
     {
       "epoch": 2.82,
-      "learning_rate": 1.9785655399835116e-05,
-      "loss": 0.5173,
+      "learning_rate": 6.574608408903546e-05,
+      "loss": 3.8697,
       "step": 5036
     },
     {
       "epoch": 2.82,
-      "learning_rate": 1.9723825226710636e-05,
-      "loss": 0.456,
+      "learning_rate": 6.553998351195384e-05,
+      "loss": 3.8761,
       "step": 5037
     },
     {
       "epoch": 2.82,
-      "learning_rate": 1.966199505358615e-05,
-      "loss": 0.548,
+      "learning_rate": 6.533388293487222e-05,
+      "loss": 4.1502,
       "step": 5038
     },
     {
       "epoch": 2.82,
-      "learning_rate": 1.9600164880461662e-05,
-      "loss": 0.6084,
+      "learning_rate": 6.51277823577906e-05,
+      "loss": 4.0038,
       "step": 5039
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.953833470733718e-05,
-      "loss": 0.4913,
+      "learning_rate": 6.4921681780709e-05,
+      "loss": 4.3661,
       "step": 5040
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.9476504534212695e-05,
-      "loss": 0.5553,
+      "learning_rate": 6.471558120362737e-05,
+      "loss": 4.0039,
       "step": 5041
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.941467436108821e-05,
-      "loss": 0.5178,
+      "learning_rate": 6.450948062654575e-05,
+      "loss": 3.933,
       "step": 5042
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.9352844187963725e-05,
-      "loss": 0.6374,
+      "learning_rate": 6.430338004946413e-05,
+      "loss": 4.3495,
       "step": 5043
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.929101401483924e-05,
-      "loss": 0.6468,
+      "learning_rate": 6.409727947238253e-05,
+      "loss": 4.3116,
       "step": 5044
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.922918384171476e-05,
-      "loss": 0.3479,
+      "learning_rate": 6.38911788953009e-05,
+      "loss": 3.8859,
       "step": 5045
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.916735366859027e-05,
-      "loss": 0.5416,
+      "learning_rate": 6.36850783182193e-05,
+      "loss": 4.1695,
       "step": 5046
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.9105523495465785e-05,
-      "loss": 0.452,
+      "learning_rate": 6.347897774113768e-05,
+      "loss": 3.977,
       "step": 5047
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.9043693322341298e-05,
-      "loss": 0.5422,
+      "learning_rate": 6.327287716405607e-05,
+      "loss": 3.8761,
       "step": 5048
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.8981863149216818e-05,
-      "loss": 0.3944,
+      "learning_rate": 6.306677658697445e-05,
+      "loss": 3.8942,
       "step": 5049
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.892003297609233e-05,
-      "loss": 0.6941,
+      "learning_rate": 6.286067600989283e-05,
+      "loss": 4.2719,
       "step": 5050
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.8858202802967848e-05,
-      "loss": 0.4215,
+      "learning_rate": 6.265457543281121e-05,
+      "loss": 3.9412,
       "step": 5051
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.879637262984336e-05,
-      "loss": 0.549,
+      "learning_rate": 6.24484748557296e-05,
+      "loss": 3.9942,
       "step": 5052
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.8734542456718878e-05,
-      "loss": 0.402,
+      "learning_rate": 6.224237427864798e-05,
+      "loss": 4.0148,
       "step": 5053
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.8672712283594394e-05,
-      "loss": 0.3587,
+      "learning_rate": 6.203627370156636e-05,
+      "loss": 3.6956,
       "step": 5054
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.8610882110469907e-05,
-      "loss": 0.7089,
+      "learning_rate": 6.183017312448476e-05,
+      "loss": 3.9805,
       "step": 5055
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.8549051937345424e-05,
-      "loss": 0.4166,
+      "learning_rate": 6.162407254740313e-05,
+      "loss": 3.8139,
       "step": 5056
     },
     {
       "epoch": 2.83,
-      "learning_rate": 1.8487221764220937e-05,
-      "loss": 1.0941,
+      "learning_rate": 6.141797197032151e-05,
+      "loss": 4.5294,
       "step": 5057
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.8425391591096454e-05,
-      "loss": 0.4487,
+      "learning_rate": 6.121187139323991e-05,
+      "loss": 3.7592,
       "step": 5058
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.8363561417971967e-05,
-      "loss": 0.758,
+      "learning_rate": 6.100577081615829e-05,
+      "loss": 4.2819,
       "step": 5059
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.8301731244847483e-05,
-      "loss": 0.4831,
+      "learning_rate": 6.079967023907667e-05,
+      "loss": 4.0447,
       "step": 5060
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.8239901071723e-05,
-      "loss": 0.4813,
+      "learning_rate": 6.059356966199506e-05,
+      "loss": 3.8971,
       "step": 5061
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.8178070898598517e-05,
-      "loss": 0.8639,
+      "learning_rate": 6.038746908491344e-05,
+      "loss": 3.9026,
       "step": 5062
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.811624072547403e-05,
-      "loss": 0.4898,
+      "learning_rate": 6.0181368507831826e-05,
+      "loss": 3.7493,
       "step": 5063
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.8054410552349546e-05,
-      "loss": 0.8664,
+      "learning_rate": 5.9975267930750205e-05,
+      "loss": 3.8572,
       "step": 5064
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.799258037922506e-05,
-      "loss": 0.4517,
+      "learning_rate": 5.976916735366859e-05,
+      "loss": 3.5972,
       "step": 5065
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.7930750206100576e-05,
-      "loss": 0.442,
+      "learning_rate": 5.956306677658697e-05,
+      "loss": 3.4988,
       "step": 5066
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.786892003297609e-05,
-      "loss": 0.4213,
+      "learning_rate": 5.9356966199505365e-05,
+      "loss": 3.2196,
       "step": 5067
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.7807089859851606e-05,
-      "loss": 0.9372,
+      "learning_rate": 5.9150865622423744e-05,
+      "loss": 3.5891,
       "step": 5068
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.7745259686727123e-05,
-      "loss": 0.4718,
+      "learning_rate": 5.894476504534213e-05,
+      "loss": 4.0832,
       "step": 5069
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.7683429513602636e-05,
-      "loss": 0.6537,
+      "learning_rate": 5.873866446826051e-05,
+      "loss": 4.3453,
       "step": 5070
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.7621599340478152e-05,
-      "loss": 0.7061,
+      "learning_rate": 5.8532563891178896e-05,
+      "loss": 4.2198,
       "step": 5071
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.755976916735367e-05,
-      "loss": 0.5397,
+      "learning_rate": 5.8326463314097276e-05,
+      "loss": 4.0503,
       "step": 5072
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.7497938994229182e-05,
-      "loss": 0.4063,
+      "learning_rate": 5.812036273701567e-05,
+      "loss": 4.0209,
       "step": 5073
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.74361088211047e-05,
-      "loss": 0.6276,
+      "learning_rate": 5.791426215993405e-05,
+      "loss": 4.1641,
       "step": 5074
     },
     {
       "epoch": 2.84,
-      "learning_rate": 1.7374278647980212e-05,
-      "loss": 0.3667,
+      "learning_rate": 5.7708161582852435e-05,
+      "loss": 3.8484,
       "step": 5075
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.731244847485573e-05,
-      "loss": 0.4192,
+      "learning_rate": 5.7502061005770815e-05,
+      "loss": 4.0826,
       "step": 5076
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.7250618301731242e-05,
-      "loss": 1.0201,
+      "learning_rate": 5.72959604286892e-05,
+      "loss": 4.3472,
       "step": 5077
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.7188788128606758e-05,
-      "loss": 0.6046,
+      "learning_rate": 5.708985985160758e-05,
+      "loss": 4.1495,
       "step": 5078
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.7126957955482275e-05,
-      "loss": 0.4954,
+      "learning_rate": 5.6883759274525974e-05,
+      "loss": 4.0718,
       "step": 5079
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.7065127782357788e-05,
-      "loss": 0.4861,
+      "learning_rate": 5.667765869744435e-05,
+      "loss": 3.9429,
       "step": 5080
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.7003297609233305e-05,
-      "loss": 0.6215,
+      "learning_rate": 5.647155812036274e-05,
+      "loss": 4.1381,
       "step": 5081
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.694146743610882e-05,
-      "loss": 0.3735,
+      "learning_rate": 5.626545754328112e-05,
+      "loss": 3.5712,
       "step": 5082
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6879637262984334e-05,
-      "loss": 0.7503,
+      "learning_rate": 5.6059356966199506e-05,
+      "loss": 4.3062,
       "step": 5083
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.681780708985985e-05,
-      "loss": 0.5742,
+      "learning_rate": 5.585325638911789e-05,
+      "loss": 4.0753,
       "step": 5084
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6755976916735364e-05,
-      "loss": 0.3972,
+      "learning_rate": 5.564715581203628e-05,
+      "loss": 3.9187,
       "step": 5085
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.669414674361088e-05,
-      "loss": 0.4552,
+      "learning_rate": 5.544105523495466e-05,
+      "loss": 3.8732,
       "step": 5086
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6632316570486394e-05,
-      "loss": 0.4831,
+      "learning_rate": 5.5234954657873044e-05,
+      "loss": 3.9193,
       "step": 5087
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.657048639736191e-05,
-      "loss": 0.432,
+      "learning_rate": 5.5028854080791424e-05,
+      "loss": 4.1241,
       "step": 5088
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6508656224237427e-05,
-      "loss": 0.4534,
+      "learning_rate": 5.482275350370981e-05,
+      "loss": 3.8449,
       "step": 5089
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6446826051112944e-05,
-      "loss": 0.5014,
+      "learning_rate": 5.46166529266282e-05,
+      "loss": 3.8768,
       "step": 5090
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6384995877988457e-05,
-      "loss": 0.2895,
+      "learning_rate": 5.441055234954658e-05,
+      "loss": 3.767,
       "step": 5091
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6323165704863974e-05,
-      "loss": 0.596,
+      "learning_rate": 5.420445177246496e-05,
+      "loss": 4.0612,
       "step": 5092
     },
     {
       "epoch": 2.85,
-      "learning_rate": 1.6261335531739487e-05,
-      "loss": 0.455,
+      "learning_rate": 5.399835119538335e-05,
+      "loss": 4.0057,
       "step": 5093
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.6199505358615003e-05,
-      "loss": 0.914,
+      "learning_rate": 5.379225061830173e-05,
+      "loss": 4.0735,
       "step": 5094
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.6137675185490517e-05,
-      "loss": 0.3143,
+      "learning_rate": 5.3586150041220115e-05,
+      "loss": 3.7822,
       "step": 5095
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.6075845012366033e-05,
-      "loss": 0.4522,
+      "learning_rate": 5.33800494641385e-05,
+      "loss": 3.906,
       "step": 5096
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.6014014839241546e-05,
-      "loss": 0.2682,
+      "learning_rate": 5.317394888705689e-05,
+      "loss": 3.8806,
       "step": 5097
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5952184666117063e-05,
-      "loss": 0.6628,
+      "learning_rate": 5.296784830997527e-05,
+      "loss": 3.9278,
       "step": 5098
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.589035449299258e-05,
-      "loss": 0.5424,
+      "learning_rate": 5.2761747732893654e-05,
+      "loss": 3.937,
       "step": 5099
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5828524319868096e-05,
-      "loss": 0.416,
+      "learning_rate": 5.255564715581203e-05,
+      "loss": 3.8321,
       "step": 5100
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.576669414674361e-05,
-      "loss": 0.5098,
+      "learning_rate": 5.234954657873042e-05,
+      "loss": 3.9065,
       "step": 5101
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5704863973619126e-05,
-      "loss": 0.3969,
+      "learning_rate": 5.2143446001648806e-05,
+      "loss": 3.8293,
       "step": 5102
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.564303380049464e-05,
-      "loss": 0.4966,
+      "learning_rate": 5.193734542456719e-05,
+      "loss": 4.2167,
       "step": 5103
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5581203627370156e-05,
-      "loss": 0.4247,
+      "learning_rate": 5.173124484748557e-05,
+      "loss": 3.7697,
       "step": 5104
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.551937345424567e-05,
-      "loss": 0.3994,
+      "learning_rate": 5.152514427040396e-05,
+      "loss": 3.6977,
       "step": 5105
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5457543281121185e-05,
-      "loss": 0.3568,
+      "learning_rate": 5.131904369332234e-05,
+      "loss": 3.8337,
       "step": 5106
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.53957131079967e-05,
-      "loss": 0.4479,
+      "learning_rate": 5.1112943116240724e-05,
+      "loss": 3.8683,
       "step": 5107
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5333882934872215e-05,
-      "loss": 0.3501,
+      "learning_rate": 5.090684253915911e-05,
+      "loss": 3.9502,
       "step": 5108
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5272052761747732e-05,
-      "loss": 0.3622,
+      "learning_rate": 5.07007419620775e-05,
+      "loss": 3.849,
       "step": 5109
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5210222588623247e-05,
-      "loss": 0.765,
+      "learning_rate": 5.0494641384995876e-05,
+      "loss": 3.7362,
       "step": 5110
     },
     {
       "epoch": 2.86,
-      "learning_rate": 1.5148392415498762e-05,
-      "loss": 0.4316,
+      "learning_rate": 5.028854080791426e-05,
+      "loss": 3.6482,
       "step": 5111
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.5086562242374278e-05,
-      "loss": 0.4463,
+      "learning_rate": 5.008244023083264e-05,
+      "loss": 3.7275,
       "step": 5112
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.5024732069249791e-05,
-      "loss": 0.4369,
+      "learning_rate": 4.9876339653751036e-05,
+      "loss": 3.5425,
       "step": 5113
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4962901896125308e-05,
-      "loss": 0.3441,
+      "learning_rate": 4.9670239076669415e-05,
+      "loss": 3.5925,
       "step": 5114
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4901071723000823e-05,
-      "loss": 0.6053,
+      "learning_rate": 4.94641384995878e-05,
+      "loss": 4.0113,
       "step": 5115
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4839241549876338e-05,
-      "loss": 0.5833,
+      "learning_rate": 4.925803792250618e-05,
+      "loss": 3.9936,
       "step": 5116
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4777411376751853e-05,
-      "loss": 0.6253,
+      "learning_rate": 4.905193734542457e-05,
+      "loss": 3.6822,
       "step": 5117
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.471558120362737e-05,
-      "loss": 0.7115,
+      "learning_rate": 4.884583676834295e-05,
+      "loss": 3.1375,
       "step": 5118
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4653751030502882e-05,
-      "loss": 0.716,
+      "learning_rate": 4.863973619126134e-05,
+      "loss": 4.5436,
       "step": 5119
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4591920857378399e-05,
-      "loss": 0.6556,
+      "learning_rate": 4.843363561417972e-05,
+      "loss": 4.263,
       "step": 5120
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4530090684253914e-05,
-      "loss": 0.4447,
+      "learning_rate": 4.8227535037098106e-05,
+      "loss": 3.9375,
       "step": 5121
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.446826051112943e-05,
-      "loss": 0.6889,
+      "learning_rate": 4.8021434460016486e-05,
+      "loss": 4.0512,
       "step": 5122
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4406430338004944e-05,
-      "loss": 0.5338,
+      "learning_rate": 4.781533388293487e-05,
+      "loss": 4.1574,
       "step": 5123
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.434460016488046e-05,
-      "loss": 0.5788,
+      "learning_rate": 4.760923330585325e-05,
+      "loss": 3.9939,
       "step": 5124
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4282769991755975e-05,
-      "loss": 0.6175,
+      "learning_rate": 4.7403132728771645e-05,
+      "loss": 4.2656,
       "step": 5125
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4220939818631492e-05,
-      "loss": 0.7322,
+      "learning_rate": 4.7197032151690024e-05,
+      "loss": 4.2514,
       "step": 5126
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4159109645507005e-05,
-      "loss": 0.4867,
+      "learning_rate": 4.699093157460841e-05,
+      "loss": 4.0657,
       "step": 5127
     },
     {
       "epoch": 2.87,
-      "learning_rate": 1.4097279472382522e-05,
-      "loss": 0.6108,
+      "learning_rate": 4.678483099752679e-05,
+      "loss": 4.2185,
       "step": 5128
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.4035449299258036e-05,
-      "loss": 0.4773,
+      "learning_rate": 4.657873042044518e-05,
+      "loss": 4.0487,
       "step": 5129
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3973619126133551e-05,
-      "loss": 0.5009,
+      "learning_rate": 4.6372629843363556e-05,
+      "loss": 3.9993,
       "step": 5130
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3911788953009066e-05,
-      "loss": 0.5973,
+      "learning_rate": 4.616652926628195e-05,
+      "loss": 4.2062,
       "step": 5131
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3849958779884583e-05,
-      "loss": 0.4925,
+      "learning_rate": 4.596042868920033e-05,
+      "loss": 3.8972,
       "step": 5132
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3788128606760096e-05,
-      "loss": 0.504,
+      "learning_rate": 4.5754328112118715e-05,
+      "loss": 4.0557,
       "step": 5133
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3726298433635613e-05,
-      "loss": 0.3913,
+      "learning_rate": 4.5548227535037095e-05,
+      "loss": 3.8048,
       "step": 5134
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3664468260511127e-05,
-      "loss": 0.4923,
+      "learning_rate": 4.534212695795548e-05,
+      "loss": 4.1197,
       "step": 5135
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3602638087386644e-05,
-      "loss": 0.4329,
+      "learning_rate": 4.513602638087387e-05,
+      "loss": 3.9957,
       "step": 5136
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3540807914262159e-05,
-      "loss": 0.5426,
+      "learning_rate": 4.4929925803792254e-05,
+      "loss": 3.9668,
       "step": 5137
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3478977741137674e-05,
-      "loss": 0.3691,
+      "learning_rate": 4.472382522671064e-05,
+      "loss": 3.8691,
       "step": 5138
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.341714756801319e-05,
-      "loss": 0.3209,
+      "learning_rate": 4.451772464962902e-05,
+      "loss": 3.9509,
       "step": 5139
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3355317394888704e-05,
-      "loss": 0.6528,
+      "learning_rate": 4.4311624072547406e-05,
+      "loss": 3.8592,
       "step": 5140
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.329348722176422e-05,
-      "loss": 0.6091,
+      "learning_rate": 4.4105523495465786e-05,
+      "loss": 4.1966,
       "step": 5141
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3231657048639735e-05,
-      "loss": 0.648,
+      "learning_rate": 4.389942291838417e-05,
+      "loss": 4.3523,
       "step": 5142
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3169826875515252e-05,
-      "loss": 0.4645,
+      "learning_rate": 4.369332234130256e-05,
+      "loss": 4.2822,
       "step": 5143
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3107996702390765e-05,
-      "loss": 0.5357,
+      "learning_rate": 4.3487221764220945e-05,
+      "loss": 4.1066,
       "step": 5144
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.3046166529266281e-05,
-      "loss": 0.4563,
+      "learning_rate": 4.3281121187139325e-05,
+      "loss": 3.9248,
       "step": 5145
     },
     {
       "epoch": 2.88,
-      "learning_rate": 1.2984336356141796e-05,
-      "loss": 0.3839,
+      "learning_rate": 4.307502061005771e-05,
+      "loss": 3.7842,
       "step": 5146
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2922506183017313e-05,
-      "loss": 0.4933,
+      "learning_rate": 4.286892003297609e-05,
+      "loss": 4.0734,
       "step": 5147
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2860676009892826e-05,
-      "loss": 0.4356,
+      "learning_rate": 4.2662819455894484e-05,
+      "loss": 3.8602,
       "step": 5148
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2798845836768343e-05,
-      "loss": 0.5353,
+      "learning_rate": 4.245671887881286e-05,
+      "loss": 3.8822,
       "step": 5149
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2737015663643858e-05,
-      "loss": 0.6845,
+      "learning_rate": 4.225061830173125e-05,
+      "loss": 4.1478,
       "step": 5150
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2675185490519373e-05,
-      "loss": 0.3222,
+      "learning_rate": 4.204451772464963e-05,
+      "loss": 3.8257,
       "step": 5151
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2613355317394887e-05,
-      "loss": 0.4722,
+      "learning_rate": 4.1838417147568016e-05,
+      "loss": 3.903,
       "step": 5152
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2551525144270404e-05,
-      "loss": 0.6055,
+      "learning_rate": 4.1632316570486395e-05,
+      "loss": 4.215,
       "step": 5153
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2489694971145917e-05,
-      "loss": 0.3939,
+      "learning_rate": 4.142621599340479e-05,
+      "loss": 4.1005,
       "step": 5154
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2427864798021434e-05,
-      "loss": 0.3619,
+      "learning_rate": 4.122011541632317e-05,
+      "loss": 3.8093,
       "step": 5155
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2366034624896949e-05,
-      "loss": 0.3118,
+      "learning_rate": 4.1014014839241554e-05,
+      "loss": 3.552,
       "step": 5156
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2304204451772465e-05,
-      "loss": 0.5782,
+      "learning_rate": 4.0807914262159934e-05,
+      "loss": 3.8922,
       "step": 5157
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2242374278647978e-05,
-      "loss": 1.0774,
+      "learning_rate": 4.060181368507832e-05,
+      "loss": 4.5108,
       "step": 5158
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2180544105523495e-05,
-      "loss": 0.3945,
+      "learning_rate": 4.03957131079967e-05,
+      "loss": 4.0738,
       "step": 5159
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.211871393239901e-05,
-      "loss": 0.5548,
+      "learning_rate": 4.018961253091509e-05,
+      "loss": 4.0982,
       "step": 5160
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.2056883759274527e-05,
-      "loss": 0.6938,
+      "learning_rate": 3.998351195383347e-05,
+      "loss": 3.9686,
       "step": 5161
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.199505358615004e-05,
-      "loss": 0.566,
+      "learning_rate": 3.977741137675186e-05,
+      "loss": 4.1081,
       "step": 5162
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.1933223413025556e-05,
-      "loss": 0.7557,
+      "learning_rate": 3.957131079967024e-05,
+      "loss": 3.6158,
       "step": 5163
     },
     {
       "epoch": 2.89,
-      "learning_rate": 1.1871393239901071e-05,
-      "loss": 0.4131,
+      "learning_rate": 3.9365210222588625e-05,
+      "loss": 3.846,
       "step": 5164
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1809563066776586e-05,
-      "loss": 0.9304,
+      "learning_rate": 3.9159109645507004e-05,
+      "loss": 3.8718,
       "step": 5165
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1747732893652101e-05,
-      "loss": 0.6202,
+      "learning_rate": 3.89530090684254e-05,
+      "loss": 3.6072,
       "step": 5166
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1685902720527618e-05,
-      "loss": 0.6093,
+      "learning_rate": 3.874690849134378e-05,
+      "loss": 3.2825,
       "step": 5167
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.162407254740313e-05,
-      "loss": 0.8875,
+      "learning_rate": 3.8540807914262164e-05,
+      "loss": 3.6551,
       "step": 5168
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1562242374278647e-05,
-      "loss": 0.6358,
+      "learning_rate": 3.833470733718054e-05,
+      "loss": 4.431,
       "step": 5169
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1500412201154162e-05,
-      "loss": 0.5532,
+      "learning_rate": 3.812860676009893e-05,
+      "loss": 3.8475,
       "step": 5170
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1438582028029679e-05,
-      "loss": 0.7188,
+      "learning_rate": 3.792250618301731e-05,
+      "loss": 3.9162,
       "step": 5171
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1376751854905192e-05,
-      "loss": 0.6027,
+      "learning_rate": 3.77164056059357e-05,
+      "loss": 4.2181,
       "step": 5172
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1314921681780709e-05,
-      "loss": 0.4115,
+      "learning_rate": 3.751030502885408e-05,
+      "loss": 3.7959,
       "step": 5173
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1253091508656224e-05,
-      "loss": 0.787,
+      "learning_rate": 3.730420445177247e-05,
+      "loss": 4.1886,
       "step": 5174
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1191261335531738e-05,
-      "loss": 0.3811,
+      "learning_rate": 3.709810387469085e-05,
+      "loss": 3.8347,
       "step": 5175
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1129431162407253e-05,
-      "loss": 0.6733,
+      "learning_rate": 3.6892003297609234e-05,
+      "loss": 4.2774,
       "step": 5176
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.106760098928277e-05,
-      "loss": 0.5076,
+      "learning_rate": 3.6685902720527614e-05,
+      "loss": 4.1383,
       "step": 5177
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.1005770816158283e-05,
-      "loss": 0.5452,
+      "learning_rate": 3.647980214344601e-05,
+      "loss": 4.1758,
       "step": 5178
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.09439406430338e-05,
-      "loss": 0.5268,
+      "learning_rate": 3.6273701566364386e-05,
+      "loss": 3.8702,
       "step": 5179
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.0882110469909315e-05,
-      "loss": 0.5481,
+      "learning_rate": 3.606760098928277e-05,
+      "loss": 4.0338,
       "step": 5180
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.0820280296784831e-05,
-      "loss": 0.5129,
+      "learning_rate": 3.586150041220115e-05,
+      "loss": 3.9308,
       "step": 5181
     },
     {
       "epoch": 2.9,
-      "learning_rate": 1.0758450123660344e-05,
-      "loss": 0.6165,
+      "learning_rate": 3.565539983511954e-05,
+      "loss": 4.1627,
       "step": 5182
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0696619950535861e-05,
-      "loss": 0.4585,
+      "learning_rate": 3.5449299258037925e-05,
+      "loss": 3.9854,
       "step": 5183
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0634789777411376e-05,
-      "loss": 0.4756,
+      "learning_rate": 3.524319868095631e-05,
+      "loss": 3.7288,
       "step": 5184
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0572959604286892e-05,
-      "loss": 0.5256,
+      "learning_rate": 3.503709810387469e-05,
+      "loss": 4.0555,
       "step": 5185
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0511129431162406e-05,
-      "loss": 0.4648,
+      "learning_rate": 3.483099752679308e-05,
+      "loss": 4.0818,
       "step": 5186
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0449299258037922e-05,
-      "loss": 0.4327,
+      "learning_rate": 3.462489694971146e-05,
+      "loss": 4.2081,
       "step": 5187
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0387469084913437e-05,
-      "loss": 0.4558,
+      "learning_rate": 3.441879637262984e-05,
+      "loss": 3.7993,
       "step": 5188
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0325638911788952e-05,
-      "loss": 0.4343,
+      "learning_rate": 3.421269579554823e-05,
+      "loss": 3.9008,
       "step": 5189
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0263808738664467e-05,
-      "loss": 0.4347,
+      "learning_rate": 3.4006595218466616e-05,
+      "loss": 3.8967,
       "step": 5190
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0201978565539983e-05,
-      "loss": 0.5299,
+      "learning_rate": 3.3800494641384996e-05,
+      "loss": 3.7799,
       "step": 5191
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0140148392415497e-05,
-      "loss": 0.4042,
+      "learning_rate": 3.359439406430338e-05,
+      "loss": 3.8599,
       "step": 5192
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0078318219291013e-05,
-      "loss": 0.4625,
+      "learning_rate": 3.338829348722176e-05,
+      "loss": 3.9188,
       "step": 5193
     },
     {
       "epoch": 2.91,
-      "learning_rate": 1.0016488046166528e-05,
-      "loss": 0.6595,
+      "learning_rate": 3.318219291014015e-05,
+      "loss": 4.1941,
       "step": 5194
     },
     {
       "epoch": 2.91,
-      "learning_rate": 9.954657873042045e-06,
-      "loss": 0.5352,
+      "learning_rate": 3.2976092333058534e-05,
+      "loss": 4.1234,
       "step": 5195
     },
     {
       "epoch": 2.91,
-      "learning_rate": 9.892827699917558e-06,
-      "loss": 0.4208,
+      "learning_rate": 3.276999175597692e-05,
+      "loss": 3.903,
       "step": 5196
     },
     {
       "epoch": 2.91,
-      "learning_rate": 9.830997526793075e-06,
-      "loss": 0.5487,
+      "learning_rate": 3.25638911788953e-05,
+      "loss": 4.0267,
       "step": 5197
     },
     {
       "epoch": 2.91,
-      "learning_rate": 9.76916735366859e-06,
-      "loss": 0.4238,
+      "learning_rate": 3.235779060181369e-05,
+      "loss": 3.8702,
       "step": 5198
     },
     {
       "epoch": 2.91,
-      "learning_rate": 9.707337180544104e-06,
-      "loss": 0.7194,
+      "learning_rate": 3.2151690024732066e-05,
+      "loss": 4.0343,
       "step": 5199
     },
     {
       "epoch": 2.91,
-      "learning_rate": 9.64550700741962e-06,
-      "loss": 0.5732,
+      "learning_rate": 3.194558944765045e-05,
+      "loss": 4.1882,
       "step": 5200
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.583676834295136e-06,
-      "loss": 0.5137,
+      "learning_rate": 3.173948887056884e-05,
+      "loss": 4.3037,
       "step": 5201
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.521846661170649e-06,
-      "loss": 0.4017,
+      "learning_rate": 3.1533388293487225e-05,
+      "loss": 3.8102,
       "step": 5202
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.460016488046166e-06,
-      "loss": 0.5552,
+      "learning_rate": 3.1327287716405605e-05,
+      "loss": 3.8718,
       "step": 5203
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.39818631492168e-06,
-      "loss": 0.6802,
+      "learning_rate": 3.112118713932399e-05,
+      "loss": 3.7969,
       "step": 5204
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.336356141797197e-06,
-      "loss": 0.5623,
+      "learning_rate": 3.091508656224238e-05,
+      "loss": 3.9047,
       "step": 5205
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.274525968672712e-06,
-      "loss": 0.3981,
+      "learning_rate": 3.070898598516076e-05,
+      "loss": 3.7546,
       "step": 5206
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.212695795548227e-06,
-      "loss": 0.3424,
+      "learning_rate": 3.0502885408079144e-05,
+      "loss": 3.7169,
       "step": 5207
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.150865622423742e-06,
-      "loss": 0.5923,
+      "learning_rate": 3.029678483099753e-05,
+      "loss": 4.0869,
       "step": 5208
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.089035449299258e-06,
-      "loss": 0.5279,
+      "learning_rate": 3.0090684253915913e-05,
+      "loss": 4.241,
       "step": 5209
     },
     {
       "epoch": 2.92,
-      "learning_rate": 9.027205276174773e-06,
-      "loss": 0.5247,
+      "learning_rate": 2.9884583676834296e-05,
+      "loss": 3.8232,
       "step": 5210
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.965375103050288e-06,
-      "loss": 0.5126,
+      "learning_rate": 2.9678483099752682e-05,
+      "loss": 3.5642,
       "step": 5211
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.903544929925803e-06,
-      "loss": 0.3921,
+      "learning_rate": 2.9472382522671065e-05,
+      "loss": 3.7639,
       "step": 5212
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.841714756801318e-06,
-      "loss": 0.2553,
+      "learning_rate": 2.9266281945589448e-05,
+      "loss": 3.3243,
       "step": 5213
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.779884583676834e-06,
-      "loss": 0.4868,
+      "learning_rate": 2.9060181368507835e-05,
+      "loss": 3.7364,
       "step": 5214
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.71805441055235e-06,
-      "loss": 0.9045,
+      "learning_rate": 2.8854080791426218e-05,
+      "loss": 4.0159,
       "step": 5215
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.656224237427864e-06,
-      "loss": 0.5109,
+      "learning_rate": 2.86479802143446e-05,
+      "loss": 3.3586,
       "step": 5216
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.594394064303379e-06,
-      "loss": 0.4722,
+      "learning_rate": 2.8441879637262987e-05,
+      "loss": 3.6416,
       "step": 5217
     },
     {
       "epoch": 2.92,
-      "learning_rate": 8.532563891178894e-06,
-      "loss": 0.5954,
+      "learning_rate": 2.823577906018137e-05,
+      "loss": 3.1922,
       "step": 5218
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.47073371805441e-06,
-      "loss": 0.5897,
+      "learning_rate": 2.8029678483099753e-05,
+      "loss": 4.3309,
       "step": 5219
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.408903544929926e-06,
-      "loss": 0.8124,
+      "learning_rate": 2.782357790601814e-05,
+      "loss": 4.2992,
       "step": 5220
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.34707337180544e-06,
-      "loss": 0.5905,
+      "learning_rate": 2.7617477328936522e-05,
+      "loss": 4.0482,
       "step": 5221
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.285243198680955e-06,
-      "loss": 0.6763,
+      "learning_rate": 2.7411376751854905e-05,
+      "loss": 3.8889,
       "step": 5222
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.223413025556472e-06,
-      "loss": 0.6881,
+      "learning_rate": 2.720527617477329e-05,
+      "loss": 4.3519,
       "step": 5223
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.161582852431987e-06,
-      "loss": 0.5508,
+      "learning_rate": 2.6999175597691674e-05,
+      "loss": 4.2582,
       "step": 5224
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.099752679307502e-06,
-      "loss": 0.739,
+      "learning_rate": 2.6793075020610057e-05,
+      "loss": 4.2566,
       "step": 5225
     },
     {
       "epoch": 2.93,
-      "learning_rate": 8.037922506183017e-06,
-      "loss": 0.477,
+      "learning_rate": 2.6586974443528444e-05,
+      "loss": 4.0225,
       "step": 5226
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.976092333058531e-06,
-      "loss": 0.7365,
+      "learning_rate": 2.6380873866446827e-05,
+      "loss": 4.2688,
       "step": 5227
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.914262159934048e-06,
-      "loss": 0.4491,
+      "learning_rate": 2.617477328936521e-05,
+      "loss": 3.8302,
       "step": 5228
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.852431986809563e-06,
-      "loss": 0.405,
+      "learning_rate": 2.5968672712283596e-05,
+      "loss": 3.8832,
       "step": 5229
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.790601813685078e-06,
-      "loss": 0.5348,
+      "learning_rate": 2.576257213520198e-05,
+      "loss": 4.0177,
       "step": 5230
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.728771640560593e-06,
-      "loss": 0.4194,
+      "learning_rate": 2.5556471558120362e-05,
+      "loss": 3.7836,
       "step": 5231
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.666941467436108e-06,
-      "loss": 0.5502,
+      "learning_rate": 2.535037098103875e-05,
+      "loss": 4.173,
       "step": 5232
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.605111294311623e-06,
-      "loss": 0.4523,
+      "learning_rate": 2.514427040395713e-05,
+      "loss": 3.8221,
       "step": 5233
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.543281121187139e-06,
-      "loss": 0.4564,
+      "learning_rate": 2.4938169826875518e-05,
+      "loss": 3.8122,
       "step": 5234
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.481450948062654e-06,
-      "loss": 0.4766,
+      "learning_rate": 2.47320692497939e-05,
+      "loss": 4.0053,
       "step": 5235
     },
     {
       "epoch": 2.93,
-      "learning_rate": 7.419620774938169e-06,
-      "loss": 0.599,
+      "learning_rate": 2.4525968672712284e-05,
+      "loss": 4.2828,
       "step": 5236
     },
     {
       "epoch": 2.94,
-      "learning_rate": 7.357790601813685e-06,
-      "loss": 0.4178,
+      "learning_rate": 2.431986809563067e-05,
+      "loss": 3.7595,
       "step": 5237
     },
     {
       "epoch": 2.94,
-      "learning_rate": 7.2959604286891995e-06,
-      "loss": 0.79,
+      "learning_rate": 2.4113767518549053e-05,
+      "loss": 4.2288,
       "step": 5238
     },
     {
       "epoch": 2.94,
-      "learning_rate": 7.234130255564715e-06,
-      "loss": 0.3439,
+      "learning_rate": 2.3907666941467436e-05,
+      "loss": 3.8431,
       "step": 5239
     },
     {
       "epoch": 2.94,
-      "learning_rate": 7.17230008244023e-06,
-      "loss": 0.3715,
+      "learning_rate": 2.3701566364385822e-05,
+      "loss": 3.8778,
       "step": 5240
     },
     {
       "epoch": 2.94,
-      "learning_rate": 7.110469909315746e-06,
-      "loss": 0.5352,
+      "learning_rate": 2.3495465787304205e-05,
+      "loss": 4.0112,
       "step": 5241
     },
     {
       "epoch": 2.94,
-      "learning_rate": 7.048639736191261e-06,
-      "loss": 0.3275,
+      "learning_rate": 2.328936521022259e-05,
+      "loss": 3.7304,
       "step": 5242
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.986809563066776e-06,
-      "loss": 0.435,
+      "learning_rate": 2.3083264633140975e-05,
+      "loss": 3.9794,
       "step": 5243
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.924979389942291e-06,
-      "loss": 0.7632,
+      "learning_rate": 2.2877164056059358e-05,
+      "loss": 3.9738,
       "step": 5244
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.863149216817806e-06,
-      "loss": 0.3378,
+      "learning_rate": 2.267106347897774e-05,
+      "loss": 3.7783,
       "step": 5245
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.801319043693322e-06,
-      "loss": 0.3624,
+      "learning_rate": 2.2464962901896127e-05,
+      "loss": 3.8678,
       "step": 5246
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.739488870568837e-06,
-      "loss": 0.5057,
+      "learning_rate": 2.225886232481451e-05,
+      "loss": 4.1737,
       "step": 5247
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.677658697444352e-06,
-      "loss": 0.4115,
+      "learning_rate": 2.2052761747732893e-05,
+      "loss": 3.8529,
       "step": 5248
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.6158285243198676e-06,
-      "loss": 0.443,
+      "learning_rate": 2.184666117065128e-05,
+      "loss": 4.0275,
       "step": 5249
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.5539983511953824e-06,
-      "loss": 0.4542,
+      "learning_rate": 2.1640560593569662e-05,
+      "loss": 3.9318,
       "step": 5250
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.492168178070898e-06,
-      "loss": 0.4243,
+      "learning_rate": 2.1434460016488045e-05,
+      "loss": 3.7196,
       "step": 5251
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.430338004946413e-06,
-      "loss": 0.4784,
+      "learning_rate": 2.122835943940643e-05,
+      "loss": 3.9434,
       "step": 5252
     },
     {
       "epoch": 2.94,
-      "learning_rate": 6.368507831821929e-06,
-      "loss": 0.6223,
+      "learning_rate": 2.1022258862324815e-05,
+      "loss": 4.0266,
       "step": 5253
     },
     {
       "epoch": 2.95,
-      "learning_rate": 6.306677658697444e-06,
-      "loss": 0.5296,
+      "learning_rate": 2.0816158285243198e-05,
+      "loss": 3.8123,
       "step": 5254
     },
     {
       "epoch": 2.95,
-      "learning_rate": 6.244847485572959e-06,
-      "loss": 0.4236,
+      "learning_rate": 2.0610057708161584e-05,
+      "loss": 3.7585,
       "step": 5255
     },
     {
       "epoch": 2.95,
-      "learning_rate": 6.183017312448474e-06,
-      "loss": 0.4329,
+      "learning_rate": 2.0403957131079967e-05,
+      "loss": 4.0173,
       "step": 5256
     },
     {
       "epoch": 2.95,
-      "learning_rate": 6.121187139323989e-06,
-      "loss": 0.5611,
+      "learning_rate": 2.019785655399835e-05,
+      "loss": 3.8176,
       "step": 5257
     },
     {
       "epoch": 2.95,
-      "learning_rate": 6.059356966199505e-06,
-      "loss": 0.4122,
+      "learning_rate": 1.9991755976916736e-05,
+      "loss": 3.799,
       "step": 5258
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.99752679307502e-06,
-      "loss": 0.5043,
+      "learning_rate": 1.978565539983512e-05,
+      "loss": 3.9159,
       "step": 5259
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.935696619950536e-06,
-      "loss": 0.7649,
+      "learning_rate": 1.9579554822753502e-05,
+      "loss": 4.1107,
       "step": 5260
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.8738664468260505e-06,
-      "loss": 0.3023,
+      "learning_rate": 1.937345424567189e-05,
+      "loss": 3.306,
       "step": 5261
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.812036273701565e-06,
-      "loss": 0.5156,
+      "learning_rate": 1.916735366859027e-05,
+      "loss": 3.9473,
       "step": 5262
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.750206100577081e-06,
-      "loss": 0.6182,
+      "learning_rate": 1.8961253091508655e-05,
+      "loss": 3.8644,
       "step": 5263
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.688375927452596e-06,
-      "loss": 0.4509,
+      "learning_rate": 1.875515251442704e-05,
+      "loss": 3.3905,
       "step": 5264
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.626545754328112e-06,
-      "loss": 0.2805,
+      "learning_rate": 1.8549051937345424e-05,
+      "loss": 3.1935,
       "step": 5265
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.564715581203627e-06,
-      "loss": 0.4682,
+      "learning_rate": 1.8342951360263807e-05,
+      "loss": 3.4585,
       "step": 5266
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.5028854080791415e-06,
-      "loss": 0.4289,
+      "learning_rate": 1.8136850783182193e-05,
+      "loss": 3.234,
       "step": 5267
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.441055234954657e-06,
-      "loss": 1.0768,
+      "learning_rate": 1.7930750206100576e-05,
+      "loss": 3.3882,
       "step": 5268
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.379225061830172e-06,
-      "loss": 0.6203,
+      "learning_rate": 1.7724649629018963e-05,
+      "loss": 4.3939,
       "step": 5269
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.317394888705688e-06,
-      "loss": 0.5593,
+      "learning_rate": 1.7518549051937346e-05,
+      "loss": 4.0379,
       "step": 5270
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.255564715581203e-06,
-      "loss": 0.5544,
+      "learning_rate": 1.731244847485573e-05,
+      "loss": 4.0738,
       "step": 5271
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.1937345424567185e-06,
-      "loss": 0.7031,
+      "learning_rate": 1.7106347897774115e-05,
+      "loss": 4.175,
       "step": 5272
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.1319043693322334e-06,
-      "loss": 0.5116,
+      "learning_rate": 1.6900247320692498e-05,
+      "loss": 3.9153,
       "step": 5273
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.070074196207748e-06,
-      "loss": 0.6186,
+      "learning_rate": 1.669414674361088e-05,
+      "loss": 4.3006,
       "step": 5274
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.008244023083264e-06,
-      "loss": 0.5841,
+      "learning_rate": 1.6488046166529267e-05,
+      "loss": 3.983,
       "step": 5275
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.946413849958779e-06,
-      "loss": 0.3557,
+      "learning_rate": 1.628194558944765e-05,
+      "loss": 4.0034,
       "step": 5276
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.884583676834295e-06,
-      "loss": 0.6748,
+      "learning_rate": 1.6075845012366033e-05,
+      "loss": 4.418,
       "step": 5277
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.82275350370981e-06,
-      "loss": 0.6133,
+      "learning_rate": 1.586974443528442e-05,
+      "loss": 4.0683,
       "step": 5278
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.7609233305853245e-06,
-      "loss": 0.4738,
+      "learning_rate": 1.5663643858202802e-05,
+      "loss": 3.8721,
       "step": 5279
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.69909315746084e-06,
-      "loss": 0.3048,
+      "learning_rate": 1.545754328112119e-05,
+      "loss": 3.7206,
       "step": 5280
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.637262984336356e-06,
-      "loss": 0.6059,
+      "learning_rate": 1.5251442704039572e-05,
+      "loss": 3.8008,
       "step": 5281
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.575432811211871e-06,
-      "loss": 0.4555,
+      "learning_rate": 1.5045342126957956e-05,
+      "loss": 4.0717,
       "step": 5282
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.513602638087387e-06,
-      "loss": 0.4808,
+      "learning_rate": 1.4839241549876341e-05,
+      "loss": 4.1223,
       "step": 5283
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.4517724649629015e-06,
-      "loss": 0.6649,
+      "learning_rate": 1.4633140972794724e-05,
+      "loss": 4.2332,
       "step": 5284
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.389942291838417e-06,
-      "loss": 0.5564,
+      "learning_rate": 1.4427040395713109e-05,
+      "loss": 3.8893,
       "step": 5285
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.328112118713932e-06,
-      "loss": 0.6392,
+      "learning_rate": 1.4220939818631493e-05,
+      "loss": 4.0638,
       "step": 5286
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.266281945589447e-06,
-      "loss": 0.3821,
+      "learning_rate": 1.4014839241549876e-05,
+      "loss": 4.0834,
       "step": 5287
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.204451772464963e-06,
-      "loss": 0.457,
+      "learning_rate": 1.3808738664468261e-05,
+      "loss": 3.8107,
       "step": 5288
     },
     {
       "epoch": 2.96,
-      "learning_rate": 4.142621599340478e-06,
-      "loss": 0.3897,
+      "learning_rate": 1.3602638087386646e-05,
+      "loss": 4.0629,
       "step": 5289
     },
     {
       "epoch": 2.97,
-      "learning_rate": 4.080791426215993e-06,
-      "loss": 0.3981,
+      "learning_rate": 1.3396537510305029e-05,
+      "loss": 3.9057,
       "step": 5290
     },
     {
       "epoch": 2.97,
-      "learning_rate": 4.018961253091508e-06,
-      "loss": 0.3939,
+      "learning_rate": 1.3190436933223413e-05,
+      "loss": 3.8925,
       "step": 5291
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.957131079967024e-06,
-      "loss": 0.4912,
+      "learning_rate": 1.2984336356141798e-05,
+      "loss": 3.965,
       "step": 5292
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.895300906842539e-06,
-      "loss": 0.5635,
+      "learning_rate": 1.2778235779060181e-05,
+      "loss": 4.0511,
       "step": 5293
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.833470733718054e-06,
-      "loss": 0.5019,
+      "learning_rate": 1.2572135201978566e-05,
+      "loss": 4.1415,
       "step": 5294
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.7716405605935695e-06,
-      "loss": 0.4495,
+      "learning_rate": 1.236603462489695e-05,
+      "loss": 4.3386,
       "step": 5295
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.7098103874690844e-06,
-      "loss": 0.5848,
+      "learning_rate": 1.2159934047815335e-05,
+      "loss": 3.8571,
       "step": 5296
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.6479802143445998e-06,
-      "loss": 0.2943,
+      "learning_rate": 1.1953833470733718e-05,
+      "loss": 3.6936,
       "step": 5297
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.586150041220115e-06,
-      "loss": 0.6911,
+      "learning_rate": 1.1747732893652103e-05,
+      "loss": 4.0286,
       "step": 5298
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.5243198680956304e-06,
-      "loss": 0.4242,
+      "learning_rate": 1.1541632316570487e-05,
+      "loss": 4.0293,
       "step": 5299
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.4624896949711457e-06,
-      "loss": 0.3796,
+      "learning_rate": 1.133553173948887e-05,
+      "loss": 4.0922,
       "step": 5300
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.400659521846661e-06,
-      "loss": 0.8423,
+      "learning_rate": 1.1129431162407255e-05,
+      "loss": 4.3619,
       "step": 5301
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.338829348722176e-06,
-      "loss": 0.4244,
+      "learning_rate": 1.092333058532564e-05,
+      "loss": 3.8714,
       "step": 5302
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.2769991755976912e-06,
-      "loss": 0.5697,
+      "learning_rate": 1.0717230008244023e-05,
+      "loss": 4.1102,
       "step": 5303
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.2151690024732065e-06,
-      "loss": 0.5291,
+      "learning_rate": 1.0511129431162407e-05,
+      "loss": 4.0078,
       "step": 5304
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.153338829348722e-06,
-      "loss": 0.2442,
+      "learning_rate": 1.0305028854080792e-05,
+      "loss": 3.5789,
       "step": 5305
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.091508656224237e-06,
-      "loss": 0.4207,
+      "learning_rate": 1.0098928276999175e-05,
+      "loss": 3.5775,
       "step": 5306
     },
     {
       "epoch": 2.97,
-      "learning_rate": 3.0296784830997525e-06,
-      "loss": 0.6357,
+      "learning_rate": 9.89282769991756e-06,
+      "loss": 4.02,
       "step": 5307
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.967848309975268e-06,
-      "loss": 0.9338,
+      "learning_rate": 9.686727122835944e-06,
+      "loss": 3.9004,
       "step": 5308
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.9060181368507827e-06,
-      "loss": 0.4472,
+      "learning_rate": 9.480626545754327e-06,
+      "loss": 3.8523,
       "step": 5309
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.844187963726298e-06,
-      "loss": 0.4593,
+      "learning_rate": 9.274525968672712e-06,
+      "loss": 4.0244,
       "step": 5310
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.7823577906018133e-06,
-      "loss": 0.3413,
+      "learning_rate": 9.068425391591097e-06,
+      "loss": 3.9133,
       "step": 5311
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.7205276174773286e-06,
-      "loss": 0.6199,
+      "learning_rate": 8.862324814509481e-06,
+      "loss": 3.8616,
       "step": 5312
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.658697444352844e-06,
-      "loss": 0.5416,
+      "learning_rate": 8.656224237427864e-06,
+      "loss": 3.7944,
       "step": 5313
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.5968672712283593e-06,
-      "loss": 0.4694,
+      "learning_rate": 8.450123660346249e-06,
+      "loss": 3.51,
       "step": 5314
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.535037098103874e-06,
-      "loss": 0.5195,
+      "learning_rate": 8.244023083264634e-06,
+      "loss": 3.4487,
       "step": 5315
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.4732069249793895e-06,
-      "loss": 0.2469,
+      "learning_rate": 8.037922506183017e-06,
+      "loss": 3.4854,
       "step": 5316
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.411376751854905e-06,
-      "loss": 0.3893,
+      "learning_rate": 7.831821929101401e-06,
+      "loss": 3.4285,
       "step": 5317
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.34954657873042e-06,
-      "loss": 0.506,
+      "learning_rate": 7.625721352019786e-06,
+      "loss": 3.1365,
       "step": 5318
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.2877164056059354e-06,
-      "loss": 0.6177,
+      "learning_rate": 7.4196207749381706e-06,
+      "loss": 4.1348,
       "step": 5319
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.2258862324814507e-06,
-      "loss": 0.5876,
+      "learning_rate": 7.213520197856554e-06,
+      "loss": 3.9531,
       "step": 5320
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.164056059356966e-06,
-      "loss": 0.475,
+      "learning_rate": 7.007419620774938e-06,
+      "loss": 3.8508,
       "step": 5321
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.1022258862324814e-06,
-      "loss": 0.5751,
+      "learning_rate": 6.801319043693323e-06,
+      "loss": 4.1207,
       "step": 5322
     },
     {
       "epoch": 2.98,
-      "learning_rate": 2.0403957131079967e-06,
-      "loss": 0.5654,
+      "learning_rate": 6.595218466611707e-06,
+      "loss": 4.0903,
       "step": 5323
     },
     {
       "epoch": 2.98,
-      "learning_rate": 1.978565539983512e-06,
-      "loss": 0.606,
+      "learning_rate": 6.3891178895300905e-06,
+      "loss": 4.1029,
       "step": 5324
     },
     {
       "epoch": 2.98,
-      "learning_rate": 1.916735366859027e-06,
-      "loss": 0.4987,
+      "learning_rate": 6.183017312448475e-06,
+      "loss": 4.0303,
       "step": 5325
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.8549051937345422e-06,
-      "loss": 0.4293,
+      "learning_rate": 5.976916735366859e-06,
+      "loss": 3.9227,
       "step": 5326
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.7930750206100575e-06,
-      "loss": 0.6319,
+      "learning_rate": 5.770816158285244e-06,
+      "loss": 3.9048,
       "step": 5327
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.7312448474855728e-06,
-      "loss": 0.415,
+      "learning_rate": 5.5647155812036275e-06,
+      "loss": 4.0601,
       "step": 5328
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.669414674361088e-06,
-      "loss": 0.6281,
+      "learning_rate": 5.358615004122011e-06,
+      "loss": 4.1536,
       "step": 5329
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.6075845012366033e-06,
-      "loss": 0.408,
+      "learning_rate": 5.152514427040396e-06,
+      "loss": 4.0253,
       "step": 5330
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.5457543281121186e-06,
-      "loss": 0.6545,
+      "learning_rate": 4.94641384995878e-06,
+      "loss": 4.1337,
       "step": 5331
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.483924154987634e-06,
-      "loss": 0.5464,
+      "learning_rate": 4.740313272877164e-06,
+      "loss": 3.8758,
       "step": 5332
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.422093981863149e-06,
-      "loss": 0.5231,
+      "learning_rate": 4.534212695795548e-06,
+      "loss": 4.0683,
       "step": 5333
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.3602638087386643e-06,
-      "loss": 0.4513,
+      "learning_rate": 4.328112118713932e-06,
+      "loss": 3.7324,
       "step": 5334
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.2984336356141796e-06,
-      "loss": 0.4987,
+      "learning_rate": 4.122011541632317e-06,
+      "loss": 3.9437,
       "step": 5335
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.2366034624896947e-06,
-      "loss": 0.384,
+      "learning_rate": 3.915910964550701e-06,
+      "loss": 3.7429,
       "step": 5336
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.17477328936521e-06,
-      "loss": 0.5228,
+      "learning_rate": 3.7098103874690853e-06,
+      "loss": 4.181,
       "step": 5337
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.1129431162407254e-06,
-      "loss": 0.4872,
+      "learning_rate": 3.503709810387469e-06,
+      "loss": 3.9541,
       "step": 5338
     },
     {
       "epoch": 2.99,
-      "learning_rate": 1.0511129431162407e-06,
-      "loss": 0.5103,
+      "learning_rate": 3.2976092333058533e-06,
+      "loss": 4.0066,
       "step": 5339
     },
     {
       "epoch": 2.99,
-      "learning_rate": 9.89282769991756e-07,
-      "loss": 0.4243,
+      "learning_rate": 3.0915086562242376e-06,
+      "loss": 3.8814,
       "step": 5340
     },
     {
       "epoch": 2.99,
-      "learning_rate": 9.274525968672711e-07,
-      "loss": 0.4658,
+      "learning_rate": 2.885408079142622e-06,
+      "loss": 3.797,
       "step": 5341
     },
     {
       "epoch": 2.99,
-      "learning_rate": 8.656224237427864e-07,
-      "loss": 0.3835,
+      "learning_rate": 2.6793075020610057e-06,
+      "loss": 3.5633,
       "step": 5342
     },
     {
       "epoch": 2.99,
-      "learning_rate": 8.037922506183016e-07,
-      "loss": 0.2729,
+      "learning_rate": 2.47320692497939e-06,
+      "loss": 3.6679,
       "step": 5343
     },
     {
       "epoch": 3.0,
-      "learning_rate": 7.41962077493817e-07,
-      "loss": 0.4649,
+      "learning_rate": 2.267106347897774e-06,
+      "loss": 3.9039,
       "step": 5344
     },
     {
       "epoch": 3.0,
-      "learning_rate": 6.801319043693322e-07,
-      "loss": 0.4508,
+      "learning_rate": 2.0610057708161584e-06,
+      "loss": 3.8726,
       "step": 5345
     },
     {
       "epoch": 3.0,
-      "learning_rate": 6.183017312448474e-07,
-      "loss": 0.7735,
+      "learning_rate": 1.8549051937345426e-06,
+      "loss": 3.9625,
       "step": 5346
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.564715581203627e-07,
-      "loss": 0.5351,
+      "learning_rate": 1.6488046166529267e-06,
+      "loss": 3.9666,
       "step": 5347
     },
     {
       "epoch": 3.0,
-      "learning_rate": 4.94641384995878e-07,
-      "loss": 0.5192,
+      "learning_rate": 1.442704039571311e-06,
+      "loss": 3.9754,
       "step": 5348
     },
     {
       "epoch": 3.0,
-      "learning_rate": 4.328112118713932e-07,
-      "loss": 0.7357,
+      "learning_rate": 1.236603462489695e-06,
+      "loss": 3.648,
       "step": 5349
     },
     {
       "epoch": 3.0,
-      "learning_rate": 3.709810387469085e-07,
-      "loss": 0.5153,
+      "learning_rate": 1.0305028854080792e-06,
+      "loss": 3.5492,
       "step": 5350
     },
     {
       "epoch": 3.0,
-      "learning_rate": 3.091508656224237e-07,
-      "loss": 0.5915,
+      "learning_rate": 8.244023083264633e-07,
+      "loss": 3.1506,
       "step": 5351
     },
     {
       "epoch": 3.0,
-      "learning_rate": 2.47320692497939e-07,
-      "loss": 0.4899,
+      "learning_rate": 6.183017312448475e-07,
+      "loss": 3.1039,
       "step": 5352
     },
     {
       "epoch": 3.0,
       "step": 5352,
       "total_flos": 0.0,
-      "train_loss": 2.796300874691685,
-      "train_runtime": 30438.7012,
-      "train_samples_per_second": 2.813,
-      "train_steps_per_second": 0.176
+      "train_loss": 4.915239981070762,
+      "train_runtime": 30235.4571,
+      "train_samples_per_second": 2.832,
+      "train_steps_per_second": 0.177
     }
   ],
   "max_steps": 5352,