diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -9,32212 +9,32212 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.5802,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 3.992,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.589,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 3.8181,
       "step": 2
     },
     {
       "epoch": 0.0,
-      "learning_rate": 0.0,
-      "loss": 5.8413,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 3.7726,
       "step": 3
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2e-06,
-      "loss": 5.2778,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 3.3629,
       "step": 4
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4e-06,
-      "loss": 5.5308,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 3.6419,
       "step": 5
     },
     {
       "epoch": 0.0,
-      "learning_rate": 6e-06,
-      "loss": 5.3403,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 3.4384,
       "step": 6
     },
     {
       "epoch": 0.0,
-      "learning_rate": 8e-06,
-      "loss": 5.3527,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 3.5227,
       "step": 7
     },
     {
       "epoch": 0.0,
-      "learning_rate": 1e-05,
-      "loss": 5.168,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 3.5379,
       "step": 8
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.2e-05,
-      "loss": 5.4157,
+      "learning_rate": 1.8e-06,
+      "loss": 3.7281,
       "step": 9
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.4e-05,
-      "loss": 4.991,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 3.519,
       "step": 10
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.6e-05,
-      "loss": 4.8795,
+      "learning_rate": 2.2e-06,
+      "loss": 3.3431,
       "step": 11
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.8e-05,
-      "loss": 4.8489,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 3.3845,
       "step": 12
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2e-05,
-      "loss": 4.9404,
+      "learning_rate": 2.6e-06,
+      "loss": 3.4649,
       "step": 13
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2.2e-05,
-      "loss": 4.5799,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 3.5441,
       "step": 14
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2.4e-05,
-      "loss": 4.7039,
+      "learning_rate": 3e-06,
+      "loss": 3.0623,
       "step": 15
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2.6e-05,
-      "loss": 4.7245,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 3.7523,
       "step": 16
     },
     {
       "epoch": 0.01,
-      "learning_rate": 2.8e-05,
-      "loss": 4.7614,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 3.5193,
       "step": 17
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3e-05,
-      "loss": 4.5656,
+      "learning_rate": 3.6e-06,
+      "loss": 3.246,
       "step": 18
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.2e-05,
-      "loss": 4.5534,
+      "learning_rate": 3.8e-06,
+      "loss": 3.363,
       "step": 19
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 4.8297,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 3.8142,
       "step": 20
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.6e-05,
-      "loss": 4.5089,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 3.5738,
       "step": 21
     },
     {
       "epoch": 0.01,
-      "learning_rate": 3.8e-05,
-      "loss": 4.9362,
+      "learning_rate": 4.4e-06,
+      "loss": 3.8996,
       "step": 22
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4e-05,
-      "loss": 4.6566,
+      "learning_rate": 4.6e-06,
+      "loss": 3.4836,
       "step": 23
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.2000000000000004e-05,
-      "loss": 4.4095,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 3.4447,
       "step": 24
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.4e-05,
-      "loss": 4.5129,
+      "learning_rate": 5e-06,
+      "loss": 3.3839,
       "step": 25
     },
     {
       "epoch": 0.01,
-      "learning_rate": 4.6e-05,
-      "loss": 4.6861,
+      "learning_rate": 5.2e-06,
+      "loss": 3.627,
       "step": 26
     },
     {
       "epoch": 0.02,
-      "learning_rate": 4.8e-05,
-      "loss": 4.4625,
+      "learning_rate": 5.4e-06,
+      "loss": 3.3744,
       "step": 27
     },
     {
       "epoch": 0.02,
-      "learning_rate": 5e-05,
-      "loss": 4.1955,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 3.1118,
       "step": 28
     },
     {
       "epoch": 0.02,
-      "learning_rate": 5.2e-05,
-      "loss": 4.542,
+      "learning_rate": 5.8e-06,
+      "loss": 3.5995,
       "step": 29
     },
     {
       "epoch": 0.02,
-      "learning_rate": 5.4e-05,
-      "loss": 4.4487,
+      "learning_rate": 6e-06,
+      "loss": 3.064,
       "step": 30
     },
     {
       "epoch": 0.02,
-      "learning_rate": 5.6e-05,
-      "loss": 4.3978,
+      "learning_rate": 6.2e-06,
+      "loss": 3.2333,
       "step": 31
     },
     {
       "epoch": 0.02,
-      "learning_rate": 5.800000000000001e-05,
-      "loss": 4.7367,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 3.6427,
       "step": 32
     },
     {
       "epoch": 0.02,
-      "learning_rate": 6e-05,
-      "loss": 4.4123,
+      "learning_rate": 6.6e-06,
+      "loss": 3.3577,
       "step": 33
     },
     {
       "epoch": 0.02,
-      "learning_rate": 6.2e-05,
-      "loss": 4.2376,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 3.137,
       "step": 34
     },
     {
       "epoch": 0.02,
-      "learning_rate": 6.4e-05,
-      "loss": 4.5116,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 3.3634,
       "step": 35
     },
     {
       "epoch": 0.02,
-      "learning_rate": 6.6e-05,
-      "loss": 4.7934,
+      "learning_rate": 7.2e-06,
+      "loss": 3.655,
       "step": 36
     },
     {
       "epoch": 0.02,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 4.4381,
+      "learning_rate": 7.4e-06,
+      "loss": 3.3534,
       "step": 37
     },
     {
       "epoch": 0.02,
-      "learning_rate": 7.000000000000001e-05,
-      "loss": 4.5595,
+      "learning_rate": 7.6e-06,
+      "loss": 3.0795,
       "step": 38
     },
     {
       "epoch": 0.02,
-      "learning_rate": 7.2e-05,
-      "loss": 4.6284,
+      "learning_rate": 7.8e-06,
+      "loss": 3.4339,
       "step": 39
     },
     {
       "epoch": 0.02,
-      "learning_rate": 7.4e-05,
-      "loss": 4.6479,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 3.3455,
       "step": 40
     },
     {
       "epoch": 0.02,
-      "learning_rate": 7.6e-05,
-      "loss": 4.4379,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 3.1673,
       "step": 41
     },
     {
       "epoch": 0.02,
-      "learning_rate": 7.8e-05,
-      "loss": 4.3186,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 3.1841,
       "step": 42
     },
     {
       "epoch": 0.02,
-      "learning_rate": 8e-05,
-      "loss": 4.5889,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 3.2396,
       "step": 43
     },
     {
       "epoch": 0.02,
-      "learning_rate": 8.2e-05,
-      "loss": 4.3214,
+      "learning_rate": 8.8e-06,
+      "loss": 2.7627,
       "step": 44
     },
     {
       "epoch": 0.03,
-      "learning_rate": 8.400000000000001e-05,
-      "loss": 4.7199,
+      "learning_rate": 9e-06,
+      "loss": 3.1939,
       "step": 45
     },
     {
       "epoch": 0.03,
-      "learning_rate": 8.599999999999999e-05,
-      "loss": 4.7496,
+      "learning_rate": 9.2e-06,
+      "loss": 3.0895,
       "step": 46
     },
     {
       "epoch": 0.03,
-      "learning_rate": 8.8e-05,
-      "loss": 4.7364,
+      "learning_rate": 9.4e-06,
+      "loss": 3.2591,
       "step": 47
     },
     {
       "epoch": 0.03,
-      "learning_rate": 8.999999999999999e-05,
-      "loss": 4.8753,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.461,
       "step": 48
     },
     {
       "epoch": 0.03,
-      "learning_rate": 9.2e-05,
-      "loss": 4.949,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.3123,
       "step": 49
     },
     {
       "epoch": 0.03,
-      "learning_rate": 9.400000000000001e-05,
-      "loss": 5.027,
+      "learning_rate": 1e-05,
+      "loss": 2.4615,
       "step": 50
     },
     {
       "epoch": 0.03,
-      "learning_rate": 9.6e-05,
-      "loss": 4.4543,
+      "learning_rate": 1.02e-05,
+      "loss": 3.5886,
       "step": 51
     },
     {
       "epoch": 0.03,
-      "learning_rate": 9.800000000000001e-05,
-      "loss": 4.6299,
+      "learning_rate": 1.04e-05,
+      "loss": 3.6806,
       "step": 52
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.0001,
-      "loss": 4.5232,
+      "learning_rate": 1.06e-05,
+      "loss": 3.5256,
       "step": 53
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000102,
-      "loss": 4.3683,
+      "learning_rate": 1.08e-05,
+      "loss": 3.4397,
       "step": 54
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000104,
-      "loss": 4.5343,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 3.3059,
       "step": 55
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000106,
-      "loss": 4.5478,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 3.318,
       "step": 56
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000108,
-      "loss": 4.4856,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 3.1923,
       "step": 57
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.00011,
-      "loss": 4.4394,
+      "learning_rate": 1.16e-05,
+      "loss": 3.4746,
       "step": 58
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000112,
-      "loss": 4.4379,
+      "learning_rate": 1.18e-05,
+      "loss": 3.4423,
       "step": 59
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000114,
-      "loss": 4.4407,
+      "learning_rate": 1.2e-05,
+      "loss": 3.452,
       "step": 60
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.00011600000000000001,
-      "loss": 4.5579,
+      "learning_rate": 1.22e-05,
+      "loss": 3.5458,
       "step": 61
     },
     {
       "epoch": 0.03,
-      "learning_rate": 0.000118,
-      "loss": 4.5174,
+      "learning_rate": 1.24e-05,
+      "loss": 3.321,
       "step": 62
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.00012,
-      "loss": 4.8199,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 3.2869,
       "step": 63
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000122,
-      "loss": 4.2503,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 2.9675,
       "step": 64
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000124,
-      "loss": 4.4843,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 3.361,
       "step": 65
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000126,
-      "loss": 4.6071,
+      "learning_rate": 1.32e-05,
+      "loss": 3.5901,
       "step": 66
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000128,
-      "loss": 4.5811,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 3.4168,
       "step": 67
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.00013000000000000002,
-      "loss": 4.5564,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 3.2635,
       "step": 68
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000132,
-      "loss": 4.5395,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 3.1926,
       "step": 69
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000134,
-      "loss": 4.4299,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 3.332,
       "step": 70
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.00013600000000000003,
-      "loss": 4.5831,
+      "learning_rate": 1.42e-05,
+      "loss": 3.0905,
       "step": 71
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.00013800000000000002,
-      "loss": 4.2683,
+      "learning_rate": 1.44e-05,
+      "loss": 3.1981,
       "step": 72
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.00014000000000000001,
-      "loss": 4.6712,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 3.2791,
       "step": 73
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.00014199999999999998,
-      "loss": 4.3491,
+      "learning_rate": 1.48e-05,
+      "loss": 3.295,
       "step": 74
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000144,
-      "loss": 4.9766,
+      "learning_rate": 1.5e-05,
+      "loss": 3.8635,
       "step": 75
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000146,
-      "loss": 4.6173,
+      "learning_rate": 1.52e-05,
+      "loss": 3.3016,
       "step": 76
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000148,
-      "loss": 4.4911,
+      "learning_rate": 1.54e-05,
+      "loss": 2.9993,
       "step": 77
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.00015,
-      "loss": 4.4158,
+      "learning_rate": 1.56e-05,
+      "loss": 3.5501,
       "step": 78
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000152,
-      "loss": 4.5255,
+      "learning_rate": 1.58e-05,
+      "loss": 3.4552,
       "step": 79
     },
     {
       "epoch": 0.04,
-      "learning_rate": 0.000154,
-      "loss": 4.729,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 3.521,
       "step": 80
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000156,
-      "loss": 4.4216,
+      "learning_rate": 1.62e-05,
+      "loss": 3.2351,
       "step": 81
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000158,
-      "loss": 4.6069,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 3.2597,
       "step": 82
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00016,
-      "loss": 4.4213,
+      "learning_rate": 1.66e-05,
+      "loss": 3.0521,
       "step": 83
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000162,
-      "loss": 4.7356,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 3.4383,
       "step": 84
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000164,
-      "loss": 4.5921,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 3.0746,
       "step": 85
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00016600000000000002,
-      "loss": 4.4656,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 3.1327,
       "step": 86
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00016800000000000002,
-      "loss": 4.8462,
+      "learning_rate": 1.74e-05,
+      "loss": 3.0685,
       "step": 87
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00017,
-      "loss": 4.4326,
+      "learning_rate": 1.76e-05,
+      "loss": 2.7704,
       "step": 88
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00017199999999999998,
-      "loss": 4.7751,
+      "learning_rate": 1.78e-05,
+      "loss": 3.0052,
       "step": 89
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000174,
-      "loss": 4.3423,
+      "learning_rate": 1.8e-05,
+      "loss": 2.9003,
       "step": 90
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000176,
-      "loss": 4.7629,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 3.0922,
       "step": 91
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000178,
-      "loss": 4.4473,
+      "learning_rate": 1.84e-05,
+      "loss": 2.9866,
       "step": 92
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00017999999999999998,
-      "loss": 4.2477,
+      "learning_rate": 1.86e-05,
+      "loss": 2.5953,
       "step": 93
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000182,
-      "loss": 4.9831,
+      "learning_rate": 1.88e-05,
+      "loss": 2.6842,
       "step": 94
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000184,
-      "loss": 4.9647,
+      "learning_rate": 1.9e-05,
+      "loss": 2.9373,
       "step": 95
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.000186,
-      "loss": 4.8039,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 2.7103,
       "step": 96
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00018800000000000002,
-      "loss": 4.8747,
+      "learning_rate": 1.94e-05,
+      "loss": 2.5005,
       "step": 97
     },
     {
       "epoch": 0.05,
-      "learning_rate": 0.00019,
-      "loss": 4.6304,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 2.5222,
       "step": 98
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000192,
-      "loss": 4.9462,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 2.5108,
       "step": 99
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000194,
-      "loss": 4.9919,
+      "learning_rate": 2e-05,
+      "loss": 2.5287,
       "step": 100
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.00019600000000000002,
-      "loss": 5.0039,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 3.5283,
       "step": 101
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.00019800000000000002,
-      "loss": 4.7123,
+      "learning_rate": 2.04e-05,
+      "loss": 3.3997,
       "step": 102
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.0002,
-      "loss": 5.1955,
+      "learning_rate": 2.06e-05,
+      "loss": 3.5594,
       "step": 103
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000202,
-      "loss": 4.7571,
+      "learning_rate": 2.08e-05,
+      "loss": 3.2444,
       "step": 104
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000204,
-      "loss": 4.7477,
+      "learning_rate": 2.1e-05,
+      "loss": 3.2784,
       "step": 105
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000206,
-      "loss": 4.6276,
+      "learning_rate": 2.12e-05,
+      "loss": 3.0602,
       "step": 106
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000208,
-      "loss": 4.754,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 3.313,
       "step": 107
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.00021,
-      "loss": 4.7858,
+      "learning_rate": 2.16e-05,
+      "loss": 3.2319,
       "step": 108
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000212,
-      "loss": 4.7258,
+      "learning_rate": 2.18e-05,
+      "loss": 3.2574,
       "step": 109
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000214,
-      "loss": 4.7978,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 3.5165,
       "step": 110
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000216,
-      "loss": 5.0629,
+      "learning_rate": 2.22e-05,
+      "loss": 3.2104,
       "step": 111
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000218,
-      "loss": 4.547,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 3.1124,
       "step": 112
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.00022,
-      "loss": 4.7776,
+      "learning_rate": 2.26e-05,
+      "loss": 3.0979,
       "step": 113
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000222,
-      "loss": 4.5737,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 2.9967,
       "step": 114
     },
     {
       "epoch": 0.06,
-      "learning_rate": 0.000224,
-      "loss": 4.4555,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 3.2227,
       "step": 115
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00022600000000000002,
-      "loss": 4.569,
+      "learning_rate": 2.32e-05,
+      "loss": 2.8693,
       "step": 116
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000228,
-      "loss": 4.9807,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 3.1598,
       "step": 117
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00023,
-      "loss": 5.4174,
+      "learning_rate": 2.36e-05,
+      "loss": 3.1608,
       "step": 118
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00023200000000000003,
-      "loss": 4.7848,
+      "learning_rate": 2.38e-05,
+      "loss": 3.1767,
       "step": 119
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00023400000000000002,
-      "loss": 5.2558,
+      "learning_rate": 2.4e-05,
+      "loss": 3.6019,
       "step": 120
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000236,
-      "loss": 5.0696,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 3.268,
       "step": 121
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00023799999999999998,
-      "loss": 4.7726,
+      "learning_rate": 2.44e-05,
+      "loss": 3.496,
       "step": 122
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00024,
-      "loss": 4.7591,
+      "learning_rate": 2.46e-05,
+      "loss": 3.1792,
       "step": 123
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000242,
-      "loss": 4.9439,
+      "learning_rate": 2.48e-05,
+      "loss": 3.3206,
       "step": 124
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000244,
-      "loss": 4.5524,
+      "learning_rate": 2.5e-05,
+      "loss": 2.9943,
       "step": 125
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000246,
-      "loss": 4.9518,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 2.829,
       "step": 126
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000248,
-      "loss": 4.674,
+      "learning_rate": 2.54e-05,
+      "loss": 3.3311,
       "step": 127
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00025,
-      "loss": 4.8241,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 3.1464,
       "step": 128
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000252,
-      "loss": 4.8413,
+      "learning_rate": 2.58e-05,
+      "loss": 3.0236,
       "step": 129
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000254,
-      "loss": 5.1165,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 3.1405,
       "step": 130
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.000256,
-      "loss": 5.6705,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 3.2847,
       "step": 131
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00025800000000000004,
-      "loss": 4.6968,
+      "learning_rate": 2.64e-05,
+      "loss": 3.1525,
       "step": 132
     },
     {
       "epoch": 0.07,
-      "learning_rate": 0.00026000000000000003,
-      "loss": 5.4421,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 3.1305,
       "step": 133
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000262,
-      "loss": 5.273,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 3.1445,
       "step": 134
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000264,
-      "loss": 4.6382,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 3.0277,
       "step": 135
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000266,
-      "loss": 5.379,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 3.0804,
       "step": 136
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000268,
-      "loss": 5.0821,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 3.0436,
       "step": 137
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00027,
-      "loss": 5.8035,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 3.0259,
       "step": 138
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00027200000000000005,
-      "loss": 5.5749,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 3.3046,
       "step": 139
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00027400000000000005,
-      "loss": 5.1669,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 3.224,
       "step": 140
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00027600000000000004,
-      "loss": 6.1823,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 3.1431,
       "step": 141
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00027800000000000004,
-      "loss": 5.2155,
+      "learning_rate": 2.84e-05,
+      "loss": 2.9416,
       "step": 142
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00028000000000000003,
-      "loss": 4.8095,
+      "learning_rate": 2.86e-05,
+      "loss": 2.5949,
       "step": 143
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00028199999999999997,
-      "loss": 6.1255,
+      "learning_rate": 2.88e-05,
+      "loss": 2.6013,
       "step": 144
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00028399999999999996,
-      "loss": 5.2863,
+      "learning_rate": 2.9e-05,
+      "loss": 2.7122,
       "step": 145
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00028599999999999996,
-      "loss": 5.1855,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 2.3622,
       "step": 146
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000288,
-      "loss": 5.2812,
+      "learning_rate": 2.94e-05,
+      "loss": 2.2801,
       "step": 147
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.00029,
-      "loss": 5.1448,
+      "learning_rate": 2.96e-05,
+      "loss": 2.0663,
       "step": 148
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000292,
-      "loss": 5.3333,
+      "learning_rate": 2.98e-05,
+      "loss": 2.7475,
       "step": 149
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000294,
-      "loss": 4.6151,
+      "learning_rate": 3e-05,
+      "loss": 2.5938,
       "step": 150
     },
     {
       "epoch": 0.08,
-      "learning_rate": 0.000296,
-      "loss": 4.945,
+      "learning_rate": 3.02e-05,
+      "loss": 3.1866,
       "step": 151
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000298,
-      "loss": 5.4402,
+      "learning_rate": 3.04e-05,
+      "loss": 3.4392,
       "step": 152
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.0003,
-      "loss": 4.599,
+      "learning_rate": 3.06e-05,
+      "loss": 3.1123,
       "step": 153
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000302,
-      "loss": 6.4004,
+      "learning_rate": 3.08e-05,
+      "loss": 3.5643,
       "step": 154
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000304,
-      "loss": 4.9015,
+      "learning_rate": 3.1e-05,
+      "loss": 3.0386,
       "step": 155
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000306,
-      "loss": 5.2098,
+      "learning_rate": 3.12e-05,
+      "loss": 3.4456,
       "step": 156
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000308,
-      "loss": 5.079,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 3.2886,
       "step": 157
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.00031,
-      "loss": 4.8843,
+      "learning_rate": 3.16e-05,
+      "loss": 3.3199,
       "step": 158
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000312,
-      "loss": 5.4663,
+      "learning_rate": 3.18e-05,
+      "loss": 3.2453,
       "step": 159
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000314,
-      "loss": 6.6276,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 3.4552,
       "step": 160
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000316,
-      "loss": 5.2692,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 3.2108,
       "step": 161
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.00031800000000000003,
-      "loss": 5.4687,
+      "learning_rate": 3.24e-05,
+      "loss": 3.2024,
       "step": 162
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.00032,
-      "loss": 5.0059,
+      "learning_rate": 3.26e-05,
+      "loss": 3.1625,
       "step": 163
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000322,
-      "loss": 5.4655,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 3.2781,
       "step": 164
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000324,
-      "loss": 4.9151,
+      "learning_rate": 3.3e-05,
+      "loss": 3.1551,
       "step": 165
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000326,
-      "loss": 5.3121,
+      "learning_rate": 3.32e-05,
+      "loss": 3.2363,
       "step": 166
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.000328,
-      "loss": 4.9021,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 3.1569,
       "step": 167
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.00033,
-      "loss": 4.7903,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 3.1388,
       "step": 168
     },
     {
       "epoch": 0.09,
-      "learning_rate": 0.00033200000000000005,
-      "loss": 4.7505,
+      "learning_rate": 3.38e-05,
+      "loss": 2.9376,
       "step": 169
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00033400000000000004,
-      "loss": 4.8679,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 2.9921,
       "step": 170
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00033600000000000004,
-      "loss": 5.3312,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 3.1882,
       "step": 171
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00033800000000000003,
-      "loss": 7.5863,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 3.2787,
       "step": 172
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00034,
-      "loss": 4.9188,
+      "learning_rate": 3.46e-05,
+      "loss": 3.3056,
       "step": 173
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000342,
-      "loss": 5.1428,
+      "learning_rate": 3.48e-05,
+      "loss": 2.8736,
       "step": 174
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00034399999999999996,
-      "loss": 6.7474,
+      "learning_rate": 3.5e-05,
+      "loss": 3.157,
       "step": 175
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000346,
-      "loss": 5.2806,
+      "learning_rate": 3.52e-05,
+      "loss": 3.0108,
       "step": 176
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000348,
-      "loss": 4.8877,
+      "learning_rate": 3.54e-05,
+      "loss": 2.8398,
       "step": 177
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00035,
-      "loss": 4.8529,
+      "learning_rate": 3.56e-05,
+      "loss": 3.0094,
       "step": 178
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000352,
-      "loss": 4.8546,
+      "learning_rate": 3.58e-05,
+      "loss": 3.0663,
       "step": 179
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000354,
-      "loss": 5.6208,
+      "learning_rate": 3.6e-05,
+      "loss": 3.265,
       "step": 180
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000356,
-      "loss": 5.7513,
+      "learning_rate": 3.62e-05,
+      "loss": 3.0226,
       "step": 181
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000358,
-      "loss": 4.8119,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 2.8337,
       "step": 182
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.00035999999999999997,
-      "loss": 5.3898,
+      "learning_rate": 3.66e-05,
+      "loss": 3.1334,
       "step": 183
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000362,
-      "loss": 4.9066,
+      "learning_rate": 3.68e-05,
+      "loss": 3.0193,
       "step": 184
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000364,
-      "loss": 5.1167,
+      "learning_rate": 3.7e-05,
+      "loss": 2.6703,
       "step": 185
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000366,
-      "loss": 5.0465,
+      "learning_rate": 3.72e-05,
+      "loss": 2.5416,
       "step": 186
     },
     {
       "epoch": 0.1,
-      "learning_rate": 0.000368,
-      "loss": 4.8059,
+      "learning_rate": 3.74e-05,
+      "loss": 2.825,
       "step": 187
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00037,
-      "loss": 5.1748,
+      "learning_rate": 3.76e-05,
+      "loss": 2.9914,
       "step": 188
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000372,
-      "loss": 5.2375,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 2.832,
       "step": 189
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000374,
-      "loss": 5.6544,
+      "learning_rate": 3.8e-05,
+      "loss": 2.7943,
       "step": 190
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00037600000000000003,
-      "loss": 5.2274,
+      "learning_rate": 3.82e-05,
+      "loss": 2.751,
       "step": 191
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000378,
-      "loss": 5.153,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 2.724,
       "step": 192
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00038,
-      "loss": 6.7931,
+      "learning_rate": 3.86e-05,
+      "loss": 2.7643,
       "step": 193
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000382,
-      "loss": 5.2208,
+      "learning_rate": 3.88e-05,
+      "loss": 2.7803,
       "step": 194
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000384,
-      "loss": 5.1046,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 2.2997,
       "step": 195
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000386,
-      "loss": 6.0011,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 2.3026,
       "step": 196
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000388,
-      "loss": 6.4511,
+      "learning_rate": 3.94e-05,
+      "loss": 2.4077,
       "step": 197
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00039000000000000005,
-      "loss": 4.8761,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 1.9257,
       "step": 198
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00039200000000000004,
-      "loss": 5.1652,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 1.7536,
       "step": 199
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00039400000000000004,
-      "loss": 6.4481,
+      "learning_rate": 4e-05,
+      "loss": 2.2137,
       "step": 200
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.00039600000000000003,
-      "loss": 5.4104,
+      "learning_rate": 4.02e-05,
+      "loss": 3.4474,
       "step": 201
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000398,
-      "loss": 5.1596,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 3.3898,
       "step": 202
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.0004,
-      "loss": 5.3031,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 3.1967,
       "step": 203
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000402,
-      "loss": 5.6399,
+      "learning_rate": 4.08e-05,
+      "loss": 2.9007,
       "step": 204
     },
     {
       "epoch": 0.11,
-      "learning_rate": 0.000404,
-      "loss": 5.4172,
+      "learning_rate": 4.1e-05,
+      "loss": 2.7127,
       "step": 205
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00040600000000000006,
-      "loss": 6.016,
+      "learning_rate": 4.12e-05,
+      "loss": 3.167,
       "step": 206
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000408,
-      "loss": 5.3531,
+      "learning_rate": 4.14e-05,
+      "loss": 3.1481,
       "step": 207
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00041,
-      "loss": 6.4945,
+      "learning_rate": 4.16e-05,
+      "loss": 3.202,
       "step": 208
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000412,
-      "loss": 5.2357,
+      "learning_rate": 4.18e-05,
+      "loss": 2.9316,
       "step": 209
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000414,
-      "loss": 5.5978,
+      "learning_rate": 4.2e-05,
+      "loss": 2.9757,
       "step": 210
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000416,
-      "loss": 5.059,
+      "learning_rate": 4.22e-05,
+      "loss": 2.97,
       "step": 211
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00041799999999999997,
-      "loss": 5.702,
+      "learning_rate": 4.24e-05,
+      "loss": 2.9356,
       "step": 212
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00042,
-      "loss": 6.4733,
+      "learning_rate": 4.26e-05,
+      "loss": 3.218,
       "step": 213
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000422,
-      "loss": 4.8425,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 2.9109,
       "step": 214
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000424,
-      "loss": 5.8915,
+      "learning_rate": 4.3e-05,
+      "loss": 3.4791,
       "step": 215
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000426,
-      "loss": 4.8864,
+      "learning_rate": 4.32e-05,
+      "loss": 2.7313,
       "step": 216
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000428,
-      "loss": 5.4242,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 3.0691,
       "step": 217
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00043,
-      "loss": 4.9671,
+      "learning_rate": 4.36e-05,
+      "loss": 2.8531,
       "step": 218
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000432,
-      "loss": 5.0474,
+      "learning_rate": 4.38e-05,
+      "loss": 2.8569,
       "step": 219
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00043400000000000003,
-      "loss": 6.2105,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 3.2709,
       "step": 220
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000436,
-      "loss": 5.2854,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 2.9622,
       "step": 221
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.000438,
-      "loss": 5.8823,
+      "learning_rate": 4.44e-05,
+      "loss": 3.2497,
       "step": 222
     },
     {
       "epoch": 0.12,
-      "learning_rate": 0.00044,
-      "loss": 5.0082,
+      "learning_rate": 4.46e-05,
+      "loss": 3.0882,
       "step": 223
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000442,
-      "loss": 5.3606,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 2.8808,
       "step": 224
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000444,
-      "loss": 5.3525,
+      "learning_rate": 4.5e-05,
+      "loss": 2.8913,
       "step": 225
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000446,
-      "loss": 5.6586,
+      "learning_rate": 4.52e-05,
+      "loss": 3.019,
       "step": 226
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000448,
-      "loss": 5.4038,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 2.877,
       "step": 227
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00045000000000000004,
-      "loss": 5.3157,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 2.7611,
       "step": 228
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00045200000000000004,
-      "loss": 5.154,
+      "learning_rate": 4.58e-05,
+      "loss": 2.8008,
       "step": 229
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00045400000000000003,
-      "loss": 5.1189,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 2.7197,
       "step": 230
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000456,
-      "loss": 4.9935,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 2.7656,
       "step": 231
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000458,
-      "loss": 5.7535,
+      "learning_rate": 4.64e-05,
+      "loss": 2.9431,
       "step": 232
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00046,
-      "loss": 5.4902,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 2.8997,
       "step": 233
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000462,
-      "loss": 5.1133,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 2.7514,
       "step": 234
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00046400000000000006,
-      "loss": 4.858,
+      "learning_rate": 4.7e-05,
+      "loss": 2.6112,
       "step": 235
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00046600000000000005,
-      "loss": 5.6959,
+      "learning_rate": 4.72e-05,
+      "loss": 3.0416,
       "step": 236
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00046800000000000005,
-      "loss": 6.2172,
+      "learning_rate": 4.74e-05,
+      "loss": 2.8515,
       "step": 237
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.00047,
-      "loss": 6.562,
+      "learning_rate": 4.76e-05,
+      "loss": 2.932,
       "step": 238
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000472,
-      "loss": 5.935,
+      "learning_rate": 4.78e-05,
+      "loss": 2.6979,
       "step": 239
     },
     {
       "epoch": 0.13,
-      "learning_rate": 0.000474,
-      "loss": 6.6483,
+      "learning_rate": 4.8e-05,
+      "loss": 2.7352,
       "step": 240
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00047599999999999997,
-      "loss": 5.2033,
+      "learning_rate": 4.82e-05,
+      "loss": 2.6192,
       "step": 241
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00047799999999999996,
-      "loss": 5.6109,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 2.4924,
       "step": 242
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00048,
-      "loss": 8.1627,
+      "learning_rate": 4.86e-05,
+      "loss": 2.2598,
       "step": 243
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000482,
-      "loss": 5.6307,
+      "learning_rate": 4.88e-05,
+      "loss": 2.3626,
       "step": 244
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000484,
-      "loss": 5.3183,
+      "learning_rate": 4.9e-05,
+      "loss": 2.1997,
       "step": 245
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000486,
-      "loss": 5.3182,
+      "learning_rate": 4.92e-05,
+      "loss": 1.9489,
       "step": 246
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000488,
-      "loss": 4.9353,
+      "learning_rate": 4.94e-05,
+      "loss": 1.9644,
       "step": 247
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00049,
-      "loss": 5.4198,
+      "learning_rate": 4.96e-05,
+      "loss": 1.8474,
       "step": 248
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000492,
-      "loss": 5.2011,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 1.8165,
       "step": 249
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000494,
-      "loss": 6.221,
+      "learning_rate": 5e-05,
+      "loss": 1.8391,
       "step": 250
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000496,
-      "loss": 5.7471,
+      "learning_rate": 5.02e-05,
+      "loss": 3.2713,
       "step": 251
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000498,
-      "loss": 6.7848,
+      "learning_rate": 5.0400000000000005e-05,
+      "loss": 3.2747,
       "step": 252
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0005,
-      "loss": 6.1404,
+      "learning_rate": 5.0600000000000003e-05,
+      "loss": 3.1071,
       "step": 253
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.0005020000000000001,
-      "loss": 5.6192,
+      "learning_rate": 5.08e-05,
+      "loss": 3.0269,
       "step": 254
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000504,
-      "loss": 5.2452,
+      "learning_rate": 5.1000000000000006e-05,
+      "loss": 2.7349,
       "step": 255
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000506,
-      "loss": 5.4712,
+      "learning_rate": 5.1200000000000004e-05,
+      "loss": 2.8675,
       "step": 256
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.000508,
-      "loss": 5.3885,
+      "learning_rate": 5.14e-05,
+      "loss": 2.9802,
       "step": 257
     },
     {
       "epoch": 0.14,
-      "learning_rate": 0.00051,
-      "loss": 6.309,
+      "learning_rate": 5.16e-05,
+      "loss": 3.4412,
       "step": 258
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000512,
-      "loss": 5.5993,
+      "learning_rate": 5.1800000000000005e-05,
+      "loss": 3.1706,
       "step": 259
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000514,
-      "loss": 5.0639,
+      "learning_rate": 5.2000000000000004e-05,
+      "loss": 2.6514,
       "step": 260
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0005160000000000001,
-      "loss": 6.3613,
+      "learning_rate": 5.22e-05,
+      "loss": 3.4016,
       "step": 261
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000518,
-      "loss": 5.6503,
+      "learning_rate": 5.2400000000000007e-05,
+      "loss": 3.3476,
       "step": 262
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0005200000000000001,
-      "loss": 6.2474,
+      "learning_rate": 5.2600000000000005e-05,
+      "loss": 3.2025,
       "step": 263
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000522,
-      "loss": 5.4449,
+      "learning_rate": 5.28e-05,
+      "loss": 3.3042,
       "step": 264
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000524,
-      "loss": 5.4925,
+      "learning_rate": 5.300000000000001e-05,
+      "loss": 2.8689,
       "step": 265
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000526,
-      "loss": 4.9186,
+      "learning_rate": 5.3200000000000006e-05,
+      "loss": 2.6642,
       "step": 266
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000528,
-      "loss": 5.8049,
+      "learning_rate": 5.3400000000000004e-05,
+      "loss": 3.0528,
       "step": 267
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0005300000000000001,
-      "loss": 5.524,
+      "learning_rate": 5.360000000000001e-05,
+      "loss": 3.4616,
       "step": 268
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000532,
-      "loss": 5.6952,
+      "learning_rate": 5.380000000000001e-05,
+      "loss": 3.0984,
       "step": 269
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0005340000000000001,
-      "loss": 5.0467,
+      "learning_rate": 5.4000000000000005e-05,
+      "loss": 2.816,
       "step": 270
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000536,
-      "loss": 5.2591,
+      "learning_rate": 5.420000000000001e-05,
+      "loss": 2.8335,
       "step": 271
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0005380000000000001,
-      "loss": 5.2361,
+      "learning_rate": 5.440000000000001e-05,
+      "loss": 2.6266,
       "step": 272
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.00054,
-      "loss": 5.6284,
+      "learning_rate": 5.4600000000000006e-05,
+      "loss": 2.7286,
       "step": 273
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0005420000000000001,
-      "loss": 7.7446,
+      "learning_rate": 5.4800000000000004e-05,
+      "loss": 2.9433,
       "step": 274
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.0005440000000000001,
-      "loss": 5.3482,
+      "learning_rate": 5.500000000000001e-05,
+      "loss": 2.6495,
       "step": 275
     },
     {
       "epoch": 0.15,
-      "learning_rate": 0.000546,
-      "loss": 6.5872,
+      "learning_rate": 5.520000000000001e-05,
+      "loss": 3.2656,
       "step": 276
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005480000000000001,
-      "loss": 6.1554,
+      "learning_rate": 5.5400000000000005e-05,
+      "loss": 2.8249,
       "step": 277
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00055,
-      "loss": 5.5187,
+      "learning_rate": 5.560000000000001e-05,
+      "loss": 3.0853,
       "step": 278
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005520000000000001,
-      "loss": 5.2059,
+      "learning_rate": 5.580000000000001e-05,
+      "loss": 2.7186,
       "step": 279
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000554,
-      "loss": 5.1271,
+      "learning_rate": 5.6000000000000006e-05,
+      "loss": 2.5272,
       "step": 280
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005560000000000001,
-      "loss": 6.0853,
+      "learning_rate": 5.620000000000001e-05,
+      "loss": 3.035,
       "step": 281
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000558,
-      "loss": 5.5123,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 2.7449,
       "step": 282
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005600000000000001,
-      "loss": 5.9598,
+      "learning_rate": 5.66e-05,
+      "loss": 3.0029,
       "step": 283
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005620000000000001,
-      "loss": 4.9705,
+      "learning_rate": 5.68e-05,
+      "loss": 2.4737,
       "step": 284
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005639999999999999,
-      "loss": 5.5854,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 2.5978,
       "step": 285
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000566,
-      "loss": 5.4473,
+      "learning_rate": 5.72e-05,
+      "loss": 2.3851,
       "step": 286
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005679999999999999,
-      "loss": 5.4138,
+      "learning_rate": 5.74e-05,
+      "loss": 2.6884,
       "step": 287
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00057,
-      "loss": 5.807,
+      "learning_rate": 5.76e-05,
+      "loss": 2.6648,
       "step": 288
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005719999999999999,
-      "loss": 5.2322,
+      "learning_rate": 5.7799999999999995e-05,
+      "loss": 2.4124,
       "step": 289
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000574,
-      "loss": 6.1189,
+      "learning_rate": 5.8e-05,
+      "loss": 2.5843,
       "step": 290
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000576,
-      "loss": 5.35,
+      "learning_rate": 5.82e-05,
+      "loss": 2.4229,
       "step": 291
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.000578,
-      "loss": 6.651,
+      "learning_rate": 5.8399999999999997e-05,
+      "loss": 2.7438,
       "step": 292
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.00058,
-      "loss": 5.4333,
+      "learning_rate": 5.86e-05,
+      "loss": 2.3469,
       "step": 293
     },
     {
       "epoch": 0.16,
-      "learning_rate": 0.0005819999999999999,
-      "loss": 5.5214,
+      "learning_rate": 5.88e-05,
+      "loss": 2.0262,
       "step": 294
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000584,
-      "loss": 5.2903,
+      "learning_rate": 5.9e-05,
+      "loss": 1.905,
       "step": 295
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.0005859999999999999,
-      "loss": 11.0393,
+      "learning_rate": 5.92e-05,
+      "loss": 2.1471,
       "step": 296
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000588,
-      "loss": 6.1775,
+      "learning_rate": 5.94e-05,
+      "loss": 2.2054,
       "step": 297
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00059,
-      "loss": 5.7617,
+      "learning_rate": 5.96e-05,
+      "loss": 1.9071,
       "step": 298
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000592,
-      "loss": 6.5483,
+      "learning_rate": 5.9800000000000003e-05,
+      "loss": 1.6484,
       "step": 299
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000594,
-      "loss": 5.1852,
+      "learning_rate": 6e-05,
+      "loss": 1.9474,
       "step": 300
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000596,
-      "loss": 6.1581,
+      "learning_rate": 6.02e-05,
+      "loss": 3.4948,
       "step": 301
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000598,
-      "loss": 5.3975,
+      "learning_rate": 6.04e-05,
+      "loss": 3.0148,
       "step": 302
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.0006,
-      "loss": 5.429,
+      "learning_rate": 6.06e-05,
+      "loss": 3.0494,
       "step": 303
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000602,
-      "loss": 5.1957,
+      "learning_rate": 6.08e-05,
+      "loss": 2.7212,
       "step": 304
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000604,
-      "loss": 6.3192,
+      "learning_rate": 6.1e-05,
+      "loss": 3.2609,
       "step": 305
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000606,
-      "loss": 5.3676,
+      "learning_rate": 6.12e-05,
+      "loss": 2.7699,
       "step": 306
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000608,
-      "loss": 5.339,
+      "learning_rate": 6.14e-05,
+      "loss": 2.8865,
       "step": 307
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.00061,
-      "loss": 6.7886,
+      "learning_rate": 6.16e-05,
+      "loss": 3.7117,
       "step": 308
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000612,
-      "loss": 5.5512,
+      "learning_rate": 6.18e-05,
+      "loss": 2.8956,
       "step": 309
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000614,
-      "loss": 5.9811,
+      "learning_rate": 6.2e-05,
+      "loss": 3.1182,
       "step": 310
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.000616,
-      "loss": 5.3142,
+      "learning_rate": 6.220000000000001e-05,
+      "loss": 2.9492,
       "step": 311
     },
     {
       "epoch": 0.17,
-      "learning_rate": 0.0006180000000000001,
-      "loss": 5.8051,
+      "learning_rate": 6.24e-05,
+      "loss": 2.7033,
       "step": 312
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00062,
-      "loss": 5.3857,
+      "learning_rate": 6.26e-05,
+      "loss": 2.6778,
       "step": 313
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000622,
-      "loss": 6.0494,
+      "learning_rate": 6.280000000000001e-05,
+      "loss": 2.8236,
       "step": 314
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000624,
-      "loss": 6.1703,
+      "learning_rate": 6.3e-05,
+      "loss": 2.9941,
       "step": 315
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000626,
-      "loss": 5.26,
+      "learning_rate": 6.32e-05,
+      "loss": 2.5915,
       "step": 316
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000628,
-      "loss": 5.3323,
+      "learning_rate": 6.340000000000001e-05,
+      "loss": 2.7083,
       "step": 317
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00063,
-      "loss": 5.1775,
+      "learning_rate": 6.36e-05,
+      "loss": 2.728,
       "step": 318
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000632,
-      "loss": 6.4257,
+      "learning_rate": 6.38e-05,
+      "loss": 2.8243,
       "step": 319
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000634,
-      "loss": 6.4121,
+      "learning_rate": 6.400000000000001e-05,
+      "loss": 2.8371,
       "step": 320
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0006360000000000001,
-      "loss": 6.3211,
+      "learning_rate": 6.42e-05,
+      "loss": 3.1351,
       "step": 321
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000638,
-      "loss": 6.3609,
+      "learning_rate": 6.440000000000001e-05,
+      "loss": 2.5313,
       "step": 322
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.00064,
-      "loss": 5.843,
+      "learning_rate": 6.460000000000001e-05,
+      "loss": 2.8606,
       "step": 323
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000642,
-      "loss": 6.328,
+      "learning_rate": 6.48e-05,
+      "loss": 2.6507,
       "step": 324
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000644,
-      "loss": 5.4115,
+      "learning_rate": 6.500000000000001e-05,
+      "loss": 3.1682,
       "step": 325
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000646,
-      "loss": 5.5552,
+      "learning_rate": 6.52e-05,
+      "loss": 2.5163,
       "step": 326
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000648,
-      "loss": 5.9054,
+      "learning_rate": 6.54e-05,
+      "loss": 2.8675,
       "step": 327
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0006500000000000001,
-      "loss": 5.8717,
+      "learning_rate": 6.560000000000001e-05,
+      "loss": 2.8676,
       "step": 328
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.000652,
-      "loss": 5.2736,
+      "learning_rate": 6.58e-05,
+      "loss": 2.6702,
       "step": 329
     },
     {
       "epoch": 0.18,
-      "learning_rate": 0.0006540000000000001,
-      "loss": 5.742,
+      "learning_rate": 6.6e-05,
+      "loss": 2.7189,
       "step": 330
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000656,
-      "loss": 5.6059,
+      "learning_rate": 6.620000000000001e-05,
+      "loss": 2.4294,
       "step": 331
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006580000000000001,
-      "loss": 5.8953,
+      "learning_rate": 6.64e-05,
+      "loss": 2.4591,
       "step": 332
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.00066,
-      "loss": 6.9948,
+      "learning_rate": 6.66e-05,
+      "loss": 3.0921,
       "step": 333
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000662,
-      "loss": 5.9059,
+      "learning_rate": 6.680000000000001e-05,
+      "loss": 2.6551,
       "step": 334
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006640000000000001,
-      "loss": 6.7979,
+      "learning_rate": 6.7e-05,
+      "loss": 2.926,
       "step": 335
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000666,
-      "loss": 6.4717,
+      "learning_rate": 6.720000000000001e-05,
+      "loss": 2.5791,
       "step": 336
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006680000000000001,
-      "loss": 6.2319,
+      "learning_rate": 6.740000000000001e-05,
+      "loss": 2.8306,
       "step": 337
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.00067,
-      "loss": 5.3135,
+      "learning_rate": 6.76e-05,
+      "loss": 2.533,
       "step": 338
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006720000000000001,
-      "loss": 5.2328,
+      "learning_rate": 6.780000000000001e-05,
+      "loss": 2.3462,
       "step": 339
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000674,
-      "loss": 5.8928,
+      "learning_rate": 6.800000000000001e-05,
+      "loss": 2.4151,
       "step": 340
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006760000000000001,
-      "loss": 5.3474,
+      "learning_rate": 6.82e-05,
+      "loss": 2.4882,
       "step": 341
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006780000000000001,
-      "loss": 5.6893,
+      "learning_rate": 6.840000000000001e-05,
+      "loss": 2.3287,
       "step": 342
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.00068,
-      "loss": 5.5719,
+      "learning_rate": 6.860000000000001e-05,
+      "loss": 2.2812,
       "step": 343
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006820000000000001,
-      "loss": 5.1262,
+      "learning_rate": 6.879999999999999e-05,
+      "loss": 1.6865,
       "step": 344
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.000684,
-      "loss": 6.4271,
+      "learning_rate": 6.9e-05,
+      "loss": 2.3276,
       "step": 345
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006860000000000001,
-      "loss": 5.284,
+      "learning_rate": 6.92e-05,
+      "loss": 1.699,
       "step": 346
     },
     {
       "epoch": 0.19,
-      "learning_rate": 0.0006879999999999999,
-      "loss": 5.8936,
+      "learning_rate": 6.939999999999999e-05,
+      "loss": 1.8784,
       "step": 347
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00069,
-      "loss": 5.7215,
+      "learning_rate": 6.96e-05,
+      "loss": 2.0265,
       "step": 348
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000692,
-      "loss": 7.5502,
+      "learning_rate": 6.98e-05,
+      "loss": 1.6271,
       "step": 349
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000694,
-      "loss": 5.2064,
+      "learning_rate": 7e-05,
+      "loss": 1.8088,
       "step": 350
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000696,
-      "loss": 6.4653,
+      "learning_rate": 7.02e-05,
+      "loss": 2.8521,
       "step": 351
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.0006979999999999999,
-      "loss": 5.8233,
+      "learning_rate": 7.04e-05,
+      "loss": 2.9789,
       "step": 352
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.0007,
-      "loss": 5.6414,
+      "learning_rate": 7.06e-05,
+      "loss": 2.652,
       "step": 353
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.0007019999999999999,
-      "loss": 5.335,
+      "learning_rate": 7.08e-05,
+      "loss": 2.4719,
       "step": 354
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000704,
-      "loss": 5.2106,
+      "learning_rate": 7.1e-05,
+      "loss": 2.3231,
       "step": 355
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.0007059999999999999,
-      "loss": 5.675,
+      "learning_rate": 7.12e-05,
+      "loss": 2.9489,
       "step": 356
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000708,
-      "loss": 5.7361,
+      "learning_rate": 7.14e-05,
+      "loss": 3.0312,
       "step": 357
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.00071,
-      "loss": 6.1298,
+      "learning_rate": 7.16e-05,
+      "loss": 3.1116,
       "step": 358
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000712,
-      "loss": 6.9985,
+      "learning_rate": 7.18e-05,
+      "loss": 2.9337,
       "step": 359
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000714,
-      "loss": 5.4424,
+      "learning_rate": 7.2e-05,
+      "loss": 2.3238,
       "step": 360
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000716,
-      "loss": 5.3975,
+      "learning_rate": 7.22e-05,
+      "loss": 2.4948,
       "step": 361
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000718,
-      "loss": 5.057,
+      "learning_rate": 7.24e-05,
+      "loss": 2.3209,
       "step": 362
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.0007199999999999999,
-      "loss": 5.4258,
+      "learning_rate": 7.26e-05,
+      "loss": 2.5843,
       "step": 363
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000722,
-      "loss": 6.0874,
+      "learning_rate": 7.280000000000001e-05,
+      "loss": 2.754,
       "step": 364
     },
     {
       "epoch": 0.2,
-      "learning_rate": 0.000724,
-      "loss": 8.2428,
+      "learning_rate": 7.3e-05,
+      "loss": 3.0635,
       "step": 365
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000726,
-      "loss": 7.0248,
+      "learning_rate": 7.32e-05,
+      "loss": 2.6711,
       "step": 366
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000728,
-      "loss": 5.8983,
+      "learning_rate": 7.340000000000001e-05,
+      "loss": 3.0105,
       "step": 367
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00073,
-      "loss": 5.5039,
+      "learning_rate": 7.36e-05,
+      "loss": 2.6296,
       "step": 368
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000732,
-      "loss": 6.0179,
+      "learning_rate": 7.38e-05,
+      "loss": 2.6187,
       "step": 369
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000734,
-      "loss": 5.5334,
+      "learning_rate": 7.4e-05,
+      "loss": 2.447,
       "step": 370
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000736,
-      "loss": 5.3457,
+      "learning_rate": 7.42e-05,
+      "loss": 2.5649,
       "step": 371
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000738,
-      "loss": 6.2481,
+      "learning_rate": 7.44e-05,
+      "loss": 2.8467,
       "step": 372
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00074,
-      "loss": 5.4737,
+      "learning_rate": 7.46e-05,
+      "loss": 2.1975,
       "step": 373
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000742,
-      "loss": 6.2183,
+      "learning_rate": 7.48e-05,
+      "loss": 2.51,
       "step": 374
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000744,
-      "loss": 5.1563,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 2.2381,
       "step": 375
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000746,
-      "loss": 5.6365,
+      "learning_rate": 7.52e-05,
+      "loss": 2.2819,
       "step": 376
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000748,
-      "loss": 5.403,
+      "learning_rate": 7.54e-05,
+      "loss": 2.2658,
       "step": 377
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00075,
-      "loss": 5.3533,
+      "learning_rate": 7.560000000000001e-05,
+      "loss": 2.7623,
       "step": 378
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.0007520000000000001,
-      "loss": 5.3927,
+      "learning_rate": 7.58e-05,
+      "loss": 2.3471,
       "step": 379
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000754,
-      "loss": 5.939,
+      "learning_rate": 7.6e-05,
+      "loss": 2.6372,
       "step": 380
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000756,
-      "loss": 5.0415,
+      "learning_rate": 7.620000000000001e-05,
+      "loss": 2.1763,
       "step": 381
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.000758,
-      "loss": 5.493,
+      "learning_rate": 7.64e-05,
+      "loss": 2.5379,
       "step": 382
     },
     {
       "epoch": 0.21,
-      "learning_rate": 0.00076,
-      "loss": 5.7137,
+      "learning_rate": 7.66e-05,
+      "loss": 2.5526,
       "step": 383
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000762,
-      "loss": 6.2573,
+      "learning_rate": 7.680000000000001e-05,
+      "loss": 2.2586,
       "step": 384
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000764,
-      "loss": 5.7064,
+      "learning_rate": 7.7e-05,
+      "loss": 3.0208,
       "step": 385
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007660000000000001,
-      "loss": 5.6286,
+      "learning_rate": 7.72e-05,
+      "loss": 2.4657,
       "step": 386
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000768,
-      "loss": 6.3579,
+      "learning_rate": 7.740000000000001e-05,
+      "loss": 2.0677,
       "step": 387
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007700000000000001,
-      "loss": 5.2865,
+      "learning_rate": 7.76e-05,
+      "loss": 2.3039,
       "step": 388
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000772,
-      "loss": 5.2695,
+      "learning_rate": 7.780000000000001e-05,
+      "loss": 2.208,
       "step": 389
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007740000000000001,
-      "loss": 6.6205,
+      "learning_rate": 7.800000000000001e-05,
+      "loss": 2.7193,
       "step": 390
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000776,
-      "loss": 5.2116,
+      "learning_rate": 7.82e-05,
+      "loss": 1.8861,
       "step": 391
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000778,
-      "loss": 6.2093,
+      "learning_rate": 7.840000000000001e-05,
+      "loss": 2.1343,
       "step": 392
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007800000000000001,
-      "loss": 5.5573,
+      "learning_rate": 7.860000000000001e-05,
+      "loss": 2.0452,
       "step": 393
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000782,
-      "loss": 5.7461,
+      "learning_rate": 7.88e-05,
+      "loss": 1.8748,
       "step": 394
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007840000000000001,
-      "loss": 5.8815,
+      "learning_rate": 7.900000000000001e-05,
+      "loss": 1.8931,
       "step": 395
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000786,
-      "loss": 7.0836,
+      "learning_rate": 7.920000000000001e-05,
+      "loss": 1.8485,
       "step": 396
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007880000000000001,
-      "loss": 5.584,
+      "learning_rate": 7.94e-05,
+      "loss": 1.5074,
       "step": 397
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.00079,
-      "loss": 9.7145,
+      "learning_rate": 7.960000000000001e-05,
+      "loss": 1.8929,
       "step": 398
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007920000000000001,
-      "loss": 11.0987,
+      "learning_rate": 7.98e-05,
+      "loss": 1.753,
       "step": 399
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.0007940000000000001,
-      "loss": 5.9893,
+      "learning_rate": 8e-05,
+      "loss": 1.4719,
       "step": 400
     },
     {
       "epoch": 0.22,
-      "learning_rate": 0.000796,
-      "loss": 5.8344,
+      "learning_rate": 8.020000000000001e-05,
+      "loss": 3.4298,
       "step": 401
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0007980000000000001,
-      "loss": 5.8467,
+      "learning_rate": 8.04e-05,
+      "loss": 2.7305,
       "step": 402
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008,
-      "loss": 5.403,
+      "learning_rate": 8.060000000000001e-05,
+      "loss": 2.7848,
       "step": 403
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008020000000000001,
-      "loss": 5.8787,
+      "learning_rate": 8.080000000000001e-05,
+      "loss": 2.7461,
       "step": 404
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000804,
-      "loss": 5.47,
+      "learning_rate": 8.1e-05,
+      "loss": 2.2443,
       "step": 405
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008060000000000001,
-      "loss": 6.4202,
+      "learning_rate": 8.120000000000001e-05,
+      "loss": 3.1046,
       "step": 406
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000808,
-      "loss": 5.7586,
+      "learning_rate": 8.14e-05,
+      "loss": 3.4058,
       "step": 407
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008100000000000001,
-      "loss": 5.5644,
+      "learning_rate": 8.16e-05,
+      "loss": 2.635,
       "step": 408
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008120000000000001,
-      "loss": 5.9939,
+      "learning_rate": 8.18e-05,
+      "loss": 3.2202,
       "step": 409
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008139999999999999,
-      "loss": 5.7594,
+      "learning_rate": 8.2e-05,
+      "loss": 2.707,
       "step": 410
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000816,
-      "loss": 6.3656,
+      "learning_rate": 8.22e-05,
+      "loss": 2.7637,
       "step": 411
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008179999999999999,
-      "loss": 5.9035,
+      "learning_rate": 8.24e-05,
+      "loss": 2.6614,
       "step": 412
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00082,
-      "loss": 6.1055,
+      "learning_rate": 8.26e-05,
+      "loss": 3.2864,
       "step": 413
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.0008219999999999999,
-      "loss": 5.7212,
+      "learning_rate": 8.28e-05,
+      "loss": 2.8326,
       "step": 414
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000824,
-      "loss": 6.1417,
+      "learning_rate": 8.3e-05,
+      "loss": 2.4039,
       "step": 415
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000826,
-      "loss": 5.8061,
+      "learning_rate": 8.32e-05,
+      "loss": 2.5581,
       "step": 416
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000828,
-      "loss": 6.2892,
+      "learning_rate": 8.34e-05,
+      "loss": 2.695,
       "step": 417
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.00083,
-      "loss": 5.7129,
+      "learning_rate": 8.36e-05,
+      "loss": 2.4815,
       "step": 418
     },
     {
       "epoch": 0.23,
-      "learning_rate": 0.000832,
-      "loss": 6.6271,
+      "learning_rate": 8.38e-05,
+      "loss": 3.034,
       "step": 419
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000834,
-      "loss": 5.8522,
+      "learning_rate": 8.4e-05,
+      "loss": 2.6319,
       "step": 420
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0008359999999999999,
-      "loss": 5.6503,
+      "learning_rate": 8.42e-05,
+      "loss": 2.4595,
       "step": 421
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000838,
-      "loss": 5.6825,
+      "learning_rate": 8.44e-05,
+      "loss": 2.8489,
       "step": 422
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00084,
-      "loss": 5.4941,
+      "learning_rate": 8.46e-05,
+      "loss": 2.1619,
       "step": 423
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000842,
-      "loss": 5.8587,
+      "learning_rate": 8.48e-05,
+      "loss": 2.5888,
       "step": 424
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000844,
-      "loss": 5.6034,
+      "learning_rate": 8.5e-05,
+      "loss": 2.2599,
       "step": 425
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000846,
-      "loss": 5.5805,
+      "learning_rate": 8.52e-05,
+      "loss": 2.6586,
       "step": 426
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000848,
-      "loss": 6.4935,
+      "learning_rate": 8.54e-05,
+      "loss": 2.2628,
       "step": 427
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00085,
-      "loss": 8.7125,
+      "learning_rate": 8.560000000000001e-05,
+      "loss": 2.6195,
       "step": 428
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000852,
-      "loss": 5.5379,
+      "learning_rate": 8.58e-05,
+      "loss": 2.2114,
       "step": 429
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000854,
-      "loss": 6.774,
+      "learning_rate": 8.6e-05,
+      "loss": 2.7246,
       "step": 430
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000856,
-      "loss": 6.1877,
+      "learning_rate": 8.620000000000001e-05,
+      "loss": 2.8309,
       "step": 431
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000858,
-      "loss": 5.7596,
+      "learning_rate": 8.64e-05,
+      "loss": 2.3465,
       "step": 432
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.00086,
-      "loss": 5.5451,
+      "learning_rate": 8.66e-05,
+      "loss": 2.1822,
       "step": 433
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000862,
-      "loss": 5.5449,
+      "learning_rate": 8.680000000000001e-05,
+      "loss": 2.0905,
       "step": 434
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000864,
-      "loss": 5.557,
+      "learning_rate": 8.7e-05,
+      "loss": 2.3345,
       "step": 435
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.000866,
-      "loss": 5.2794,
+      "learning_rate": 8.72e-05,
+      "loss": 2.2854,
       "step": 436
     },
     {
       "epoch": 0.24,
-      "learning_rate": 0.0008680000000000001,
-      "loss": 5.6317,
+      "learning_rate": 8.740000000000001e-05,
+      "loss": 2.1054,
       "step": 437
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00087,
-      "loss": 6.0562,
+      "learning_rate": 8.76e-05,
+      "loss": 2.4255,
       "step": 438
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000872,
-      "loss": 5.3547,
+      "learning_rate": 8.78e-05,
+      "loss": 2.0983,
       "step": 439
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000874,
-      "loss": 5.578,
+      "learning_rate": 8.800000000000001e-05,
+      "loss": 2.1981,
       "step": 440
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000876,
-      "loss": 5.8526,
+      "learning_rate": 8.82e-05,
+      "loss": 2.1805,
       "step": 441
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000878,
-      "loss": 6.8698,
+      "learning_rate": 8.840000000000001e-05,
+      "loss": 2.0666,
       "step": 442
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.00088,
-      "loss": 5.4054,
+      "learning_rate": 8.86e-05,
+      "loss": 2.0275,
       "step": 443
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000882,
-      "loss": 5.7684,
+      "learning_rate": 8.88e-05,
+      "loss": 1.7247,
       "step": 444
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000884,
-      "loss": 5.5407,
+      "learning_rate": 8.900000000000001e-05,
+      "loss": 1.7722,
       "step": 445
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0008860000000000001,
-      "loss": 7.7708,
+      "learning_rate": 8.92e-05,
+      "loss": 2.1902,
       "step": 446
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000888,
-      "loss": 6.5105,
+      "learning_rate": 8.94e-05,
+      "loss": 2.1527,
       "step": 447
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0008900000000000001,
-      "loss": 6.0654,
+      "learning_rate": 8.960000000000001e-05,
+      "loss": 1.7806,
       "step": 448
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000892,
-      "loss": 10.2675,
+      "learning_rate": 8.98e-05,
+      "loss": 1.4787,
       "step": 449
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000894,
-      "loss": 6.4337,
+      "learning_rate": 9e-05,
+      "loss": 1.6105,
       "step": 450
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000896,
-      "loss": 5.6065,
+      "learning_rate": 9.020000000000001e-05,
+      "loss": 2.5647,
       "step": 451
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000898,
-      "loss": 6.0042,
+      "learning_rate": 9.04e-05,
+      "loss": 2.5917,
       "step": 452
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.0009000000000000001,
-      "loss": 7.2876,
+      "learning_rate": 9.06e-05,
+      "loss": 2.6298,
       "step": 453
     },
     {
       "epoch": 0.25,
-      "learning_rate": 0.000902,
-      "loss": 5.697,
+      "learning_rate": 9.080000000000001e-05,
+      "loss": 2.9009,
       "step": 454
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009040000000000001,
-      "loss": 6.536,
+      "learning_rate": 9.1e-05,
+      "loss": 2.4801,
       "step": 455
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.000906,
-      "loss": 5.861,
+      "learning_rate": 9.120000000000001e-05,
+      "loss": 2.4988,
       "step": 456
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009080000000000001,
-      "loss": 5.9541,
+      "learning_rate": 9.140000000000001e-05,
+      "loss": 2.9022,
       "step": 457
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00091,
-      "loss": 6.7376,
+      "learning_rate": 9.16e-05,
+      "loss": 2.4766,
       "step": 458
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.000912,
-      "loss": 5.5201,
+      "learning_rate": 9.180000000000001e-05,
+      "loss": 2.3251,
       "step": 459
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009140000000000001,
-      "loss": 6.2328,
+      "learning_rate": 9.200000000000001e-05,
+      "loss": 2.3137,
       "step": 460
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.000916,
-      "loss": 6.746,
+      "learning_rate": 9.22e-05,
+      "loss": 2.7779,
       "step": 461
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009180000000000001,
-      "loss": 5.5854,
+      "learning_rate": 9.240000000000001e-05,
+      "loss": 2.4092,
       "step": 462
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00092,
-      "loss": 6.1128,
+      "learning_rate": 9.260000000000001e-05,
+      "loss": 2.3993,
       "step": 463
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009220000000000001,
-      "loss": 5.6173,
+      "learning_rate": 9.28e-05,
+      "loss": 2.3057,
       "step": 464
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.000924,
-      "loss": 5.556,
+      "learning_rate": 9.300000000000001e-05,
+      "loss": 2.4424,
       "step": 465
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009260000000000001,
-      "loss": 6.0281,
+      "learning_rate": 9.320000000000002e-05,
+      "loss": 2.6183,
       "step": 466
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009280000000000001,
-      "loss": 5.6352,
+      "learning_rate": 9.340000000000001e-05,
+      "loss": 2.2283,
       "step": 467
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.00093,
-      "loss": 6.1857,
+      "learning_rate": 9.360000000000001e-05,
+      "loss": 2.552,
       "step": 468
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009320000000000001,
-      "loss": 5.3981,
+      "learning_rate": 9.38e-05,
+      "loss": 2.4589,
       "step": 469
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.000934,
-      "loss": 5.7451,
+      "learning_rate": 9.4e-05,
+      "loss": 2.4157,
       "step": 470
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009360000000000001,
-      "loss": 6.0019,
+      "learning_rate": 9.42e-05,
+      "loss": 2.5367,
       "step": 471
     },
     {
       "epoch": 0.26,
-      "learning_rate": 0.0009379999999999999,
-      "loss": 5.4895,
+      "learning_rate": 9.44e-05,
+      "loss": 2.0546,
       "step": 472
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00094,
-      "loss": 5.5495,
+      "learning_rate": 9.46e-05,
+      "loss": 2.4708,
       "step": 473
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000942,
-      "loss": 6.0448,
+      "learning_rate": 9.48e-05,
+      "loss": 2.8027,
       "step": 474
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000944,
-      "loss": 5.8212,
+      "learning_rate": 9.5e-05,
+      "loss": 2.4798,
       "step": 475
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000946,
-      "loss": 6.0345,
+      "learning_rate": 9.52e-05,
+      "loss": 2.3653,
       "step": 476
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000948,
-      "loss": 5.9918,
+      "learning_rate": 9.54e-05,
+      "loss": 2.5442,
       "step": 477
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00095,
-      "loss": 5.8179,
+      "learning_rate": 9.56e-05,
+      "loss": 2.6838,
       "step": 478
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0009519999999999999,
-      "loss": 5.8431,
+      "learning_rate": 9.58e-05,
+      "loss": 2.4813,
       "step": 479
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000954,
-      "loss": 7.2752,
+      "learning_rate": 9.6e-05,
+      "loss": 2.2251,
       "step": 480
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0009559999999999999,
-      "loss": 5.6978,
+      "learning_rate": 9.620000000000001e-05,
+      "loss": 2.1992,
       "step": 481
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000958,
-      "loss": 7.397,
+      "learning_rate": 9.64e-05,
+      "loss": 2.8325,
       "step": 482
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.00096,
-      "loss": 6.5424,
+      "learning_rate": 9.66e-05,
+      "loss": 2.5211,
       "step": 483
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000962,
-      "loss": 5.8226,
+      "learning_rate": 9.680000000000001e-05,
+      "loss": 2.3187,
       "step": 484
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000964,
-      "loss": 6.5297,
+      "learning_rate": 9.7e-05,
+      "loss": 2.473,
       "step": 485
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000966,
-      "loss": 6.5972,
+      "learning_rate": 9.72e-05,
+      "loss": 2.7107,
       "step": 486
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000968,
-      "loss": 5.6045,
+      "learning_rate": 9.74e-05,
+      "loss": 2.3596,
       "step": 487
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.0009699999999999999,
-      "loss": 6.1922,
+      "learning_rate": 9.76e-05,
+      "loss": 2.6862,
       "step": 488
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000972,
-      "loss": 6.0956,
+      "learning_rate": 9.78e-05,
+      "loss": 2.897,
       "step": 489
     },
     {
       "epoch": 0.27,
-      "learning_rate": 0.000974,
-      "loss": 5.7588,
+      "learning_rate": 9.8e-05,
+      "loss": 2.3983,
       "step": 490
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000976,
-      "loss": 5.5526,
+      "learning_rate": 9.82e-05,
+      "loss": 2.3651,
       "step": 491
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000978,
-      "loss": 6.2006,
+      "learning_rate": 9.84e-05,
+      "loss": 2.609,
       "step": 492
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00098,
-      "loss": 5.8593,
+      "learning_rate": 9.86e-05,
+      "loss": 2.2454,
       "step": 493
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000982,
-      "loss": 6.4826,
+      "learning_rate": 9.88e-05,
+      "loss": 2.6882,
       "step": 494
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000984,
-      "loss": 6.0277,
+      "learning_rate": 9.900000000000001e-05,
+      "loss": 2.7959,
       "step": 495
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0009860000000000001,
-      "loss": 6.0805,
+      "learning_rate": 9.92e-05,
+      "loss": 2.1536,
       "step": 496
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000988,
-      "loss": 5.9525,
+      "learning_rate": 9.94e-05,
+      "loss": 2.0828,
       "step": 497
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.00099,
-      "loss": 7.538,
+      "learning_rate": 9.960000000000001e-05,
+      "loss": 2.1012,
       "step": 498
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000992,
-      "loss": 5.6965,
+      "learning_rate": 9.98e-05,
+      "loss": 1.9902,
       "step": 499
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000994,
-      "loss": 5.4068,
+      "learning_rate": 0.0001,
+      "loss": 1.9364,
       "step": 500
     },
     {
       "epoch": 0.28,
-      "eval_loss": 6.093479156494141,
-      "eval_runtime": 1249.6964,
-      "eval_samples_per_second": 2.114,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.9428203094010312,
+      "eval_loss": 6.361258506774902,
+      "eval_runtime": 1235.8089,
+      "eval_samples_per_second": 2.138,
+      "eval_steps_per_second": 0.268,
+      "eval_wer": 1.9833399444664817,
       "step": 500
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000996,
-      "loss": 6.9039,
+      "learning_rate": 9.997938994229185e-05,
+      "loss": 4.0044,
       "step": 501
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000998,
-      "loss": 7.2881,
+      "learning_rate": 9.995877988458369e-05,
+      "loss": 4.2251,
       "step": 502
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.001,
-      "loss": 6.454,
+      "learning_rate": 9.993816982687552e-05,
+      "loss": 3.6946,
       "step": 503
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0009997938994229184,
-      "loss": 5.8195,
+      "learning_rate": 9.991755976916735e-05,
+      "loss": 3.4519,
       "step": 504
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0009995877988458368,
-      "loss": 6.1896,
+      "learning_rate": 9.98969497114592e-05,
+      "loss": 3.8298,
       "step": 505
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0009993816982687552,
-      "loss": 9.0881,
+      "learning_rate": 9.987633965375104e-05,
+      "loss": 4.0296,
       "step": 506
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.0009991755976916736,
-      "loss": 5.5124,
+      "learning_rate": 9.985572959604287e-05,
+      "loss": 3.4455,
       "step": 507
     },
     {
       "epoch": 0.28,
-      "learning_rate": 0.000998969497114592,
-      "loss": 5.7427,
+      "learning_rate": 9.983511953833471e-05,
+      "loss": 3.6073,
       "step": 508
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009987633965375105,
-      "loss": 6.6938,
+      "learning_rate": 9.981450948062655e-05,
+      "loss": 3.6601,
       "step": 509
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009985572959604287,
-      "loss": 5.789,
+      "learning_rate": 9.979389942291838e-05,
+      "loss": 3.6538,
       "step": 510
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.000998351195383347,
-      "loss": 5.7052,
+      "learning_rate": 9.977328936521023e-05,
+      "loss": 3.3946,
       "step": 511
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009981450948062655,
-      "loss": 6.1583,
+      "learning_rate": 9.975267930750206e-05,
+      "loss": 3.3129,
       "step": 512
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009979389942291839,
-      "loss": 6.2109,
+      "learning_rate": 9.97320692497939e-05,
+      "loss": 3.4938,
       "step": 513
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009977328936521023,
-      "loss": 5.2647,
+      "learning_rate": 9.971145919208574e-05,
+      "loss": 3.2177,
       "step": 514
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009975267930750207,
-      "loss": 5.5793,
+      "learning_rate": 9.969084913437759e-05,
+      "loss": 3.0427,
       "step": 515
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.000997320692497939,
-      "loss": 7.4105,
+      "learning_rate": 9.967023907666943e-05,
+      "loss": 3.9953,
       "step": 516
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009971145919208575,
-      "loss": 6.5091,
+      "learning_rate": 9.964962901896126e-05,
+      "loss": 3.7048,
       "step": 517
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.000996908491343776,
-      "loss": 6.1413,
+      "learning_rate": 9.962901896125309e-05,
+      "loss": 3.6099,
       "step": 518
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009967023907666943,
-      "loss": 5.7145,
+      "learning_rate": 9.960840890354493e-05,
+      "loss": 3.5195,
       "step": 519
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009964962901896125,
-      "loss": 6.0263,
+      "learning_rate": 9.958779884583678e-05,
+      "loss": 3.4493,
       "step": 520
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.000996290189612531,
-      "loss": 5.3656,
+      "learning_rate": 9.956718878812861e-05,
+      "loss": 3.216,
       "step": 521
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009960840890354493,
-      "loss": 6.0043,
+      "learning_rate": 9.954657873042045e-05,
+      "loss": 3.2475,
       "step": 522
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009958779884583677,
-      "loss": 6.4915,
+      "learning_rate": 9.95259686727123e-05,
+      "loss": 3.3079,
       "step": 523
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009956718878812861,
-      "loss": 5.9178,
+      "learning_rate": 9.950535861500412e-05,
+      "loss": 3.3853,
       "step": 524
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.0009954657873042045,
-      "loss": 7.8161,
+      "learning_rate": 9.948474855729597e-05,
+      "loss": 3.6501,
       "step": 525
     },
     {
       "epoch": 0.29,
-      "learning_rate": 0.000995259686727123,
-      "loss": 5.6401,
+      "learning_rate": 9.94641384995878e-05,
+      "loss": 3.1468,
       "step": 526
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009950535861500413,
-      "loss": 6.6302,
+      "learning_rate": 9.944352844187964e-05,
+      "loss": 3.4505,
       "step": 527
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009948474855729598,
-      "loss": 5.6324,
+      "learning_rate": 9.942291838417148e-05,
+      "loss": 3.2271,
       "step": 528
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.000994641384995878,
-      "loss": 6.2261,
+      "learning_rate": 9.940230832646333e-05,
+      "loss": 3.5235,
       "step": 529
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009944352844187964,
-      "loss": 7.1911,
+      "learning_rate": 9.938169826875516e-05,
+      "loss": 3.746,
       "step": 530
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009942291838417148,
-      "loss": 5.6159,
+      "learning_rate": 9.936108821104699e-05,
+      "loss": 3.4608,
       "step": 531
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009940230832646332,
-      "loss": 5.4483,
+      "learning_rate": 9.934047815333883e-05,
+      "loss": 2.9581,
       "step": 532
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009938169826875516,
-      "loss": 5.6703,
+      "learning_rate": 9.931986809563067e-05,
+      "loss": 3.2081,
       "step": 533
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.00099361088211047,
-      "loss": 6.1458,
+      "learning_rate": 9.929925803792252e-05,
+      "loss": 3.4153,
       "step": 534
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009934047815333884,
-      "loss": 6.1687,
+      "learning_rate": 9.927864798021435e-05,
+      "loss": 3.4724,
       "step": 535
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009931986809563068,
-      "loss": 5.733,
+      "learning_rate": 9.925803792250619e-05,
+      "loss": 3.24,
       "step": 536
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009929925803792252,
-      "loss": 5.6151,
+      "learning_rate": 9.923742786479803e-05,
+      "loss": 3.2011,
       "step": 537
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009927864798021434,
-      "loss": 5.9331,
+      "learning_rate": 9.921681780708986e-05,
+      "loss": 3.3735,
       "step": 538
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009925803792250618,
-      "loss": 5.6894,
+      "learning_rate": 9.91962077493817e-05,
+      "loss": 3.0658,
       "step": 539
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009923742786479802,
-      "loss": 5.7838,
+      "learning_rate": 9.917559769167354e-05,
+      "loss": 3.1974,
       "step": 540
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009921681780708986,
-      "loss": 7.2282,
+      "learning_rate": 9.915498763396538e-05,
+      "loss": 3.2856,
       "step": 541
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.000991962077493817,
-      "loss": 6.2396,
+      "learning_rate": 9.913437757625722e-05,
+      "loss": 3.2759,
       "step": 542
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009917559769167354,
-      "loss": 7.2881,
+      "learning_rate": 9.911376751854907e-05,
+      "loss": 3.3345,
       "step": 543
     },
     {
       "epoch": 0.3,
-      "learning_rate": 0.0009915498763396538,
-      "loss": 5.969,
+      "learning_rate": 9.90931574608409e-05,
+      "loss": 2.9336,
       "step": 544
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009913437757625722,
-      "loss": 5.6602,
+      "learning_rate": 9.907254740313273e-05,
+      "loss": 2.905,
       "step": 545
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009911376751854906,
-      "loss": 7.742,
+      "learning_rate": 9.905193734542457e-05,
+      "loss": 2.8371,
       "step": 546
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009909315746084088,
-      "loss": 6.4301,
+      "learning_rate": 9.903132728771641e-05,
+      "loss": 2.5109,
       "step": 547
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009907254740313272,
-      "loss": 7.2254,
+      "learning_rate": 9.901071723000826e-05,
+      "loss": 2.9694,
       "step": 548
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009905193734542456,
-      "loss": 7.9274,
+      "learning_rate": 9.899010717230009e-05,
+      "loss": 3.0376,
       "step": 549
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.000990313272877164,
-      "loss": 6.0151,
+      "learning_rate": 9.896949711459193e-05,
+      "loss": 2.5348,
       "step": 550
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009901071723000825,
-      "loss": 6.6425,
+      "learning_rate": 9.894888705688376e-05,
+      "loss": 4.0025,
       "step": 551
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009899010717230009,
-      "loss": 6.7184,
+      "learning_rate": 9.89282769991756e-05,
+      "loss": 3.6007,
       "step": 552
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009896949711459193,
-      "loss": 6.3715,
+      "learning_rate": 9.890766694146743e-05,
+      "loss": 3.5299,
       "step": 553
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009894888705688377,
-      "loss": 5.9396,
+      "learning_rate": 9.888705688375928e-05,
+      "loss": 3.4078,
       "step": 554
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.000989282769991756,
-      "loss": 5.7681,
+      "learning_rate": 9.886644682605112e-05,
+      "loss": 3.4435,
       "step": 555
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009890766694146743,
-      "loss": 6.3127,
+      "learning_rate": 9.884583676834296e-05,
+      "loss": 3.424,
       "step": 556
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009888705688375927,
-      "loss": 6.206,
+      "learning_rate": 9.882522671063479e-05,
+      "loss": 3.5493,
       "step": 557
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.000988664468260511,
-      "loss": 6.8175,
+      "learning_rate": 9.880461665292662e-05,
+      "loss": 3.4035,
       "step": 558
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009884583676834295,
-      "loss": 5.5251,
+      "learning_rate": 9.878400659521847e-05,
+      "loss": 3.3964,
       "step": 559
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.000988252267106348,
-      "loss": 5.6403,
+      "learning_rate": 9.876339653751031e-05,
+      "loss": 3.1389,
       "step": 560
     },
     {
       "epoch": 0.31,
-      "learning_rate": 0.0009880461665292663,
-      "loss": 5.882,
+      "learning_rate": 9.874278647980215e-05,
+      "loss": 3.6905,
       "step": 561
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009878400659521847,
-      "loss": 5.891,
+      "learning_rate": 9.872217642209398e-05,
+      "loss": 3.4614,
       "step": 562
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009876339653751031,
-      "loss": 5.5482,
+      "learning_rate": 9.870156636438583e-05,
+      "loss": 3.0831,
       "step": 563
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009874278647980215,
-      "loss": 5.9301,
+      "learning_rate": 9.868095630667767e-05,
+      "loss": 3.4066,
       "step": 564
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009872217642209397,
-      "loss": 5.9511,
+      "learning_rate": 9.86603462489695e-05,
+      "loss": 3.1213,
       "step": 565
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009870156636438581,
-      "loss": 5.6083,
+      "learning_rate": 9.863973619126134e-05,
+      "loss": 3.3376,
       "step": 566
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009868095630667765,
-      "loss": 6.5232,
+      "learning_rate": 9.861912613355317e-05,
+      "loss": 3.4888,
       "step": 567
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.000986603462489695,
-      "loss": 5.9832,
+      "learning_rate": 9.859851607584502e-05,
+      "loss": 3.9008,
       "step": 568
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009863973619126133,
-      "loss": 5.7986,
+      "learning_rate": 9.857790601813686e-05,
+      "loss": 3.5956,
       "step": 569
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009861912613355318,
-      "loss": 5.7474,
+      "learning_rate": 9.85572959604287e-05,
+      "loss": 3.4843,
       "step": 570
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009859851607584502,
-      "loss": 6.1929,
+      "learning_rate": 9.853668590272053e-05,
+      "loss": 3.3147,
       "step": 571
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009857790601813686,
-      "loss": 5.623,
+      "learning_rate": 9.851607584501236e-05,
+      "loss": 3.424,
       "step": 572
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.000985572959604287,
-      "loss": 6.2775,
+      "learning_rate": 9.84954657873042e-05,
+      "loss": 3.4497,
       "step": 573
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009853668590272052,
-      "loss": 6.6127,
+      "learning_rate": 9.847485572959605e-05,
+      "loss": 3.5343,
       "step": 574
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009851607584501236,
-      "loss": 5.7773,
+      "learning_rate": 9.845424567188789e-05,
+      "loss": 3.2311,
       "step": 575
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.000984954657873042,
-      "loss": 5.8412,
+      "learning_rate": 9.843363561417972e-05,
+      "loss": 3.4899,
       "step": 576
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009847485572959604,
-      "loss": 5.8474,
+      "learning_rate": 9.841302555647157e-05,
+      "loss": 3.3373,
       "step": 577
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009845424567188788,
-      "loss": 5.5494,
+      "learning_rate": 9.83924154987634e-05,
+      "loss": 3.2763,
       "step": 578
     },
     {
       "epoch": 0.32,
-      "learning_rate": 0.0009843363561417972,
-      "loss": 5.8369,
+      "learning_rate": 9.837180544105524e-05,
+      "loss": 3.0815,
       "step": 579
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009841302555647156,
-      "loss": 6.1954,
+      "learning_rate": 9.835119538334708e-05,
+      "loss": 3.6294,
       "step": 580
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.000983924154987634,
-      "loss": 5.5802,
+      "learning_rate": 9.833058532563891e-05,
+      "loss": 3.0888,
       "step": 581
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009837180544105524,
-      "loss": 5.6219,
+      "learning_rate": 9.830997526793076e-05,
+      "loss": 3.185,
       "step": 582
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009835119538334708,
-      "loss": 5.7712,
+      "learning_rate": 9.82893652102226e-05,
+      "loss": 3.2505,
       "step": 583
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.000983305853256389,
-      "loss": 5.9541,
+      "learning_rate": 9.826875515251444e-05,
+      "loss": 3.4471,
       "step": 584
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009830997526793074,
-      "loss": 5.8814,
+      "learning_rate": 9.824814509480627e-05,
+      "loss": 3.5605,
       "step": 585
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009828936521022258,
-      "loss": 5.3727,
+      "learning_rate": 9.82275350370981e-05,
+      "loss": 2.9726,
       "step": 586
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009826875515251442,
-      "loss": 5.9545,
+      "learning_rate": 9.820692497938995e-05,
+      "loss": 3.1855,
       "step": 587
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009824814509480626,
-      "loss": 5.6051,
+      "learning_rate": 9.818631492168179e-05,
+      "loss": 3.1231,
       "step": 588
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.000982275350370981,
-      "loss": 6.2995,
+      "learning_rate": 9.816570486397363e-05,
+      "loss": 3.1023,
       "step": 589
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009820692497938995,
-      "loss": 6.0274,
+      "learning_rate": 9.814509480626546e-05,
+      "loss": 3.2682,
       "step": 590
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009818631492168179,
-      "loss": 5.613,
+      "learning_rate": 9.81244847485573e-05,
+      "loss": 2.9668,
       "step": 591
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009816570486397363,
-      "loss": 5.6562,
+      "learning_rate": 9.810387469084913e-05,
+      "loss": 2.968,
       "step": 592
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009814509480626545,
-      "loss": 5.5516,
+      "learning_rate": 9.808326463314098e-05,
+      "loss": 2.7085,
       "step": 593
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009812448474855729,
-      "loss": 5.7627,
+      "learning_rate": 9.806265457543281e-05,
+      "loss": 2.7702,
       "step": 594
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009810387469084913,
-      "loss": 5.5813,
+      "learning_rate": 9.804204451772465e-05,
+      "loss": 2.3152,
       "step": 595
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.0009808326463314099,
-      "loss": 5.636,
+      "learning_rate": 9.80214344600165e-05,
+      "loss": 2.4291,
       "step": 596
     },
     {
       "epoch": 0.33,
-      "learning_rate": 0.000980626545754328,
-      "loss": 10.2422,
+      "learning_rate": 9.800082440230834e-05,
+      "loss": 2.9401,
       "step": 597
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009804204451772465,
-      "loss": 8.1904,
+      "learning_rate": 9.798021434460017e-05,
+      "loss": 2.5082,
       "step": 598
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.000980214344600165,
-      "loss": 5.8852,
+      "learning_rate": 9.7959604286892e-05,
+      "loss": 2.5991,
       "step": 599
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009800082440230833,
-      "loss": 6.0553,
+      "learning_rate": 9.793899422918384e-05,
+      "loss": 2.3768,
       "step": 600
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009798021434460017,
-      "loss": 6.3463,
+      "learning_rate": 9.791838417147568e-05,
+      "loss": 3.5405,
       "step": 601
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009795960428689201,
-      "loss": 6.1136,
+      "learning_rate": 9.789777411376753e-05,
+      "loss": 3.5755,
       "step": 602
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009793899422918385,
-      "loss": 5.5926,
+      "learning_rate": 9.787716405605936e-05,
+      "loss": 3.4884,
       "step": 603
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.000979183841714757,
-      "loss": 5.7269,
+      "learning_rate": 9.78565539983512e-05,
+      "loss": 3.2946,
       "step": 604
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009789777411376753,
-      "loss": 6.3031,
+      "learning_rate": 9.783594394064303e-05,
+      "loss": 3.6134,
       "step": 605
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009787716405605935,
-      "loss": 5.541,
+      "learning_rate": 9.781533388293487e-05,
+      "loss": 3.181,
       "step": 606
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.000978565539983512,
-      "loss": 6.0213,
+      "learning_rate": 9.779472382522672e-05,
+      "loss": 3.2648,
       "step": 607
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009783594394064303,
-      "loss": 6.3093,
+      "learning_rate": 9.777411376751855e-05,
+      "loss": 3.4505,
       "step": 608
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009781533388293487,
-      "loss": 6.9093,
+      "learning_rate": 9.775350370981039e-05,
+      "loss": 3.7735,
       "step": 609
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009779472382522672,
-      "loss": 5.7258,
+      "learning_rate": 9.773289365210223e-05,
+      "loss": 3.4957,
       "step": 610
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009777411376751856,
-      "loss": 6.208,
+      "learning_rate": 9.771228359439408e-05,
+      "loss": 3.7905,
       "step": 611
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.000977535037098104,
-      "loss": 5.7205,
+      "learning_rate": 9.769167353668591e-05,
+      "loss": 3.5635,
       "step": 612
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009773289365210224,
-      "loss": 6.4296,
+      "learning_rate": 9.767106347897774e-05,
+      "loss": 3.6682,
       "step": 613
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009771228359439408,
-      "loss": 5.8435,
+      "learning_rate": 9.765045342126958e-05,
+      "loss": 3.4985,
       "step": 614
     },
     {
       "epoch": 0.34,
-      "learning_rate": 0.0009769167353668592,
-      "loss": 6.3646,
+      "learning_rate": 9.762984336356142e-05,
+      "loss": 3.6138,
       "step": 615
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009767106347897774,
-      "loss": 5.683,
+      "learning_rate": 9.760923330585327e-05,
+      "loss": 3.3438,
       "step": 616
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009765045342126958,
-      "loss": 5.9436,
+      "learning_rate": 9.75886232481451e-05,
+      "loss": 3.157,
       "step": 617
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009762984336356142,
-      "loss": 5.8275,
+      "learning_rate": 9.756801319043694e-05,
+      "loss": 3.4407,
       "step": 618
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009760923330585326,
-      "loss": 6.4333,
+      "learning_rate": 9.754740313272877e-05,
+      "loss": 3.143,
       "step": 619
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009758862324814509,
-      "loss": 6.5196,
+      "learning_rate": 9.752679307502061e-05,
+      "loss": 3.5754,
       "step": 620
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009756801319043693,
-      "loss": 5.7596,
+      "learning_rate": 9.750618301731246e-05,
+      "loss": 3.3723,
       "step": 621
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009754740313272877,
-      "loss": 5.7417,
+      "learning_rate": 9.748557295960429e-05,
+      "loss": 3.4178,
       "step": 622
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009752679307502061,
-      "loss": 5.7944,
+      "learning_rate": 9.746496290189613e-05,
+      "loss": 3.5491,
       "step": 623
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009750618301731245,
-      "loss": 6.7521,
+      "learning_rate": 9.744435284418797e-05,
+      "loss": 3.6676,
       "step": 624
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009748557295960428,
-      "loss": 5.6838,
+      "learning_rate": 9.74237427864798e-05,
+      "loss": 3.348,
       "step": 625
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009746496290189612,
-      "loss": 5.5583,
+      "learning_rate": 9.740313272877163e-05,
+      "loss": 3.2028,
       "step": 626
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009744435284418796,
-      "loss": 5.5746,
+      "learning_rate": 9.738252267106348e-05,
+      "loss": 3.1043,
       "step": 627
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.000974237427864798,
-      "loss": 7.0217,
+      "learning_rate": 9.736191261335532e-05,
+      "loss": 3.8329,
       "step": 628
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009740313272877163,
-      "loss": 7.019,
+      "learning_rate": 9.734130255564716e-05,
+      "loss": 3.7316,
       "step": 629
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009738252267106347,
-      "loss": 5.555,
+      "learning_rate": 9.732069249793901e-05,
+      "loss": 3.1397,
       "step": 630
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009736191261335531,
-      "loss": 5.9149,
+      "learning_rate": 9.730008244023084e-05,
+      "loss": 3.62,
       "step": 631
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.0009734130255564716,
-      "loss": 5.7774,
+      "learning_rate": 9.727947238252268e-05,
+      "loss": 3.396,
       "step": 632
     },
     {
       "epoch": 0.35,
-      "learning_rate": 0.00097320692497939,
-      "loss": 5.7222,
+      "learning_rate": 9.725886232481451e-05,
+      "loss": 3.3123,
       "step": 633
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009730008244023083,
-      "loss": 5.9334,
+      "learning_rate": 9.723825226710635e-05,
+      "loss": 3.4306,
       "step": 634
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009727947238252267,
-      "loss": 5.6841,
+      "learning_rate": 9.72176422093982e-05,
+      "loss": 3.2584,
       "step": 635
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009725886232481451,
-      "loss": 5.8093,
+      "learning_rate": 9.719703215169003e-05,
+      "loss": 3.1437,
       "step": 636
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009723825226710635,
-      "loss": 5.5973,
+      "learning_rate": 9.717642209398187e-05,
+      "loss": 2.8176,
       "step": 637
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009721764220939819,
-      "loss": 5.3166,
+      "learning_rate": 9.715581203627371e-05,
+      "loss": 2.8513,
       "step": 638
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009719703215169002,
-      "loss": 5.7042,
+      "learning_rate": 9.713520197856554e-05,
+      "loss": 3.2283,
       "step": 639
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009717642209398186,
-      "loss": 6.4014,
+      "learning_rate": 9.711459192085737e-05,
+      "loss": 3.2215,
       "step": 640
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.000971558120362737,
-      "loss": 5.773,
+      "learning_rate": 9.709398186314922e-05,
+      "loss": 3.1071,
       "step": 641
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009713520197856554,
-      "loss": 6.1479,
+      "learning_rate": 9.707337180544106e-05,
+      "loss": 3.1907,
       "step": 642
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009711459192085737,
-      "loss": 9.3503,
+      "learning_rate": 9.70527617477329e-05,
+      "loss": 3.0012,
       "step": 643
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009709398186314922,
-      "loss": 6.1811,
+      "learning_rate": 9.703215169002475e-05,
+      "loss": 3.1511,
       "step": 644
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009707337180544106,
-      "loss": 5.3168,
+      "learning_rate": 9.701154163231658e-05,
+      "loss": 2.5023,
       "step": 645
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.000970527617477329,
-      "loss": 5.5771,
+      "learning_rate": 9.69909315746084e-05,
+      "loss": 2.8662,
       "step": 646
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009703215169002474,
-      "loss": 8.9244,
+      "learning_rate": 9.697032151690025e-05,
+      "loss": 2.9503,
       "step": 647
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009701154163231657,
-      "loss": 9.2924,
+      "learning_rate": 9.694971145919209e-05,
+      "loss": 3.2732,
       "step": 648
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009699093157460841,
-      "loss": 6.0032,
+      "learning_rate": 9.692910140148392e-05,
+      "loss": 2.2599,
       "step": 649
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.0009697032151690026,
-      "loss": 5.9065,
+      "learning_rate": 9.690849134377577e-05,
+      "loss": 2.3047,
       "step": 650
     },
     {
       "epoch": 0.36,
-      "learning_rate": 0.000969497114591921,
-      "loss": 5.7178,
+      "learning_rate": 9.688788128606761e-05,
+      "loss": 3.5165,
       "step": 651
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009692910140148393,
-      "loss": 5.5791,
+      "learning_rate": 9.686727122835945e-05,
+      "loss": 3.7542,
       "step": 652
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009690849134377577,
-      "loss": 5.8403,
+      "learning_rate": 9.684666117065128e-05,
+      "loss": 3.6448,
       "step": 653
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009688788128606761,
-      "loss": 5.8589,
+      "learning_rate": 9.682605111294311e-05,
+      "loss": 3.7066,
       "step": 654
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009686727122835945,
-      "loss": 5.8483,
+      "learning_rate": 9.680544105523496e-05,
+      "loss": 3.6231,
       "step": 655
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009684666117065129,
-      "loss": 5.6194,
+      "learning_rate": 9.67848309975268e-05,
+      "loss": 3.3567,
       "step": 656
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009682605111294312,
-      "loss": 5.5117,
+      "learning_rate": 9.676422093981864e-05,
+      "loss": 3.317,
       "step": 657
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009680544105523496,
-      "loss": 8.8586,
+      "learning_rate": 9.674361088211047e-05,
+      "loss": 3.8371,
       "step": 658
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.000967848309975268,
-      "loss": 5.7808,
+      "learning_rate": 9.672300082440232e-05,
+      "loss": 3.4645,
       "step": 659
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009676422093981864,
-      "loss": 5.7761,
+      "learning_rate": 9.670239076669415e-05,
+      "loss": 3.3794,
       "step": 660
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009674361088211047,
-      "loss": 6.0116,
+      "learning_rate": 9.668178070898599e-05,
+      "loss": 3.3211,
       "step": 661
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009672300082440231,
-      "loss": 5.6008,
+      "learning_rate": 9.666117065127783e-05,
+      "loss": 3.3556,
       "step": 662
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009670239076669415,
-      "loss": 6.5432,
+      "learning_rate": 9.664056059356966e-05,
+      "loss": 3.677,
       "step": 663
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009668178070898599,
-      "loss": 5.8096,
+      "learning_rate": 9.66199505358615e-05,
+      "loss": 3.5356,
       "step": 664
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009666117065127783,
-      "loss": 5.4905,
+      "learning_rate": 9.659934047815335e-05,
+      "loss": 3.2039,
       "step": 665
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009664056059356966,
-      "loss": 6.0899,
+      "learning_rate": 9.657873042044518e-05,
+      "loss": 3.6271,
       "step": 666
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.000966199505358615,
-      "loss": 5.7675,
+      "learning_rate": 9.655812036273702e-05,
+      "loss": 3.5479,
       "step": 667
     },
     {
       "epoch": 0.37,
-      "learning_rate": 0.0009659934047815334,
-      "loss": 6.1804,
+      "learning_rate": 9.653751030502885e-05,
+      "loss": 3.3746,
       "step": 668
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009657873042044518,
-      "loss": 5.5204,
+      "learning_rate": 9.65169002473207e-05,
+      "loss": 3.1679,
       "step": 669
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009655812036273703,
-      "loss": 5.6149,
+      "learning_rate": 9.649629018961254e-05,
+      "loss": 3.4887,
       "step": 670
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009653751030502886,
-      "loss": 6.0129,
+      "learning_rate": 9.647568013190438e-05,
+      "loss": 3.5134,
       "step": 671
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.000965169002473207,
-      "loss": 6.4508,
+      "learning_rate": 9.645507007419621e-05,
+      "loss": 3.4698,
       "step": 672
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009649629018961254,
-      "loss": 5.4699,
+      "learning_rate": 9.643446001648804e-05,
+      "loss": 3.1711,
       "step": 673
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009647568013190438,
-      "loss": 5.4832,
+      "learning_rate": 9.641384995877989e-05,
+      "loss": 3.2413,
       "step": 674
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009645507007419621,
-      "loss": 5.7112,
+      "learning_rate": 9.639323990107173e-05,
+      "loss": 3.0991,
       "step": 675
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009643446001648805,
-      "loss": 5.6204,
+      "learning_rate": 9.637262984336357e-05,
+      "loss": 3.26,
       "step": 676
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009641384995877989,
-      "loss": 5.4164,
+      "learning_rate": 9.63520197856554e-05,
+      "loss": 3.2756,
       "step": 677
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009639323990107173,
-      "loss": 5.6556,
+      "learning_rate": 9.633140972794725e-05,
+      "loss": 3.3138,
       "step": 678
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009637262984336357,
-      "loss": 6.1959,
+      "learning_rate": 9.631079967023909e-05,
+      "loss": 3.2328,
       "step": 679
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.000963520197856554,
-      "loss": 5.9093,
+      "learning_rate": 9.629018961253092e-05,
+      "loss": 3.239,
       "step": 680
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009633140972794724,
-      "loss": 5.9239,
+      "learning_rate": 9.626957955482275e-05,
+      "loss": 3.141,
       "step": 681
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009631079967023908,
-      "loss": 5.5869,
+      "learning_rate": 9.624896949711459e-05,
+      "loss": 3.5407,
       "step": 682
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009629018961253092,
-      "loss": 6.027,
+      "learning_rate": 9.622835943940644e-05,
+      "loss": 3.4267,
       "step": 683
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009626957955482275,
-      "loss": 5.6705,
+      "learning_rate": 9.620774938169828e-05,
+      "loss": 3.1044,
       "step": 684
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009624896949711459,
-      "loss": 6.0041,
+      "learning_rate": 9.618713932399012e-05,
+      "loss": 3.2873,
       "step": 685
     },
     {
       "epoch": 0.38,
-      "learning_rate": 0.0009622835943940643,
-      "loss": 5.9309,
+      "learning_rate": 9.616652926628195e-05,
+      "loss": 3.3791,
       "step": 686
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009620774938169827,
-      "loss": 6.27,
+      "learning_rate": 9.614591920857378e-05,
+      "loss": 3.4528,
       "step": 687
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009618713932399011,
-      "loss": 5.6478,
+      "learning_rate": 9.612530915086563e-05,
+      "loss": 3.1482,
       "step": 688
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009616652926628194,
-      "loss": 7.2664,
+      "learning_rate": 9.610469909315747e-05,
+      "loss": 3.6231,
       "step": 689
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009614591920857378,
-      "loss": 5.6514,
+      "learning_rate": 9.60840890354493e-05,
+      "loss": 3.2904,
       "step": 690
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009612530915086563,
-      "loss": 5.4726,
+      "learning_rate": 9.606347897774114e-05,
+      "loss": 2.8751,
       "step": 691
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009610469909315747,
-      "loss": 6.023,
+      "learning_rate": 9.604286892003299e-05,
+      "loss": 3.1832,
       "step": 692
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.000960840890354493,
-      "loss": 5.8477,
+      "learning_rate": 9.602225886232481e-05,
+      "loss": 2.7187,
       "step": 693
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009606347897774114,
-      "loss": 5.6152,
+      "learning_rate": 9.600164880461666e-05,
+      "loss": 2.8087,
       "step": 694
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009604286892003298,
-      "loss": 5.6259,
+      "learning_rate": 9.598103874690849e-05,
+      "loss": 2.8381,
       "step": 695
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009602225886232482,
-      "loss": 5.6599,
+      "learning_rate": 9.596042868920033e-05,
+      "loss": 2.5424,
       "step": 696
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009600164880461666,
-      "loss": 5.4909,
+      "learning_rate": 9.593981863149217e-05,
+      "loss": 2.2969,
       "step": 697
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009598103874690849,
-      "loss": 5.7744,
+      "learning_rate": 9.591920857378402e-05,
+      "loss": 2.3119,
       "step": 698
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009596042868920033,
-      "loss": 5.7638,
+      "learning_rate": 9.589859851607586e-05,
+      "loss": 2.2324,
       "step": 699
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009593981863149217,
-      "loss": 5.7855,
+      "learning_rate": 9.587798845836769e-05,
+      "loss": 2.3583,
       "step": 700
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009591920857378401,
-      "loss": 5.6992,
+      "learning_rate": 9.585737840065952e-05,
+      "loss": 3.6291,
       "step": 701
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009589859851607585,
-      "loss": 5.5817,
+      "learning_rate": 9.583676834295136e-05,
+      "loss": 3.3372,
       "step": 702
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009587798845836768,
-      "loss": 6.1692,
+      "learning_rate": 9.581615828524321e-05,
+      "loss": 3.4945,
       "step": 703
     },
     {
       "epoch": 0.39,
-      "learning_rate": 0.0009585737840065952,
-      "loss": 7.3786,
+      "learning_rate": 9.579554822753504e-05,
+      "loss": 3.8717,
       "step": 704
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009583676834295136,
-      "loss": 8.9289,
+      "learning_rate": 9.577493816982688e-05,
+      "loss": 3.643,
       "step": 705
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.000958161582852432,
-      "loss": 6.1429,
+      "learning_rate": 9.575432811211872e-05,
+      "loss": 3.4809,
       "step": 706
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009579554822753503,
-      "loss": 7.9422,
+      "learning_rate": 9.573371805441055e-05,
+      "loss": 3.6176,
       "step": 707
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009577493816982687,
-      "loss": 6.3416,
+      "learning_rate": 9.57131079967024e-05,
+      "loss": 3.8602,
       "step": 708
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009575432811211871,
-      "loss": 5.7308,
+      "learning_rate": 9.569249793899423e-05,
+      "loss": 3.2871,
       "step": 709
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009573371805441055,
-      "loss": 6.0273,
+      "learning_rate": 9.567188788128607e-05,
+      "loss": 3.7482,
       "step": 710
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.000957131079967024,
-      "loss": 5.9464,
+      "learning_rate": 9.565127782357791e-05,
+      "loss": 3.6941,
       "step": 711
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009569249793899422,
-      "loss": 5.7031,
+      "learning_rate": 9.563066776586976e-05,
+      "loss": 3.4712,
       "step": 712
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009567188788128607,
-      "loss": 6.1895,
+      "learning_rate": 9.561005770816159e-05,
+      "loss": 3.5186,
       "step": 713
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009565127782357791,
-      "loss": 6.2069,
+      "learning_rate": 9.558944765045342e-05,
+      "loss": 3.7385,
       "step": 714
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009563066776586975,
-      "loss": 6.162,
+      "learning_rate": 9.556883759274526e-05,
+      "loss": 3.3137,
       "step": 715
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009561005770816158,
-      "loss": 5.4371,
+      "learning_rate": 9.55482275350371e-05,
+      "loss": 3.1186,
       "step": 716
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009558944765045342,
-      "loss": 5.5628,
+      "learning_rate": 9.552761747732895e-05,
+      "loss": 3.6201,
       "step": 717
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009556883759274526,
-      "loss": 6.9056,
+      "learning_rate": 9.550700741962078e-05,
+      "loss": 3.9445,
       "step": 718
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.000955482275350371,
-      "loss": 5.5642,
+      "learning_rate": 9.548639736191262e-05,
+      "loss": 3.7017,
       "step": 719
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009552761747732894,
-      "loss": 5.9009,
+      "learning_rate": 9.546578730420445e-05,
+      "loss": 3.3281,
       "step": 720
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009550700741962077,
-      "loss": 6.071,
+      "learning_rate": 9.54451772464963e-05,
+      "loss": 3.4388,
       "step": 721
     },
     {
       "epoch": 0.4,
-      "learning_rate": 0.0009548639736191261,
-      "loss": 6.0097,
+      "learning_rate": 9.542456718878814e-05,
+      "loss": 3.4915,
       "step": 722
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009546578730420445,
-      "loss": 5.6494,
+      "learning_rate": 9.540395713107997e-05,
+      "loss": 3.3369,
       "step": 723
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009544517724649629,
-      "loss": 6.4281,
+      "learning_rate": 9.538334707337181e-05,
+      "loss": 3.7413,
       "step": 724
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009542456718878813,
-      "loss": 5.8058,
+      "learning_rate": 9.536273701566365e-05,
+      "loss": 3.375,
       "step": 725
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009540395713107996,
-      "loss": 5.9138,
+      "learning_rate": 9.53421269579555e-05,
+      "loss": 3.3654,
       "step": 726
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.000953833470733718,
-      "loss": 5.5967,
+      "learning_rate": 9.532151690024733e-05,
+      "loss": 3.3488,
       "step": 727
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009536273701566364,
-      "loss": 6.0279,
+      "learning_rate": 9.530090684253916e-05,
+      "loss": 3.4605,
       "step": 728
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009534212695795548,
-      "loss": 5.4258,
+      "learning_rate": 9.5280296784831e-05,
+      "loss": 3.3555,
       "step": 729
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009532151690024731,
-      "loss": 5.5545,
+      "learning_rate": 9.525968672712284e-05,
+      "loss": 3.0789,
       "step": 730
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009530090684253915,
-      "loss": 5.568,
+      "learning_rate": 9.523907666941469e-05,
+      "loss": 3.1928,
       "step": 731
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.00095280296784831,
-      "loss": 5.6758,
+      "learning_rate": 9.521846661170652e-05,
+      "loss": 3.5207,
       "step": 732
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009525968672712284,
-      "loss": 7.9411,
+      "learning_rate": 9.519785655399836e-05,
+      "loss": 3.306,
       "step": 733
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009523907666941468,
-      "loss": 5.7769,
+      "learning_rate": 9.517724649629019e-05,
+      "loss": 3.0849,
       "step": 734
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009521846661170651,
-      "loss": 5.7837,
+      "learning_rate": 9.515663643858203e-05,
+      "loss": 3.0036,
       "step": 735
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009519785655399835,
-      "loss": 5.6159,
+      "learning_rate": 9.513602638087386e-05,
+      "loss": 3.0052,
       "step": 736
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.000951772464962902,
-      "loss": 5.4329,
+      "learning_rate": 9.51154163231657e-05,
+      "loss": 3.0359,
       "step": 737
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009515663643858204,
-      "loss": 5.5327,
+      "learning_rate": 9.509480626545755e-05,
+      "loss": 2.835,
       "step": 738
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009513602638087387,
-      "loss": 7.4568,
+      "learning_rate": 9.50741962077494e-05,
+      "loss": 3.3056,
       "step": 739
     },
     {
       "epoch": 0.41,
-      "learning_rate": 0.0009511541632316571,
-      "loss": 5.4145,
+      "learning_rate": 9.505358615004122e-05,
+      "loss": 2.8262,
       "step": 740
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009509480626545755,
-      "loss": 6.4719,
+      "learning_rate": 9.503297609233305e-05,
+      "loss": 3.2274,
       "step": 741
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009507419620774939,
-      "loss": 5.7968,
+      "learning_rate": 9.50123660346249e-05,
+      "loss": 3.2776,
       "step": 742
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009505358615004123,
-      "loss": 9.9958,
+      "learning_rate": 9.499175597691674e-05,
+      "loss": 3.6053,
       "step": 743
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009503297609233306,
-      "loss": 6.1593,
+      "learning_rate": 9.497114591920858e-05,
+      "loss": 3.4556,
       "step": 744
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.000950123660346249,
-      "loss": 7.4034,
+      "learning_rate": 9.495053586150041e-05,
+      "loss": 3.5009,
       "step": 745
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009499175597691674,
-      "loss": 5.7294,
+      "learning_rate": 9.492992580379226e-05,
+      "loss": 2.7513,
       "step": 746
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009497114591920858,
-      "loss": 5.9461,
+      "learning_rate": 9.49093157460841e-05,
+      "loss": 2.9825,
       "step": 747
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009495053586150041,
-      "loss": 8.9013,
+      "learning_rate": 9.488870568837593e-05,
+      "loss": 2.6713,
       "step": 748
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009492992580379225,
-      "loss": 5.4214,
+      "learning_rate": 9.486809563066777e-05,
+      "loss": 2.2714,
       "step": 749
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009490931574608409,
-      "loss": 5.6228,
+      "learning_rate": 9.48474855729596e-05,
+      "loss": 2.4971,
       "step": 750
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009488870568837594,
-      "loss": 6.6634,
+      "learning_rate": 9.482687551525145e-05,
+      "loss": 3.9118,
       "step": 751
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009486809563066778,
-      "loss": 5.9863,
+      "learning_rate": 9.480626545754329e-05,
+      "loss": 3.7359,
       "step": 752
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009484748557295961,
-      "loss": 5.6832,
+      "learning_rate": 9.478565539983513e-05,
+      "loss": 3.4583,
       "step": 753
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009482687551525145,
-      "loss": 6.0156,
+      "learning_rate": 9.476504534212696e-05,
+      "loss": 3.7454,
       "step": 754
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009480626545754329,
-      "loss": 5.986,
+      "learning_rate": 9.474443528441879e-05,
+      "loss": 3.6765,
       "step": 755
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009478565539983513,
-      "loss": 5.996,
+      "learning_rate": 9.472382522671064e-05,
+      "loss": 3.6645,
       "step": 756
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.0009476504534212697,
-      "loss": 5.9839,
+      "learning_rate": 9.470321516900248e-05,
+      "loss": 3.3929,
       "step": 757
     },
     {
       "epoch": 0.42,
-      "learning_rate": 0.000947444352844188,
-      "loss": 5.7588,
+      "learning_rate": 9.468260511129432e-05,
+      "loss": 3.3115,
       "step": 758
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009472382522671064,
-      "loss": 5.4257,
+      "learning_rate": 9.466199505358615e-05,
+      "loss": 3.6653,
       "step": 759
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009470321516900248,
-      "loss": 5.5268,
+      "learning_rate": 9.4641384995878e-05,
+      "loss": 3.3536,
       "step": 760
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009468260511129432,
-      "loss": 5.7849,
+      "learning_rate": 9.462077493816983e-05,
+      "loss": 3.5744,
       "step": 761
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009466199505358615,
-      "loss": 6.3809,
+      "learning_rate": 9.460016488046167e-05,
+      "loss": 3.2477,
       "step": 762
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009464138499587799,
-      "loss": 5.6375,
+      "learning_rate": 9.457955482275351e-05,
+      "loss": 3.4668,
       "step": 763
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009462077493816983,
-      "loss": 5.8323,
+      "learning_rate": 9.455894476504534e-05,
+      "loss": 3.2338,
       "step": 764
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009460016488046167,
-      "loss": 5.6393,
+      "learning_rate": 9.453833470733719e-05,
+      "loss": 3.1595,
       "step": 765
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009457955482275351,
-      "loss": 6.0319,
+      "learning_rate": 9.451772464962903e-05,
+      "loss": 3.3488,
       "step": 766
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009455894476504534,
-      "loss": 6.3085,
+      "learning_rate": 9.449711459192086e-05,
+      "loss": 3.5833,
       "step": 767
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009453833470733718,
-      "loss": 5.9428,
+      "learning_rate": 9.447650453421269e-05,
+      "loss": 3.2968,
       "step": 768
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009451772464962902,
-      "loss": 6.3527,
+      "learning_rate": 9.445589447650453e-05,
+      "loss": 3.8563,
       "step": 769
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009449711459192086,
-      "loss": 5.9388,
+      "learning_rate": 9.443528441879638e-05,
+      "loss": 3.5692,
       "step": 770
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009447650453421269,
-      "loss": 5.9063,
+      "learning_rate": 9.441467436108822e-05,
+      "loss": 3.2429,
       "step": 771
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009445589447650453,
-      "loss": 5.6981,
+      "learning_rate": 9.439406430338006e-05,
+      "loss": 3.1946,
       "step": 772
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009443528441879638,
-      "loss": 5.8204,
+      "learning_rate": 9.437345424567189e-05,
+      "loss": 3.249,
       "step": 773
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009441467436108822,
-      "loss": 6.3106,
+      "learning_rate": 9.435284418796374e-05,
+      "loss": 3.487,
       "step": 774
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009439406430338006,
-      "loss": 5.7092,
+      "learning_rate": 9.433223413025557e-05,
+      "loss": 3.5384,
       "step": 775
     },
     {
       "epoch": 0.43,
-      "learning_rate": 0.0009437345424567189,
-      "loss": 6.1343,
+      "learning_rate": 9.431162407254741e-05,
+      "loss": 3.3086,
       "step": 776
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009435284418796373,
-      "loss": 6.0211,
+      "learning_rate": 9.429101401483924e-05,
+      "loss": 3.3577,
       "step": 777
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009433223413025557,
-      "loss": 5.7363,
+      "learning_rate": 9.427040395713108e-05,
+      "loss": 3.4437,
       "step": 778
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009431162407254741,
-      "loss": 5.7902,
+      "learning_rate": 9.424979389942293e-05,
+      "loss": 3.2118,
       "step": 779
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009429101401483924,
-      "loss": 5.7293,
+      "learning_rate": 9.422918384171477e-05,
+      "loss": 3.3966,
       "step": 780
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009427040395713108,
-      "loss": 5.8983,
+      "learning_rate": 9.42085737840066e-05,
+      "loss": 3.2085,
       "step": 781
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009424979389942292,
-      "loss": 5.7028,
+      "learning_rate": 9.418796372629843e-05,
+      "loss": 3.3043,
       "step": 782
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009422918384171476,
-      "loss": 5.5494,
+      "learning_rate": 9.416735366859027e-05,
+      "loss": 2.9048,
       "step": 783
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.000942085737840066,
-      "loss": 5.6193,
+      "learning_rate": 9.414674361088212e-05,
+      "loss": 3.2445,
       "step": 784
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009418796372629843,
-      "loss": 5.9452,
+      "learning_rate": 9.412613355317396e-05,
+      "loss": 3.0468,
       "step": 785
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009416735366859027,
-      "loss": 6.301,
+      "learning_rate": 9.41055234954658e-05,
+      "loss": 3.4235,
       "step": 786
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009414674361088211,
-      "loss": 5.9136,
+      "learning_rate": 9.408491343775763e-05,
+      "loss": 3.2875,
       "step": 787
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009412613355317395,
-      "loss": 5.9246,
+      "learning_rate": 9.406430338004946e-05,
+      "loss": 3.394,
       "step": 788
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009410552349546579,
-      "loss": 6.4458,
+      "learning_rate": 9.40436933223413e-05,
+      "loss": 3.7793,
       "step": 789
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009408491343775762,
-      "loss": 6.2647,
+      "learning_rate": 9.402308326463315e-05,
+      "loss": 3.3472,
       "step": 790
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009406430338004946,
-      "loss": 5.8964,
+      "learning_rate": 9.400247320692498e-05,
+      "loss": 3.3987,
       "step": 791
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.000940436933223413,
-      "loss": 5.638,
+      "learning_rate": 9.398186314921682e-05,
+      "loss": 3.1164,
       "step": 792
     },
     {
       "epoch": 0.44,
-      "learning_rate": 0.0009402308326463315,
-      "loss": 5.8369,
+      "learning_rate": 9.396125309150866e-05,
+      "loss": 2.9103,
       "step": 793
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009400247320692498,
-      "loss": 5.8889,
+      "learning_rate": 9.394064303380051e-05,
+      "loss": 3.0303,
       "step": 794
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009398186314921682,
-      "loss": 6.2636,
+      "learning_rate": 9.392003297609234e-05,
+      "loss": 2.9989,
       "step": 795
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009396125309150866,
-      "loss": 7.3261,
+      "learning_rate": 9.389942291838417e-05,
+      "loss": 3.0285,
       "step": 796
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.000939406430338005,
-      "loss": 6.69,
+      "learning_rate": 9.387881286067601e-05,
+      "loss": 2.5637,
       "step": 797
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009392003297609234,
-      "loss": 5.7478,
+      "learning_rate": 9.385820280296785e-05,
+      "loss": 2.495,
       "step": 798
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009389942291838417,
-      "loss": 5.9056,
+      "learning_rate": 9.38375927452597e-05,
+      "loss": 2.6581,
       "step": 799
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009387881286067601,
-      "loss": 5.7171,
+      "learning_rate": 9.381698268755153e-05,
+      "loss": 2.2975,
       "step": 800
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009385820280296785,
-      "loss": 7.3809,
+      "learning_rate": 9.379637262984337e-05,
+      "loss": 3.8789,
       "step": 801
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009383759274525969,
-      "loss": 5.8822,
+      "learning_rate": 9.37757625721352e-05,
+      "loss": 3.6901,
       "step": 802
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009381698268755152,
-      "loss": 6.8818,
+      "learning_rate": 9.375515251442704e-05,
+      "loss": 3.7884,
       "step": 803
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009379637262984336,
-      "loss": 5.5852,
+      "learning_rate": 9.373454245671889e-05,
+      "loss": 3.4235,
       "step": 804
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.000937757625721352,
-      "loss": 7.3812,
+      "learning_rate": 9.371393239901072e-05,
+      "loss": 3.5482,
       "step": 805
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009375515251442704,
-      "loss": 6.2432,
+      "learning_rate": 9.369332234130256e-05,
+      "loss": 3.4869,
       "step": 806
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009373454245671888,
-      "loss": 5.7977,
+      "learning_rate": 9.36727122835944e-05,
+      "loss": 3.3789,
       "step": 807
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009371393239901071,
-      "loss": 5.4737,
+      "learning_rate": 9.365210222588623e-05,
+      "loss": 3.2205,
       "step": 808
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009369332234130255,
-      "loss": 10.2999,
+      "learning_rate": 9.363149216817806e-05,
+      "loss": 3.9502,
       "step": 809
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009367271228359439,
-      "loss": 6.8698,
+      "learning_rate": 9.361088211046991e-05,
+      "loss": 3.8815,
       "step": 810
     },
     {
       "epoch": 0.45,
-      "learning_rate": 0.0009365210222588623,
-      "loss": 5.6809,
+      "learning_rate": 9.359027205276175e-05,
+      "loss": 3.6275,
       "step": 811
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009363149216817806,
-      "loss": 6.2355,
+      "learning_rate": 9.35696619950536e-05,
+      "loss": 3.5267,
       "step": 812
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.000936108821104699,
-      "loss": 6.2455,
+      "learning_rate": 9.354905193734544e-05,
+      "loss": 3.6031,
       "step": 813
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009359027205276175,
-      "loss": 5.5169,
+      "learning_rate": 9.352844187963727e-05,
+      "loss": 3.1118,
       "step": 814
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009356966199505359,
-      "loss": 5.7886,
+      "learning_rate": 9.35078318219291e-05,
+      "loss": 3.3453,
       "step": 815
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009354905193734543,
-      "loss": 5.8093,
+      "learning_rate": 9.348722176422094e-05,
+      "loss": 3.4692,
       "step": 816
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009352844187963726,
-      "loss": 6.4745,
+      "learning_rate": 9.346661170651278e-05,
+      "loss": 3.1591,
       "step": 817
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.000935078318219291,
-      "loss": 5.9947,
+      "learning_rate": 9.344600164880463e-05,
+      "loss": 3.8063,
       "step": 818
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009348722176422094,
-      "loss": 5.7777,
+      "learning_rate": 9.342539159109646e-05,
+      "loss": 3.1555,
       "step": 819
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009346661170651278,
-      "loss": 5.8253,
+      "learning_rate": 9.34047815333883e-05,
+      "loss": 3.5492,
       "step": 820
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009344600164880462,
-      "loss": 5.978,
+      "learning_rate": 9.338417147568014e-05,
+      "loss": 3.4635,
       "step": 821
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009342539159109645,
-      "loss": 5.7328,
+      "learning_rate": 9.336356141797197e-05,
+      "loss": 3.588,
       "step": 822
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009340478153338829,
-      "loss": 7.2603,
+      "learning_rate": 9.33429513602638e-05,
+      "loss": 3.6487,
       "step": 823
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009338417147568013,
-      "loss": 5.9169,
+      "learning_rate": 9.332234130255565e-05,
+      "loss": 3.3559,
       "step": 824
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009336356141797197,
-      "loss": 6.128,
+      "learning_rate": 9.330173124484749e-05,
+      "loss": 3.5244,
       "step": 825
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.000933429513602638,
-      "loss": 5.7228,
+      "learning_rate": 9.328112118713933e-05,
+      "loss": 3.2988,
       "step": 826
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009332234130255564,
-      "loss": 6.1841,
+      "learning_rate": 9.326051112943118e-05,
+      "loss": 3.5475,
       "step": 827
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009330173124484748,
-      "loss": 6.1751,
+      "learning_rate": 9.323990107172301e-05,
+      "loss": 3.5077,
       "step": 828
     },
     {
       "epoch": 0.46,
-      "learning_rate": 0.0009328112118713932,
-      "loss": 6.1,
+      "learning_rate": 9.321929101401484e-05,
+      "loss": 3.5652,
       "step": 829
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009326051112943117,
-      "loss": 5.7972,
+      "learning_rate": 9.319868095630668e-05,
+      "loss": 3.4221,
       "step": 830
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.00093239901071723,
-      "loss": 6.0542,
+      "learning_rate": 9.317807089859852e-05,
+      "loss": 3.7942,
       "step": 831
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009321929101401484,
-      "loss": 6.0632,
+      "learning_rate": 9.315746084089035e-05,
+      "loss": 3.6429,
       "step": 832
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009319868095630669,
-      "loss": 5.9715,
+      "learning_rate": 9.31368507831822e-05,
+      "loss": 3.353,
       "step": 833
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009317807089859853,
-      "loss": 5.2951,
+      "learning_rate": 9.311624072547404e-05,
+      "loss": 3.172,
       "step": 834
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009315746084089036,
-      "loss": 5.4006,
+      "learning_rate": 9.309563066776587e-05,
+      "loss": 3.3052,
       "step": 835
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.000931368507831822,
-      "loss": 5.7872,
+      "learning_rate": 9.307502061005771e-05,
+      "loss": 3.2143,
       "step": 836
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009311624072547404,
-      "loss": 5.9822,
+      "learning_rate": 9.305441055234954e-05,
+      "loss": 3.1466,
       "step": 837
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009309563066776588,
-      "loss": 5.4688,
+      "learning_rate": 9.303380049464139e-05,
+      "loss": 3.0927,
       "step": 838
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009307502061005772,
-      "loss": 6.0008,
+      "learning_rate": 9.301319043693323e-05,
+      "loss": 2.845,
       "step": 839
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009305441055234955,
-      "loss": 6.3812,
+      "learning_rate": 9.299258037922507e-05,
+      "loss": 3.1799,
       "step": 840
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009303380049464139,
-      "loss": 7.813,
+      "learning_rate": 9.297197032151692e-05,
+      "loss": 3.3647,
       "step": 841
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009301319043693323,
-      "loss": 5.4515,
+      "learning_rate": 9.295136026380875e-05,
+      "loss": 3.1602,
       "step": 842
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009299258037922507,
-      "loss": 6.1169,
+      "learning_rate": 9.293075020610058e-05,
+      "loss": 3.4143,
       "step": 843
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009297197032151691,
-      "loss": 6.059,
+      "learning_rate": 9.291014014839242e-05,
+      "loss": 3.0437,
       "step": 844
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009295136026380874,
-      "loss": 5.5557,
+      "learning_rate": 9.288953009068426e-05,
+      "loss": 2.9555,
       "step": 845
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009293075020610058,
-      "loss": 5.468,
+      "learning_rate": 9.286892003297609e-05,
+      "loss": 2.7015,
       "step": 846
     },
     {
       "epoch": 0.47,
-      "learning_rate": 0.0009291014014839242,
-      "loss": 5.4248,
+      "learning_rate": 9.284830997526794e-05,
+      "loss": 2.3103,
       "step": 847
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009288953009068426,
-      "loss": 6.0027,
+      "learning_rate": 9.282769991755978e-05,
+      "loss": 2.781,
       "step": 848
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009286892003297609,
-      "loss": 5.6318,
+      "learning_rate": 9.280708985985161e-05,
+      "loss": 2.719,
       "step": 849
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009284830997526793,
-      "loss": 9.163,
+      "learning_rate": 9.278647980214345e-05,
+      "loss": 2.7673,
       "step": 850
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009282769991755977,
-      "loss": 5.8012,
+      "learning_rate": 9.276586974443528e-05,
+      "loss": 3.7928,
       "step": 851
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009280708985985161,
-      "loss": 5.7268,
+      "learning_rate": 9.274525968672713e-05,
+      "loss": 3.6403,
       "step": 852
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009278647980214346,
-      "loss": 6.8307,
+      "learning_rate": 9.272464962901897e-05,
+      "loss": 3.3394,
       "step": 853
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009276586974443529,
-      "loss": 5.7482,
+      "learning_rate": 9.270403957131081e-05,
+      "loss": 3.6337,
       "step": 854
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009274525968672713,
-      "loss": 5.7373,
+      "learning_rate": 9.268342951360264e-05,
+      "loss": 3.7976,
       "step": 855
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009272464962901897,
-      "loss": 6.0101,
+      "learning_rate": 9.266281945589447e-05,
+      "loss": 3.38,
       "step": 856
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009270403957131081,
-      "loss": 5.6004,
+      "learning_rate": 9.264220939818632e-05,
+      "loss": 3.4907,
       "step": 857
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009268342951360264,
-      "loss": 5.7602,
+      "learning_rate": 9.262159934047816e-05,
+      "loss": 3.5892,
       "step": 858
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009266281945589448,
-      "loss": 5.771,
+      "learning_rate": 9.260098928277e-05,
+      "loss": 3.2856,
       "step": 859
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009264220939818632,
-      "loss": 6.2244,
+      "learning_rate": 9.258037922506183e-05,
+      "loss": 3.6338,
       "step": 860
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009262159934047816,
-      "loss": 5.3677,
+      "learning_rate": 9.255976916735368e-05,
+      "loss": 3.1704,
       "step": 861
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009260098928277,
-      "loss": 5.4577,
+      "learning_rate": 9.25391591096455e-05,
+      "loss": 3.5372,
       "step": 862
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009258037922506183,
-      "loss": 6.3185,
+      "learning_rate": 9.251854905193735e-05,
+      "loss": 3.3983,
       "step": 863
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009255976916735367,
-      "loss": 5.4712,
+      "learning_rate": 9.249793899422918e-05,
+      "loss": 3.1698,
       "step": 864
     },
     {
       "epoch": 0.48,
-      "learning_rate": 0.0009253915910964551,
-      "loss": 5.9647,
+      "learning_rate": 9.247732893652102e-05,
+      "loss": 3.3441,
       "step": 865
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009251854905193735,
-      "loss": 5.6616,
+      "learning_rate": 9.245671887881287e-05,
+      "loss": 3.368,
       "step": 866
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009249793899422918,
-      "loss": 5.525,
+      "learning_rate": 9.243610882110471e-05,
+      "loss": 3.3155,
       "step": 867
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009247732893652102,
-      "loss": 5.972,
+      "learning_rate": 9.241549876339655e-05,
+      "loss": 3.6475,
       "step": 868
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009245671887881286,
-      "loss": 5.5719,
+      "learning_rate": 9.239488870568838e-05,
+      "loss": 3.0096,
       "step": 869
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.000924361088211047,
-      "loss": 5.5685,
+      "learning_rate": 9.237427864798021e-05,
+      "loss": 3.3112,
       "step": 870
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009241549876339654,
-      "loss": 5.5242,
+      "learning_rate": 9.235366859027206e-05,
+      "loss": 3.1641,
       "step": 871
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009239488870568837,
-      "loss": 5.6932,
+      "learning_rate": 9.23330585325639e-05,
+      "loss": 3.35,
       "step": 872
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009237427864798021,
-      "loss": 5.6847,
+      "learning_rate": 9.231244847485574e-05,
+      "loss": 3.2469,
       "step": 873
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009235366859027206,
-      "loss": 5.6326,
+      "learning_rate": 9.229183841714757e-05,
+      "loss": 3.1721,
       "step": 874
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.000923330585325639,
-      "loss": 5.8906,
+      "learning_rate": 9.227122835943942e-05,
+      "loss": 3.43,
       "step": 875
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009231244847485574,
-      "loss": 5.7659,
+      "learning_rate": 9.225061830173125e-05,
+      "loss": 3.0988,
       "step": 876
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009229183841714757,
-      "loss": 6.3473,
+      "learning_rate": 9.223000824402309e-05,
+      "loss": 3.397,
       "step": 877
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009227122835943941,
-      "loss": 5.5877,
+      "learning_rate": 9.220939818631492e-05,
+      "loss": 3.1963,
       "step": 878
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009225061830173125,
-      "loss": 5.4683,
+      "learning_rate": 9.218878812860676e-05,
+      "loss": 3.0762,
       "step": 879
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009223000824402309,
-      "loss": 5.643,
+      "learning_rate": 9.21681780708986e-05,
+      "loss": 3.1986,
       "step": 880
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009220939818631492,
-      "loss": 5.8368,
+      "learning_rate": 9.214756801319045e-05,
+      "loss": 3.3929,
       "step": 881
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.0009218878812860676,
-      "loss": 5.3025,
+      "learning_rate": 9.212695795548228e-05,
+      "loss": 3.0023,
       "step": 882
     },
     {
       "epoch": 0.49,
-      "learning_rate": 0.000921681780708986,
-      "loss": 5.9461,
+      "learning_rate": 9.210634789777411e-05,
+      "loss": 3.3198,
       "step": 883
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009214756801319044,
-      "loss": 7.3355,
+      "learning_rate": 9.208573784006595e-05,
+      "loss": 3.3553,
       "step": 884
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009212695795548228,
-      "loss": 6.2517,
+      "learning_rate": 9.20651277823578e-05,
+      "loss": 3.2671,
       "step": 885
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009210634789777411,
-      "loss": 5.9753,
+      "learning_rate": 9.204451772464964e-05,
+      "loss": 3.1548,
       "step": 886
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009208573784006595,
-      "loss": 6.046,
+      "learning_rate": 9.202390766694147e-05,
+      "loss": 3.2787,
       "step": 887
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009206512778235779,
-      "loss": 5.4745,
+      "learning_rate": 9.200329760923331e-05,
+      "loss": 3.0068,
       "step": 888
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009204451772464963,
-      "loss": 5.7831,
+      "learning_rate": 9.198268755152516e-05,
+      "loss": 3.0546,
       "step": 889
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009202390766694146,
-      "loss": 6.2167,
+      "learning_rate": 9.196207749381698e-05,
+      "loss": 3.1841,
       "step": 890
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.000920032976092333,
-      "loss": 5.7482,
+      "learning_rate": 9.194146743610883e-05,
+      "loss": 3.2482,
       "step": 891
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009198268755152514,
-      "loss": 5.7933,
+      "learning_rate": 9.192085737840066e-05,
+      "loss": 3.0538,
       "step": 892
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009196207749381698,
-      "loss": 6.4929,
+      "learning_rate": 9.19002473206925e-05,
+      "loss": 3.1881,
       "step": 893
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009194146743610883,
-      "loss": 5.5808,
+      "learning_rate": 9.187963726298434e-05,
+      "loss": 2.8013,
       "step": 894
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009192085737840066,
-      "loss": 5.5091,
+      "learning_rate": 9.185902720527619e-05,
+      "loss": 2.6844,
       "step": 895
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.000919002473206925,
-      "loss": 5.7982,
+      "learning_rate": 9.183841714756802e-05,
+      "loss": 2.9867,
       "step": 896
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009187963726298434,
-      "loss": 5.6103,
+      "learning_rate": 9.181780708985985e-05,
+      "loss": 2.7275,
       "step": 897
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009185902720527618,
-      "loss": 7.3187,
+      "learning_rate": 9.179719703215169e-05,
+      "loss": 2.6466,
       "step": 898
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009183841714756801,
-      "loss": 6.7444,
+      "learning_rate": 9.177658697444353e-05,
+      "loss": 2.7976,
       "step": 899
     },
     {
       "epoch": 0.5,
-      "learning_rate": 0.0009181780708985985,
-      "loss": 7.599,
+      "learning_rate": 9.175597691673538e-05,
+      "loss": 2.9838,
       "step": 900
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009179719703215169,
-      "loss": 5.7065,
+      "learning_rate": 9.173536685902721e-05,
+      "loss": 3.838,
       "step": 901
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009177658697444353,
-      "loss": 5.9893,
+      "learning_rate": 9.171475680131905e-05,
+      "loss": 3.7071,
       "step": 902
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009175597691673537,
-      "loss": 6.0056,
+      "learning_rate": 9.169414674361088e-05,
+      "loss": 3.6604,
       "step": 903
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.000917353668590272,
-      "loss": 5.8115,
+      "learning_rate": 9.167353668590272e-05,
+      "loss": 3.3526,
       "step": 904
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009171475680131904,
-      "loss": 5.9716,
+      "learning_rate": 9.165292662819457e-05,
+      "loss": 3.7185,
       "step": 905
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009169414674361088,
-      "loss": 5.988,
+      "learning_rate": 9.16323165704864e-05,
+      "loss": 3.6342,
       "step": 906
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009167353668590272,
-      "loss": 6.2609,
+      "learning_rate": 9.161170651277824e-05,
+      "loss": 3.5777,
       "step": 907
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009165292662819456,
-      "loss": 5.4386,
+      "learning_rate": 9.159109645507008e-05,
+      "loss": 3.4302,
       "step": 908
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009163231657048639,
-      "loss": 6.1184,
+      "learning_rate": 9.157048639736191e-05,
+      "loss": 3.696,
       "step": 909
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009161170651277823,
-      "loss": 5.5782,
+      "learning_rate": 9.154987633965374e-05,
+      "loss": 3.3064,
       "step": 910
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009159109645507007,
-      "loss": 6.4379,
+      "learning_rate": 9.152926628194559e-05,
+      "loss": 3.5066,
       "step": 911
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009157048639736191,
-      "loss": 5.57,
+      "learning_rate": 9.150865622423743e-05,
+      "loss": 3.1913,
       "step": 912
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009154987633965374,
-      "loss": 6.7615,
+      "learning_rate": 9.148804616652927e-05,
+      "loss": 3.2263,
       "step": 913
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009152926628194558,
-      "loss": 5.6496,
+      "learning_rate": 9.146743610882112e-05,
+      "loss": 3.7163,
       "step": 914
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009150865622423743,
-      "loss": 6.6595,
+      "learning_rate": 9.144682605111295e-05,
+      "loss": 3.3963,
       "step": 915
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009148804616652927,
-      "loss": 5.81,
+      "learning_rate": 9.142621599340479e-05,
+      "loss": 3.5391,
       "step": 916
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009146743610882111,
-      "loss": 5.5627,
+      "learning_rate": 9.140560593569662e-05,
+      "loss": 3.1392,
       "step": 917
     },
     {
       "epoch": 0.51,
-      "learning_rate": 0.0009144682605111294,
-      "loss": 6.9672,
+      "learning_rate": 9.138499587798846e-05,
+      "loss": 3.3801,
       "step": 918
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009142621599340478,
-      "loss": 6.2755,
+      "learning_rate": 9.13643858202803e-05,
+      "loss": 3.4323,
       "step": 919
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009140560593569662,
-      "loss": 5.3061,
+      "learning_rate": 9.134377576257214e-05,
+      "loss": 3.2242,
       "step": 920
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009138499587798846,
-      "loss": 5.728,
+      "learning_rate": 9.132316570486398e-05,
+      "loss": 3.5483,
       "step": 921
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009136438582028029,
-      "loss": 5.9284,
+      "learning_rate": 9.130255564715582e-05,
+      "loss": 3.2557,
       "step": 922
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009134377576257213,
-      "loss": 5.4885,
+      "learning_rate": 9.128194558944765e-05,
+      "loss": 3.4928,
       "step": 923
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009132316570486398,
-      "loss": 6.7174,
+      "learning_rate": 9.126133553173948e-05,
+      "loss": 3.584,
       "step": 924
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009130255564715582,
-      "loss": 5.909,
+      "learning_rate": 9.124072547403133e-05,
+      "loss": 3.4436,
       "step": 925
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009128194558944766,
-      "loss": 6.3805,
+      "learning_rate": 9.122011541632317e-05,
+      "loss": 3.3985,
       "step": 926
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009126133553173949,
-      "loss": 5.4519,
+      "learning_rate": 9.119950535861501e-05,
+      "loss": 3.0288,
       "step": 927
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009124072547403133,
-      "loss": 5.5004,
+      "learning_rate": 9.117889530090684e-05,
+      "loss": 3.403,
       "step": 928
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009122011541632317,
-      "loss": 5.3908,
+      "learning_rate": 9.115828524319869e-05,
+      "loss": 3.2132,
       "step": 929
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009119950535861501,
-      "loss": 5.737,
+      "learning_rate": 9.113767518549052e-05,
+      "loss": 3.8091,
       "step": 930
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009117889530090684,
-      "loss": 5.9619,
+      "learning_rate": 9.111706512778236e-05,
+      "loss": 3.4118,
       "step": 931
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009115828524319868,
-      "loss": 5.8696,
+      "learning_rate": 9.10964550700742e-05,
+      "loss": 3.0655,
       "step": 932
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009113767518549052,
-      "loss": 6.2842,
+      "learning_rate": 9.107584501236603e-05,
+      "loss": 3.4295,
       "step": 933
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009111706512778237,
-      "loss": 5.6991,
+      "learning_rate": 9.105523495465788e-05,
+      "loss": 2.9979,
       "step": 934
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009109645507007421,
-      "loss": 5.9549,
+      "learning_rate": 9.103462489694972e-05,
+      "loss": 3.3802,
       "step": 935
     },
     {
       "epoch": 0.52,
-      "learning_rate": 0.0009107584501236604,
-      "loss": 5.7849,
+      "learning_rate": 9.101401483924156e-05,
+      "loss": 3.2158,
       "step": 936
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009105523495465788,
-      "loss": 5.6284,
+      "learning_rate": 9.09934047815334e-05,
+      "loss": 2.9365,
       "step": 937
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009103462489694972,
-      "loss": 5.375,
+      "learning_rate": 9.097279472382522e-05,
+      "loss": 2.7736,
       "step": 938
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009101401483924156,
-      "loss": 5.2918,
+      "learning_rate": 9.095218466611707e-05,
+      "loss": 3.0588,
       "step": 939
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.000909934047815334,
-      "loss": 5.8196,
+      "learning_rate": 9.093157460840891e-05,
+      "loss": 3.1984,
       "step": 940
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009097279472382523,
-      "loss": 5.8173,
+      "learning_rate": 9.091096455070075e-05,
+      "loss": 3.6396,
       "step": 941
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009095218466611707,
-      "loss": 5.5793,
+      "learning_rate": 9.089035449299258e-05,
+      "loss": 2.939,
       "step": 942
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009093157460840891,
-      "loss": 6.2229,
+      "learning_rate": 9.086974443528443e-05,
+      "loss": 3.1641,
       "step": 943
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009091096455070075,
-      "loss": 5.5593,
+      "learning_rate": 9.084913437757626e-05,
+      "loss": 2.5622,
       "step": 944
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009089035449299258,
-      "loss": 5.9367,
+      "learning_rate": 9.08285243198681e-05,
+      "loss": 2.6021,
       "step": 945
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009086974443528442,
-      "loss": 5.749,
+      "learning_rate": 9.080791426215994e-05,
+      "loss": 2.9091,
       "step": 946
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009084913437757626,
-      "loss": 5.2255,
+      "learning_rate": 9.078730420445177e-05,
+      "loss": 2.4744,
       "step": 947
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.000908285243198681,
-      "loss": 5.6387,
+      "learning_rate": 9.076669414674362e-05,
+      "loss": 2.3444,
       "step": 948
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009080791426215994,
-      "loss": 7.885,
+      "learning_rate": 9.074608408903546e-05,
+      "loss": 2.9281,
       "step": 949
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009078730420445177,
-      "loss": 4.9916,
+      "learning_rate": 9.072547403132729e-05,
+      "loss": 2.2496,
       "step": 950
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009076669414674361,
-      "loss": 5.6046,
+      "learning_rate": 9.070486397361912e-05,
+      "loss": 3.6247,
       "step": 951
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009074608408903545,
-      "loss": 5.6585,
+      "learning_rate": 9.068425391591096e-05,
+      "loss": 3.4801,
       "step": 952
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.000907254740313273,
-      "loss": 5.683,
+      "learning_rate": 9.06636438582028e-05,
+      "loss": 3.531,
       "step": 953
     },
     {
       "epoch": 0.53,
-      "learning_rate": 0.0009070486397361912,
-      "loss": 5.2602,
+      "learning_rate": 9.064303380049465e-05,
+      "loss": 3.189,
       "step": 954
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009068425391591097,
-      "loss": 5.645,
+      "learning_rate": 9.062242374278649e-05,
+      "loss": 3.3838,
       "step": 955
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009066364385820281,
-      "loss": 5.4953,
+      "learning_rate": 9.060181368507832e-05,
+      "loss": 3.0949,
       "step": 956
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009064303380049465,
-      "loss": 5.6824,
+      "learning_rate": 9.058120362737015e-05,
+      "loss": 3.4329,
       "step": 957
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009062242374278649,
-      "loss": 5.3621,
+      "learning_rate": 9.0560593569662e-05,
+      "loss": 3.2028,
       "step": 958
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009060181368507832,
-      "loss": 6.2408,
+      "learning_rate": 9.053998351195384e-05,
+      "loss": 3.724,
       "step": 959
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009058120362737016,
-      "loss": 5.3227,
+      "learning_rate": 9.051937345424568e-05,
+      "loss": 3.0876,
       "step": 960
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.00090560593569662,
-      "loss": 5.3431,
+      "learning_rate": 9.049876339653751e-05,
+      "loss": 3.2367,
       "step": 961
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009053998351195384,
-      "loss": 5.5201,
+      "learning_rate": 9.047815333882936e-05,
+      "loss": 3.2578,
       "step": 962
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009051937345424568,
-      "loss": 5.9274,
+      "learning_rate": 9.04575432811212e-05,
+      "loss": 3.4783,
       "step": 963
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009049876339653751,
-      "loss": 5.6072,
+      "learning_rate": 9.043693322341303e-05,
+      "loss": 3.2374,
       "step": 964
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009047815333882935,
-      "loss": 5.419,
+      "learning_rate": 9.041632316570486e-05,
+      "loss": 3.3835,
       "step": 965
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009045754328112119,
-      "loss": 5.424,
+      "learning_rate": 9.03957131079967e-05,
+      "loss": 3.1274,
       "step": 966
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009043693322341303,
-      "loss": 5.7684,
+      "learning_rate": 9.037510305028855e-05,
+      "loss": 3.4147,
       "step": 967
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009041632316570486,
-      "loss": 5.5606,
+      "learning_rate": 9.035449299258039e-05,
+      "loss": 3.169,
       "step": 968
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.000903957131079967,
-      "loss": 5.6427,
+      "learning_rate": 9.033388293487223e-05,
+      "loss": 3.3618,
       "step": 969
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009037510305028854,
-      "loss": 5.5729,
+      "learning_rate": 9.031327287716406e-05,
+      "loss": 3.4292,
       "step": 970
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009035449299258038,
-      "loss": 6.3085,
+      "learning_rate": 9.029266281945589e-05,
+      "loss": 3.7134,
       "step": 971
     },
     {
       "epoch": 0.54,
-      "learning_rate": 0.0009033388293487222,
-      "loss": 5.432,
+      "learning_rate": 9.027205276174774e-05,
+      "loss": 3.2785,
       "step": 972
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009031327287716405,
-      "loss": 5.7285,
+      "learning_rate": 9.025144270403958e-05,
+      "loss": 3.6726,
       "step": 973
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.000902926628194559,
-      "loss": 6.4965,
+      "learning_rate": 9.023083264633141e-05,
+      "loss": 3.7698,
       "step": 974
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009027205276174774,
-      "loss": 5.6621,
+      "learning_rate": 9.021022258862325e-05,
+      "loss": 3.271,
       "step": 975
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009025144270403958,
-      "loss": 5.7753,
+      "learning_rate": 9.01896125309151e-05,
+      "loss": 3.5855,
       "step": 976
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009023083264633141,
-      "loss": 6.1264,
+      "learning_rate": 9.016900247320693e-05,
+      "loss": 3.5439,
       "step": 977
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009021022258862325,
-      "loss": 5.5589,
+      "learning_rate": 9.014839241549877e-05,
+      "loss": 2.985,
       "step": 978
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009018961253091509,
-      "loss": 6.0841,
+      "learning_rate": 9.01277823577906e-05,
+      "loss": 3.3217,
       "step": 979
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009016900247320693,
-      "loss": 5.5604,
+      "learning_rate": 9.010717230008244e-05,
+      "loss": 3.4783,
       "step": 980
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009014839241549877,
-      "loss": 5.5459,
+      "learning_rate": 9.008656224237429e-05,
+      "loss": 3.2106,
       "step": 981
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.000901277823577906,
-      "loss": 5.2837,
+      "learning_rate": 9.006595218466613e-05,
+      "loss": 3.053,
       "step": 982
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009010717230008244,
-      "loss": 6.0762,
+      "learning_rate": 9.004534212695796e-05,
+      "loss": 3.5851,
       "step": 983
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009008656224237428,
-      "loss": 5.8828,
+      "learning_rate": 9.00247320692498e-05,
+      "loss": 3.499,
       "step": 984
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009006595218466612,
-      "loss": 5.3435,
+      "learning_rate": 9.000412201154163e-05,
+      "loss": 3.2667,
       "step": 985
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009004534212695795,
-      "loss": 6.0708,
+      "learning_rate": 8.998351195383347e-05,
+      "loss": 2.9846,
       "step": 986
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009002473206924979,
-      "loss": 6.2976,
+      "learning_rate": 8.996290189612532e-05,
+      "loss": 3.2223,
       "step": 987
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0009000412201154163,
-      "loss": 5.588,
+      "learning_rate": 8.994229183841715e-05,
+      "loss": 3.0042,
       "step": 988
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0008998351195383347,
-      "loss": 5.9371,
+      "learning_rate": 8.992168178070899e-05,
+      "loss": 3.0108,
       "step": 989
     },
     {
       "epoch": 0.55,
-      "learning_rate": 0.0008996290189612531,
-      "loss": 5.5429,
+      "learning_rate": 8.990107172300083e-05,
+      "loss": 3.1631,
       "step": 990
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008994229183841714,
-      "loss": 5.7105,
+      "learning_rate": 8.988046166529266e-05,
+      "loss": 3.2895,
       "step": 991
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008992168178070898,
-      "loss": 5.9454,
+      "learning_rate": 8.985985160758451e-05,
+      "loss": 3.0456,
       "step": 992
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008990107172300082,
-      "loss": 5.7403,
+      "learning_rate": 8.983924154987634e-05,
+      "loss": 2.9358,
       "step": 993
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008988046166529266,
-      "loss": 6.7284,
+      "learning_rate": 8.981863149216818e-05,
+      "loss": 2.9285,
       "step": 994
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.000898598516075845,
-      "loss": 5.5366,
+      "learning_rate": 8.979802143446002e-05,
+      "loss": 2.5571,
       "step": 995
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008983924154987634,
-      "loss": 4.9742,
+      "learning_rate": 8.977741137675187e-05,
+      "loss": 2.219,
       "step": 996
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008981863149216818,
-      "loss": 5.5503,
+      "learning_rate": 8.97568013190437e-05,
+      "loss": 2.2913,
       "step": 997
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008979802143446002,
-      "loss": 6.2854,
+      "learning_rate": 8.973619126133553e-05,
+      "loss": 2.4287,
       "step": 998
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008977741137675186,
-      "loss": 6.0931,
+      "learning_rate": 8.971558120362737e-05,
+      "loss": 2.8476,
       "step": 999
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008975680131904369,
-      "loss": 5.609,
+      "learning_rate": 8.969497114591921e-05,
+      "loss": 1.941,
       "step": 1000
     },
     {
       "epoch": 0.56,
-      "eval_loss": 6.030221462249756,
-      "eval_runtime": 1248.2223,
-      "eval_samples_per_second": 2.117,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.971717572391908,
+      "eval_loss": 5.697424411773682,
+      "eval_runtime": 1234.2317,
+      "eval_samples_per_second": 2.141,
+      "eval_steps_per_second": 0.268,
+      "eval_wer": 1.9745934153113844,
       "step": 1000
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008973619126133553,
-      "loss": 6.0239,
+      "learning_rate": 8.967436108821106e-05,
+      "loss": 3.7661,
       "step": 1001
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008971558120362737,
-      "loss": 5.5556,
+      "learning_rate": 8.965375103050289e-05,
+      "loss": 3.5206,
       "step": 1002
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008969497114591921,
-      "loss": 5.6206,
+      "learning_rate": 8.963314097279473e-05,
+      "loss": 3.4685,
       "step": 1003
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008967436108821105,
-      "loss": 5.7808,
+      "learning_rate": 8.961253091508657e-05,
+      "loss": 3.1543,
       "step": 1004
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008965375103050288,
-      "loss": 5.7128,
+      "learning_rate": 8.95919208573784e-05,
+      "loss": 3.1543,
       "step": 1005
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008963314097279472,
-      "loss": 5.6755,
+      "learning_rate": 8.957131079967023e-05,
+      "loss": 3.4414,
       "step": 1006
     },
     {
       "epoch": 0.56,
-      "learning_rate": 0.0008961253091508656,
-      "loss": 6.3792,
+      "learning_rate": 8.955070074196208e-05,
+      "loss": 3.5235,
       "step": 1007
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.000895919208573784,
-      "loss": 7.4117,
+      "learning_rate": 8.953009068425392e-05,
+      "loss": 3.4979,
       "step": 1008
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008957131079967023,
-      "loss": 6.9529,
+      "learning_rate": 8.950948062654576e-05,
+      "loss": 3.5608,
       "step": 1009
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008955070074196207,
-      "loss": 5.8751,
+      "learning_rate": 8.948887056883761e-05,
+      "loss": 3.7816,
       "step": 1010
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008953009068425391,
-      "loss": 5.7386,
+      "learning_rate": 8.946826051112944e-05,
+      "loss": 3.3953,
       "step": 1011
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008950948062654575,
-      "loss": 5.8392,
+      "learning_rate": 8.944765045342127e-05,
+      "loss": 3.4447,
       "step": 1012
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008948887056883759,
-      "loss": 5.6897,
+      "learning_rate": 8.942704039571311e-05,
+      "loss": 3.4132,
       "step": 1013
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008946826051112942,
-      "loss": 5.6693,
+      "learning_rate": 8.940643033800495e-05,
+      "loss": 3.2093,
       "step": 1014
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008944765045342126,
-      "loss": 6.4492,
+      "learning_rate": 8.938582028029678e-05,
+      "loss": 3.9962,
       "step": 1015
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.000894270403957131,
-      "loss": 5.3406,
+      "learning_rate": 8.936521022258863e-05,
+      "loss": 3.1874,
       "step": 1016
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008940643033800496,
-      "loss": 5.6562,
+      "learning_rate": 8.934460016488047e-05,
+      "loss": 3.2628,
       "step": 1017
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008938582028029679,
-      "loss": 5.4921,
+      "learning_rate": 8.93239901071723e-05,
+      "loss": 3.3212,
       "step": 1018
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008936521022258863,
-      "loss": 6.6062,
+      "learning_rate": 8.930338004946414e-05,
+      "loss": 3.7217,
       "step": 1019
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008934460016488047,
-      "loss": 6.914,
+      "learning_rate": 8.928276999175597e-05,
+      "loss": 3.7309,
       "step": 1020
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008932399010717231,
-      "loss": 5.7628,
+      "learning_rate": 8.926215993404782e-05,
+      "loss": 3.445,
       "step": 1021
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008930338004946415,
-      "loss": 5.9083,
+      "learning_rate": 8.924154987633966e-05,
+      "loss": 3.2489,
       "step": 1022
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008928276999175598,
-      "loss": 5.6599,
+      "learning_rate": 8.92209398186315e-05,
+      "loss": 3.2354,
       "step": 1023
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008926215993404782,
-      "loss": 5.6024,
+      "learning_rate": 8.920032976092333e-05,
+      "loss": 3.0843,
       "step": 1024
     },
     {
       "epoch": 0.57,
-      "learning_rate": 0.0008924154987633966,
-      "loss": 6.2904,
+      "learning_rate": 8.917971970321516e-05,
+      "loss": 3.4829,
       "step": 1025
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.000892209398186315,
-      "loss": 5.6106,
+      "learning_rate": 8.915910964550701e-05,
+      "loss": 3.4818,
       "step": 1026
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008920032976092334,
-      "loss": 5.659,
+      "learning_rate": 8.913849958779885e-05,
+      "loss": 3.355,
       "step": 1027
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008917971970321517,
-      "loss": 5.8208,
+      "learning_rate": 8.91178895300907e-05,
+      "loss": 3.6192,
       "step": 1028
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008915910964550701,
-      "loss": 5.5091,
+      "learning_rate": 8.909727947238252e-05,
+      "loss": 3.1739,
       "step": 1029
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008913849958779885,
-      "loss": 5.752,
+      "learning_rate": 8.907666941467437e-05,
+      "loss": 3.2286,
       "step": 1030
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008911788953009069,
-      "loss": 5.7708,
+      "learning_rate": 8.905605935696621e-05,
+      "loss": 3.3742,
       "step": 1031
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008909727947238252,
-      "loss": 5.5729,
+      "learning_rate": 8.903544929925804e-05,
+      "loss": 3.0271,
       "step": 1032
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008907666941467436,
-      "loss": 5.5487,
+      "learning_rate": 8.901483924154988e-05,
+      "loss": 3.3232,
       "step": 1033
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.000890560593569662,
-      "loss": 5.3115,
+      "learning_rate": 8.899422918384171e-05,
+      "loss": 2.9305,
       "step": 1034
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008903544929925805,
-      "loss": 5.6185,
+      "learning_rate": 8.897361912613356e-05,
+      "loss": 3.4084,
       "step": 1035
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008901483924154989,
-      "loss": 6.773,
+      "learning_rate": 8.89530090684254e-05,
+      "loss": 3.4995,
       "step": 1036
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008899422918384172,
-      "loss": 5.5839,
+      "learning_rate": 8.893239901071724e-05,
+      "loss": 3.1686,
       "step": 1037
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008897361912613356,
-      "loss": 5.6614,
+      "learning_rate": 8.891178895300907e-05,
+      "loss": 2.8955,
       "step": 1038
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.000889530090684254,
-      "loss": 6.7267,
+      "learning_rate": 8.88911788953009e-05,
+      "loss": 3.4434,
       "step": 1039
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008893239901071724,
-      "loss": 6.3976,
+      "learning_rate": 8.887056883759275e-05,
+      "loss": 3.5838,
       "step": 1040
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008891178895300907,
-      "loss": 6.5246,
+      "learning_rate": 8.884995877988459e-05,
+      "loss": 3.4113,
       "step": 1041
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008889117889530091,
-      "loss": 5.6608,
+      "learning_rate": 8.882934872217643e-05,
+      "loss": 3.1229,
       "step": 1042
     },
     {
       "epoch": 0.58,
-      "learning_rate": 0.0008887056883759275,
-      "loss": 5.6547,
+      "learning_rate": 8.880873866446826e-05,
+      "loss": 2.7681,
       "step": 1043
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008884995877988459,
-      "loss": 7.4612,
+      "learning_rate": 8.87881286067601e-05,
+      "loss": 2.8454,
       "step": 1044
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008882934872217643,
-      "loss": 5.9616,
+      "learning_rate": 8.876751854905194e-05,
+      "loss": 2.9663,
       "step": 1045
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008880873866446826,
-      "loss": 5.6414,
+      "learning_rate": 8.874690849134378e-05,
+      "loss": 2.4622,
       "step": 1046
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.000887881286067601,
-      "loss": 5.7465,
+      "learning_rate": 8.872629843363561e-05,
+      "loss": 2.3872,
       "step": 1047
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008876751854905194,
-      "loss": 5.4455,
+      "learning_rate": 8.870568837592745e-05,
+      "loss": 2.6328,
       "step": 1048
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008874690849134378,
-      "loss": 5.9119,
+      "learning_rate": 8.86850783182193e-05,
+      "loss": 2.4866,
       "step": 1049
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008872629843363561,
-      "loss": 6.497,
+      "learning_rate": 8.866446826051114e-05,
+      "loss": 2.8767,
       "step": 1050
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008870568837592745,
-      "loss": 6.7875,
+      "learning_rate": 8.864385820280298e-05,
+      "loss": 4.2511,
       "step": 1051
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008868507831821929,
-      "loss": 5.4351,
+      "learning_rate": 8.862324814509481e-05,
+      "loss": 3.6358,
       "step": 1052
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008866446826051113,
-      "loss": 5.9467,
+      "learning_rate": 8.860263808738664e-05,
+      "loss": 3.3766,
       "step": 1053
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008864385820280297,
-      "loss": 5.3352,
+      "learning_rate": 8.858202802967849e-05,
+      "loss": 3.3501,
       "step": 1054
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.000886232481450948,
-      "loss": 5.9635,
+      "learning_rate": 8.856141797197033e-05,
+      "loss": 3.3732,
       "step": 1055
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008860263808738665,
-      "loss": 5.5758,
+      "learning_rate": 8.854080791426217e-05,
+      "loss": 3.5853,
       "step": 1056
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008858202802967849,
-      "loss": 5.929,
+      "learning_rate": 8.8520197856554e-05,
+      "loss": 3.4861,
       "step": 1057
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008856141797197033,
-      "loss": 5.502,
+      "learning_rate": 8.849958779884585e-05,
+      "loss": 3.551,
       "step": 1058
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008854080791426217,
-      "loss": 6.1336,
+      "learning_rate": 8.847897774113768e-05,
+      "loss": 3.8358,
       "step": 1059
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.00088520197856554,
-      "loss": 5.6216,
+      "learning_rate": 8.845836768342952e-05,
+      "loss": 3.6411,
       "step": 1060
     },
     {
       "epoch": 0.59,
-      "learning_rate": 0.0008849958779884584,
-      "loss": 5.7415,
+      "learning_rate": 8.843775762572135e-05,
+      "loss": 3.3941,
       "step": 1061
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008847897774113768,
-      "loss": 5.5518,
+      "learning_rate": 8.841714756801319e-05,
+      "loss": 3.2044,
       "step": 1062
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008845836768342952,
-      "loss": 6.3973,
+      "learning_rate": 8.839653751030504e-05,
+      "loss": 3.637,
       "step": 1063
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008843775762572135,
-      "loss": 5.2913,
+      "learning_rate": 8.837592745259688e-05,
+      "loss": 3.0184,
       "step": 1064
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008841714756801319,
-      "loss": 5.9452,
+      "learning_rate": 8.835531739488871e-05,
+      "loss": 3.5977,
       "step": 1065
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008839653751030503,
-      "loss": 5.3736,
+      "learning_rate": 8.833470733718054e-05,
+      "loss": 3.149,
       "step": 1066
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008837592745259687,
-      "loss": 5.7383,
+      "learning_rate": 8.831409727947238e-05,
+      "loss": 3.3731,
       "step": 1067
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008835531739488871,
-      "loss": 5.4393,
+      "learning_rate": 8.829348722176423e-05,
+      "loss": 3.0377,
       "step": 1068
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008833470733718054,
-      "loss": 5.8006,
+      "learning_rate": 8.827287716405607e-05,
+      "loss": 3.3289,
       "step": 1069
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008831409727947238,
-      "loss": 5.4522,
+      "learning_rate": 8.82522671063479e-05,
+      "loss": 3.1975,
       "step": 1070
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008829348722176422,
-      "loss": 5.5037,
+      "learning_rate": 8.823165704863974e-05,
+      "loss": 3.2073,
       "step": 1071
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008827287716405606,
-      "loss": 5.2662,
+      "learning_rate": 8.821104699093157e-05,
+      "loss": 3.0744,
       "step": 1072
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008825226710634789,
-      "loss": 5.4944,
+      "learning_rate": 8.819043693322342e-05,
+      "loss": 2.7886,
       "step": 1073
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008823165704863973,
-      "loss": 6.4176,
+      "learning_rate": 8.816982687551526e-05,
+      "loss": 3.522,
       "step": 1074
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008821104699093157,
-      "loss": 5.66,
+      "learning_rate": 8.814921681780709e-05,
+      "loss": 3.0449,
       "step": 1075
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008819043693322342,
-      "loss": 6.5868,
+      "learning_rate": 8.812860676009893e-05,
+      "loss": 3.4532,
       "step": 1076
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008816982687551526,
-      "loss": 5.6613,
+      "learning_rate": 8.810799670239078e-05,
+      "loss": 3.5137,
       "step": 1077
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008814921681780709,
-      "loss": 5.4285,
+      "learning_rate": 8.808738664468262e-05,
+      "loss": 3.5442,
       "step": 1078
     },
     {
       "epoch": 0.6,
-      "learning_rate": 0.0008812860676009893,
-      "loss": 5.5742,
+      "learning_rate": 8.806677658697445e-05,
+      "loss": 3.2347,
       "step": 1079
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008810799670239077,
-      "loss": 5.6072,
+      "learning_rate": 8.804616652926628e-05,
+      "loss": 3.0601,
       "step": 1080
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008808738664468261,
-      "loss": 6.9596,
+      "learning_rate": 8.802555647155812e-05,
+      "loss": 3.7674,
       "step": 1081
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008806677658697445,
-      "loss": 6.0076,
+      "learning_rate": 8.800494641384997e-05,
+      "loss": 3.4238,
       "step": 1082
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008804616652926628,
-      "loss": 5.7447,
+      "learning_rate": 8.798433635614181e-05,
+      "loss": 3.079,
       "step": 1083
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008802555647155812,
-      "loss": 5.8632,
+      "learning_rate": 8.796372629843364e-05,
+      "loss": 3.4755,
       "step": 1084
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008800494641384996,
-      "loss": 5.3555,
+      "learning_rate": 8.794311624072548e-05,
+      "loss": 3.0411,
       "step": 1085
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.000879843363561418,
-      "loss": 6.2399,
+      "learning_rate": 8.792250618301731e-05,
+      "loss": 3.311,
       "step": 1086
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008796372629843363,
-      "loss": 6.6308,
+      "learning_rate": 8.790189612530915e-05,
+      "loss": 3.5964,
       "step": 1087
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008794311624072547,
-      "loss": 6.1564,
+      "learning_rate": 8.7881286067601e-05,
+      "loss": 3.2349,
       "step": 1088
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008792250618301731,
-      "loss": 6.0152,
+      "learning_rate": 8.786067600989283e-05,
+      "loss": 3.3471,
       "step": 1089
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008790189612530915,
-      "loss": 5.4628,
+      "learning_rate": 8.784006595218467e-05,
+      "loss": 2.8104,
       "step": 1090
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008788128606760099,
-      "loss": 5.6731,
+      "learning_rate": 8.781945589447651e-05,
+      "loss": 2.8942,
       "step": 1091
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008786067600989282,
-      "loss": 5.7508,
+      "learning_rate": 8.779884583676834e-05,
+      "loss": 3.0485,
       "step": 1092
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008784006595218466,
-      "loss": 5.6051,
+      "learning_rate": 8.777823577906017e-05,
+      "loss": 3.1529,
       "step": 1093
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.000878194558944765,
-      "loss": 6.0181,
+      "learning_rate": 8.775762572135202e-05,
+      "loss": 3.1476,
       "step": 1094
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008779884583676834,
-      "loss": 5.6472,
+      "learning_rate": 8.773701566364386e-05,
+      "loss": 2.4036,
       "step": 1095
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008777823577906017,
-      "loss": 6.2686,
+      "learning_rate": 8.77164056059357e-05,
+      "loss": 2.8857,
       "step": 1096
     },
     {
       "epoch": 0.61,
-      "learning_rate": 0.0008775762572135202,
-      "loss": 6.6742,
+      "learning_rate": 8.769579554822755e-05,
+      "loss": 2.8912,
       "step": 1097
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008773701566364386,
-      "loss": 7.2678,
+      "learning_rate": 8.767518549051938e-05,
+      "loss": 2.9443,
       "step": 1098
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.000877164056059357,
-      "loss": 12.335,
+      "learning_rate": 8.765457543281122e-05,
+      "loss": 2.911,
       "step": 1099
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008769579554822754,
-      "loss": 7.8698,
+      "learning_rate": 8.763396537510305e-05,
+      "loss": 2.9345,
       "step": 1100
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008767518549051937,
-      "loss": 5.7204,
+      "learning_rate": 8.76133553173949e-05,
+      "loss": 3.4451,
       "step": 1101
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008765457543281121,
-      "loss": 5.7686,
+      "learning_rate": 8.759274525968672e-05,
+      "loss": 3.7984,
       "step": 1102
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008763396537510305,
-      "loss": 5.7966,
+      "learning_rate": 8.757213520197857e-05,
+      "loss": 3.6451,
       "step": 1103
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008761335531739489,
-      "loss": 5.5083,
+      "learning_rate": 8.755152514427041e-05,
+      "loss": 3.1422,
       "step": 1104
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008759274525968672,
-      "loss": 5.5322,
+      "learning_rate": 8.753091508656225e-05,
+      "loss": 3.2204,
       "step": 1105
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008757213520197856,
-      "loss": 5.2557,
+      "learning_rate": 8.751030502885408e-05,
+      "loss": 3.1598,
       "step": 1106
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.000875515251442704,
-      "loss": 5.7281,
+      "learning_rate": 8.748969497114591e-05,
+      "loss": 3.1061,
       "step": 1107
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008753091508656224,
-      "loss": 5.4922,
+      "learning_rate": 8.746908491343776e-05,
+      "loss": 3.3079,
       "step": 1108
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008751030502885409,
-      "loss": 5.5839,
+      "learning_rate": 8.74484748557296e-05,
+      "loss": 3.2018,
       "step": 1109
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008748969497114591,
-      "loss": 5.6359,
+      "learning_rate": 8.742786479802144e-05,
+      "loss": 3.2696,
       "step": 1110
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008746908491343776,
-      "loss": 5.2382,
+      "learning_rate": 8.740725474031329e-05,
+      "loss": 3.2826,
       "step": 1111
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.000874484748557296,
-      "loss": 5.6426,
+      "learning_rate": 8.738664468260512e-05,
+      "loss": 3.4711,
       "step": 1112
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008742786479802144,
-      "loss": 6.0734,
+      "learning_rate": 8.736603462489695e-05,
+      "loss": 3.3272,
       "step": 1113
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008740725474031328,
-      "loss": 6.0719,
+      "learning_rate": 8.734542456718879e-05,
+      "loss": 3.4329,
       "step": 1114
     },
     {
       "epoch": 0.62,
-      "learning_rate": 0.0008738664468260511,
-      "loss": 5.948,
+      "learning_rate": 8.732481450948063e-05,
+      "loss": 3.7789,
       "step": 1115
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008736603462489696,
-      "loss": 5.7175,
+      "learning_rate": 8.730420445177246e-05,
+      "loss": 3.4099,
       "step": 1116
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.000873454245671888,
-      "loss": 5.5157,
+      "learning_rate": 8.728359439406431e-05,
+      "loss": 3.2189,
       "step": 1117
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008732481450948064,
-      "loss": 5.6237,
+      "learning_rate": 8.726298433635615e-05,
+      "loss": 3.3196,
       "step": 1118
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008730420445177247,
-      "loss": 5.6129,
+      "learning_rate": 8.724237427864798e-05,
+      "loss": 3.655,
       "step": 1119
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008728359439406431,
-      "loss": 5.8905,
+      "learning_rate": 8.722176422093982e-05,
+      "loss": 3.4146,
       "step": 1120
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008726298433635615,
-      "loss": 5.6367,
+      "learning_rate": 8.720115416323165e-05,
+      "loss": 3.2245,
       "step": 1121
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008724237427864799,
-      "loss": 5.5964,
+      "learning_rate": 8.71805441055235e-05,
+      "loss": 3.0978,
       "step": 1122
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008722176422093983,
-      "loss": 5.7005,
+      "learning_rate": 8.715993404781534e-05,
+      "loss": 3.3254,
       "step": 1123
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008720115416323166,
-      "loss": 5.3189,
+      "learning_rate": 8.713932399010718e-05,
+      "loss": 2.8672,
       "step": 1124
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.000871805441055235,
-      "loss": 5.712,
+      "learning_rate": 8.711871393239901e-05,
+      "loss": 3.3341,
       "step": 1125
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008715993404781534,
-      "loss": 5.4733,
+      "learning_rate": 8.709810387469086e-05,
+      "loss": 3.1392,
       "step": 1126
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008713932399010718,
-      "loss": 5.2996,
+      "learning_rate": 8.707749381698269e-05,
+      "loss": 3.2823,
       "step": 1127
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008711871393239901,
-      "loss": 6.1,
+      "learning_rate": 8.705688375927453e-05,
+      "loss": 3.569,
       "step": 1128
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008709810387469085,
-      "loss": 5.1543,
+      "learning_rate": 8.703627370156637e-05,
+      "loss": 2.8221,
       "step": 1129
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008707749381698269,
-      "loss": 6.7132,
+      "learning_rate": 8.70156636438582e-05,
+      "loss": 3.5554,
       "step": 1130
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008705688375927453,
-      "loss": 5.7305,
+      "learning_rate": 8.699505358615005e-05,
+      "loss": 3.4302,
       "step": 1131
     },
     {
       "epoch": 0.63,
-      "learning_rate": 0.0008703627370156637,
-      "loss": 6.1443,
+      "learning_rate": 8.697444352844189e-05,
+      "loss": 3.4033,
       "step": 1132
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.000870156636438582,
-      "loss": 6.1438,
+      "learning_rate": 8.695383347073372e-05,
+      "loss": 3.3311,
       "step": 1133
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008699505358615004,
-      "loss": 5.527,
+      "learning_rate": 8.693322341302555e-05,
+      "loss": 3.2016,
       "step": 1134
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008697444352844188,
-      "loss": 6.0826,
+      "learning_rate": 8.691261335531739e-05,
+      "loss": 3.4786,
       "step": 1135
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008695383347073373,
-      "loss": 5.2393,
+      "learning_rate": 8.689200329760924e-05,
+      "loss": 3.0091,
       "step": 1136
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008693322341302556,
-      "loss": 6.2021,
+      "learning_rate": 8.687139323990108e-05,
+      "loss": 3.2272,
       "step": 1137
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.000869126133553174,
-      "loss": 5.5513,
+      "learning_rate": 8.685078318219292e-05,
+      "loss": 3.0734,
       "step": 1138
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008689200329760924,
-      "loss": 5.335,
+      "learning_rate": 8.683017312448475e-05,
+      "loss": 2.7516,
       "step": 1139
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008687139323990108,
-      "loss": 5.3823,
+      "learning_rate": 8.680956306677658e-05,
+      "loss": 2.9301,
       "step": 1140
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008685078318219292,
-      "loss": 5.5602,
+      "learning_rate": 8.678895300906843e-05,
+      "loss": 3.2064,
       "step": 1141
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008683017312448475,
-      "loss": 6.554,
+      "learning_rate": 8.676834295136027e-05,
+      "loss": 3.2646,
       "step": 1142
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008680956306677659,
-      "loss": 6.7542,
+      "learning_rate": 8.674773289365211e-05,
+      "loss": 3.1517,
       "step": 1143
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008678895300906843,
-      "loss": 5.7537,
+      "learning_rate": 8.672712283594394e-05,
+      "loss": 2.7706,
       "step": 1144
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008676834295136027,
-      "loss": 5.3978,
+      "learning_rate": 8.670651277823579e-05,
+      "loss": 2.5815,
       "step": 1145
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008674773289365211,
-      "loss": 5.555,
+      "learning_rate": 8.668590272052763e-05,
+      "loss": 2.3234,
       "step": 1146
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008672712283594394,
-      "loss": 5.6886,
+      "learning_rate": 8.666529266281946e-05,
+      "loss": 3.005,
       "step": 1147
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008670651277823578,
-      "loss": 5.4934,
+      "learning_rate": 8.664468260511129e-05,
+      "loss": 2.1639,
       "step": 1148
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008668590272052762,
-      "loss": 6.3331,
+      "learning_rate": 8.662407254740313e-05,
+      "loss": 2.7387,
       "step": 1149
     },
     {
       "epoch": 0.64,
-      "learning_rate": 0.0008666529266281946,
-      "loss": 5.794,
+      "learning_rate": 8.660346248969498e-05,
+      "loss": 2.23,
       "step": 1150
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008664468260511129,
-      "loss": 5.789,
+      "learning_rate": 8.658285243198682e-05,
+      "loss": 3.6973,
       "step": 1151
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008662407254740313,
-      "loss": 5.9289,
+      "learning_rate": 8.656224237427866e-05,
+      "loss": 3.6621,
       "step": 1152
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008660346248969497,
-      "loss": 5.6205,
+      "learning_rate": 8.654163231657049e-05,
+      "loss": 3.385,
       "step": 1153
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008658285243198681,
-      "loss": 5.849,
+      "learning_rate": 8.652102225886232e-05,
+      "loss": 3.3895,
       "step": 1154
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008656224237427865,
-      "loss": 5.9778,
+      "learning_rate": 8.650041220115417e-05,
+      "loss": 3.2502,
       "step": 1155
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008654163231657048,
-      "loss": 5.5916,
+      "learning_rate": 8.647980214344601e-05,
+      "loss": 3.5346,
       "step": 1156
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008652102225886233,
-      "loss": 5.5011,
+      "learning_rate": 8.645919208573784e-05,
+      "loss": 3.2881,
       "step": 1157
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008650041220115417,
-      "loss": 5.8971,
+      "learning_rate": 8.643858202802968e-05,
+      "loss": 3.5045,
       "step": 1158
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008647980214344601,
-      "loss": 5.3912,
+      "learning_rate": 8.641797197032153e-05,
+      "loss": 3.4182,
       "step": 1159
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008645919208573784,
-      "loss": 5.3985,
+      "learning_rate": 8.639736191261336e-05,
+      "loss": 3.3742,
       "step": 1160
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008643858202802968,
-      "loss": 5.5576,
+      "learning_rate": 8.63767518549052e-05,
+      "loss": 3.462,
       "step": 1161
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008641797197032152,
-      "loss": 5.3502,
+      "learning_rate": 8.635614179719703e-05,
+      "loss": 3.3557,
       "step": 1162
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008639736191261336,
-      "loss": 5.4327,
+      "learning_rate": 8.633553173948887e-05,
+      "loss": 3.223,
       "step": 1163
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.000863767518549052,
-      "loss": 5.2577,
+      "learning_rate": 8.631492168178072e-05,
+      "loss": 3.12,
       "step": 1164
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008635614179719703,
-      "loss": 5.3807,
+      "learning_rate": 8.629431162407256e-05,
+      "loss": 3.0721,
       "step": 1165
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008633553173948887,
-      "loss": 5.5532,
+      "learning_rate": 8.627370156636439e-05,
+      "loss": 3.4076,
       "step": 1166
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008631492168178071,
-      "loss": 5.5996,
+      "learning_rate": 8.625309150865622e-05,
+      "loss": 3.1134,
       "step": 1167
     },
     {
       "epoch": 0.65,
-      "learning_rate": 0.0008629431162407255,
-      "loss": 4.9879,
+      "learning_rate": 8.623248145094806e-05,
+      "loss": 3.0363,
       "step": 1168
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008627370156636439,
-      "loss": 5.4891,
+      "learning_rate": 8.62118713932399e-05,
+      "loss": 3.1258,
       "step": 1169
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008625309150865622,
-      "loss": 5.8581,
+      "learning_rate": 8.619126133553175e-05,
+      "loss": 3.4309,
       "step": 1170
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008623248145094806,
-      "loss": 6.013,
+      "learning_rate": 8.617065127782358e-05,
+      "loss": 3.2652,
       "step": 1171
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.000862118713932399,
-      "loss": 5.1599,
+      "learning_rate": 8.615004122011542e-05,
+      "loss": 3.1549,
       "step": 1172
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008619126133553174,
-      "loss": 5.8124,
+      "learning_rate": 8.612943116240727e-05,
+      "loss": 3.4153,
       "step": 1173
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008617065127782357,
-      "loss": 5.2817,
+      "learning_rate": 8.61088211046991e-05,
+      "loss": 3.2004,
       "step": 1174
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008615004122011541,
-      "loss": 6.0596,
+      "learning_rate": 8.608821104699094e-05,
+      "loss": 3.0633,
       "step": 1175
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008612943116240725,
-      "loss": 6.4745,
+      "learning_rate": 8.606760098928277e-05,
+      "loss": 3.4094,
       "step": 1176
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.000861088211046991,
-      "loss": 5.5368,
+      "learning_rate": 8.604699093157461e-05,
+      "loss": 3.2517,
       "step": 1177
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008608821104699094,
-      "loss": 5.5205,
+      "learning_rate": 8.602638087386646e-05,
+      "loss": 3.1458,
       "step": 1178
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008606760098928277,
-      "loss": 5.3163,
+      "learning_rate": 8.60057708161583e-05,
+      "loss": 3.133,
       "step": 1179
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008604699093157461,
-      "loss": 5.5853,
+      "learning_rate": 8.598516075845013e-05,
+      "loss": 3.0775,
       "step": 1180
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008602638087386645,
-      "loss": 5.4125,
+      "learning_rate": 8.596455070074196e-05,
+      "loss": 2.967,
       "step": 1181
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008600577081615829,
-      "loss": 5.9092,
+      "learning_rate": 8.59439406430338e-05,
+      "loss": 3.4245,
       "step": 1182
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008598516075845012,
-      "loss": 5.8494,
+      "learning_rate": 8.592333058532564e-05,
+      "loss": 3.2345,
       "step": 1183
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008596455070074196,
-      "loss": 5.4406,
+      "learning_rate": 8.590272052761749e-05,
+      "loss": 3.0378,
       "step": 1184
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.000859439406430338,
-      "loss": 6.0505,
+      "learning_rate": 8.588211046990932e-05,
+      "loss": 3.4276,
       "step": 1185
     },
     {
       "epoch": 0.66,
-      "learning_rate": 0.0008592333058532564,
-      "loss": 5.8445,
+      "learning_rate": 8.586150041220116e-05,
+      "loss": 3.1127,
       "step": 1186
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008590272052761748,
-      "loss": 5.8123,
+      "learning_rate": 8.584089035449299e-05,
+      "loss": 3.0393,
       "step": 1187
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008588211046990931,
-      "loss": 5.5666,
+      "learning_rate": 8.582028029678483e-05,
+      "loss": 2.8792,
       "step": 1188
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008586150041220115,
-      "loss": 5.7071,
+      "learning_rate": 8.579967023907666e-05,
+      "loss": 2.9587,
       "step": 1189
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008584089035449299,
-      "loss": 6.5352,
+      "learning_rate": 8.577906018136851e-05,
+      "loss": 3.3851,
       "step": 1190
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008582028029678483,
-      "loss": 5.4081,
+      "learning_rate": 8.575845012366035e-05,
+      "loss": 3.0215,
       "step": 1191
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008579967023907666,
-      "loss": 6.5511,
+      "learning_rate": 8.57378400659522e-05,
+      "loss": 3.3088,
       "step": 1192
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.000857790601813685,
-      "loss": 5.934,
+      "learning_rate": 8.571723000824404e-05,
+      "loss": 3.5431,
       "step": 1193
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008575845012366034,
-      "loss": 5.5478,
+      "learning_rate": 8.569661995053587e-05,
+      "loss": 2.9121,
       "step": 1194
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008573784006595218,
-      "loss": 5.4186,
+      "learning_rate": 8.56760098928277e-05,
+      "loss": 2.3031,
       "step": 1195
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008571723000824402,
-      "loss": 5.3277,
+      "learning_rate": 8.565539983511954e-05,
+      "loss": 2.6933,
       "step": 1196
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008569661995053585,
-      "loss": 5.6445,
+      "learning_rate": 8.563478977741138e-05,
+      "loss": 2.73,
       "step": 1197
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.000856760098928277,
-      "loss": 5.5742,
+      "learning_rate": 8.561417971970323e-05,
+      "loss": 2.3506,
       "step": 1198
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008565539983511954,
-      "loss": 5.4332,
+      "learning_rate": 8.559356966199506e-05,
+      "loss": 2.4098,
       "step": 1199
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008563478977741138,
-      "loss": 5.0615,
+      "learning_rate": 8.55729596042869e-05,
+      "loss": 2.2267,
       "step": 1200
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008561417971970322,
-      "loss": 5.667,
+      "learning_rate": 8.555234954657873e-05,
+      "loss": 3.6371,
       "step": 1201
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008559356966199505,
-      "loss": 5.4229,
+      "learning_rate": 8.553173948887057e-05,
+      "loss": 3.0107,
       "step": 1202
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.000855729596042869,
-      "loss": 5.3872,
+      "learning_rate": 8.55111294311624e-05,
+      "loss": 3.2842,
       "step": 1203
     },
     {
       "epoch": 0.67,
-      "learning_rate": 0.0008555234954657874,
-      "loss": 5.3896,
+      "learning_rate": 8.549051937345425e-05,
+      "loss": 2.9608,
       "step": 1204
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008553173948887058,
-      "loss": 5.1914,
+      "learning_rate": 8.546990931574609e-05,
+      "loss": 2.9225,
       "step": 1205
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008551112943116241,
-      "loss": 5.4791,
+      "learning_rate": 8.544929925803793e-05,
+      "loss": 3.2767,
       "step": 1206
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008549051937345425,
-      "loss": 7.4053,
+      "learning_rate": 8.542868920032976e-05,
+      "loss": 4.0952,
       "step": 1207
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008546990931574609,
-      "loss": 5.8224,
+      "learning_rate": 8.54080791426216e-05,
+      "loss": 3.4923,
       "step": 1208
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008544929925803793,
-      "loss": 6.301,
+      "learning_rate": 8.538746908491344e-05,
+      "loss": 3.5404,
       "step": 1209
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008542868920032977,
-      "loss": 5.4241,
+      "learning_rate": 8.536685902720528e-05,
+      "loss": 3.2502,
       "step": 1210
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.000854080791426216,
-      "loss": 5.7755,
+      "learning_rate": 8.534624896949712e-05,
+      "loss": 3.2619,
       "step": 1211
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008538746908491344,
-      "loss": 5.9801,
+      "learning_rate": 8.532563891178895e-05,
+      "loss": 3.4307,
       "step": 1212
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008536685902720528,
-      "loss": 5.5397,
+      "learning_rate": 8.53050288540808e-05,
+      "loss": 3.6312,
       "step": 1213
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008534624896949712,
-      "loss": 5.7723,
+      "learning_rate": 8.528441879637263e-05,
+      "loss": 3.714,
       "step": 1214
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008532563891178895,
-      "loss": 5.385,
+      "learning_rate": 8.526380873866447e-05,
+      "loss": 3.6503,
       "step": 1215
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008530502885408079,
-      "loss": 5.6722,
+      "learning_rate": 8.524319868095631e-05,
+      "loss": 3.3536,
       "step": 1216
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008528441879637264,
-      "loss": 5.434,
+      "learning_rate": 8.522258862324814e-05,
+      "loss": 3.3792,
       "step": 1217
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008526380873866448,
-      "loss": 5.1889,
+      "learning_rate": 8.520197856553999e-05,
+      "loss": 2.9902,
       "step": 1218
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008524319868095632,
-      "loss": 5.3302,
+      "learning_rate": 8.518136850783183e-05,
+      "loss": 3.3397,
       "step": 1219
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008522258862324815,
-      "loss": 5.6648,
+      "learning_rate": 8.516075845012367e-05,
+      "loss": 3.3587,
       "step": 1220
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008520197856553999,
-      "loss": 5.4091,
+      "learning_rate": 8.51401483924155e-05,
+      "loss": 2.9795,
       "step": 1221
     },
     {
       "epoch": 0.68,
-      "learning_rate": 0.0008518136850783183,
-      "loss": 6.3325,
+      "learning_rate": 8.511953833470733e-05,
+      "loss": 3.0601,
       "step": 1222
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008516075845012367,
-      "loss": 5.5858,
+      "learning_rate": 8.509892827699918e-05,
+      "loss": 3.0481,
       "step": 1223
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.000851401483924155,
-      "loss": 6.14,
+      "learning_rate": 8.507831821929102e-05,
+      "loss": 3.2475,
       "step": 1224
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008511953833470734,
-      "loss": 6.4742,
+      "learning_rate": 8.505770816158286e-05,
+      "loss": 3.2918,
       "step": 1225
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008509892827699918,
-      "loss": 5.4455,
+      "learning_rate": 8.50370981038747e-05,
+      "loss": 3.5073,
       "step": 1226
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008507831821929102,
-      "loss": 5.4558,
+      "learning_rate": 8.501648804616654e-05,
+      "loss": 3.278,
       "step": 1227
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008505770816158286,
-      "loss": 5.6804,
+      "learning_rate": 8.499587798845837e-05,
+      "loss": 2.9644,
       "step": 1228
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008503709810387469,
-      "loss": 5.3473,
+      "learning_rate": 8.497526793075021e-05,
+      "loss": 2.8506,
       "step": 1229
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008501648804616653,
-      "loss": 6.164,
+      "learning_rate": 8.495465787304205e-05,
+      "loss": 3.6093,
       "step": 1230
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008499587798845837,
-      "loss": 5.4779,
+      "learning_rate": 8.493404781533388e-05,
+      "loss": 3.2947,
       "step": 1231
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008497526793075021,
-      "loss": 6.1431,
+      "learning_rate": 8.491343775762573e-05,
+      "loss": 3.2853,
       "step": 1232
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008495465787304205,
-      "loss": 5.6048,
+      "learning_rate": 8.489282769991757e-05,
+      "loss": 3.3072,
       "step": 1233
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008493404781533388,
-      "loss": 5.502,
+      "learning_rate": 8.48722176422094e-05,
+      "loss": 3.2354,
       "step": 1234
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008491343775762572,
-      "loss": 5.7898,
+      "learning_rate": 8.485160758450123e-05,
+      "loss": 3.3764,
       "step": 1235
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008489282769991756,
-      "loss": 5.3421,
+      "learning_rate": 8.483099752679307e-05,
+      "loss": 2.9744,
       "step": 1236
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.000848722176422094,
-      "loss": 5.9001,
+      "learning_rate": 8.481038746908492e-05,
+      "loss": 3.0591,
       "step": 1237
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008485160758450124,
-      "loss": 5.4223,
+      "learning_rate": 8.478977741137676e-05,
+      "loss": 2.9038,
       "step": 1238
     },
     {
       "epoch": 0.69,
-      "learning_rate": 0.0008483099752679308,
-      "loss": 5.9017,
+      "learning_rate": 8.47691673536686e-05,
+      "loss": 3.1885,
       "step": 1239
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008481038746908492,
-      "loss": 6.7111,
+      "learning_rate": 8.474855729596043e-05,
+      "loss": 3.266,
       "step": 1240
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008478977741137676,
-      "loss": 5.745,
+      "learning_rate": 8.472794723825228e-05,
+      "loss": 3.5284,
       "step": 1241
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.000847691673536686,
-      "loss": 6.1242,
+      "learning_rate": 8.47073371805441e-05,
+      "loss": 3.5993,
       "step": 1242
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008474855729596043,
-      "loss": 5.5578,
+      "learning_rate": 8.468672712283595e-05,
+      "loss": 3.3414,
       "step": 1243
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008472794723825227,
-      "loss": 5.5023,
+      "learning_rate": 8.466611706512778e-05,
+      "loss": 2.709,
       "step": 1244
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008470733718054411,
-      "loss": 5.1116,
+      "learning_rate": 8.464550700741962e-05,
+      "loss": 2.2346,
       "step": 1245
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008468672712283595,
-      "loss": 5.657,
+      "learning_rate": 8.462489694971147e-05,
+      "loss": 3.2507,
       "step": 1246
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008466611706512778,
-      "loss": 5.6068,
+      "learning_rate": 8.460428689200331e-05,
+      "loss": 2.4299,
       "step": 1247
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008464550700741962,
-      "loss": 5.4356,
+      "learning_rate": 8.458367683429514e-05,
+      "loss": 2.6532,
       "step": 1248
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008462489694971146,
-      "loss": 5.6828,
+      "learning_rate": 8.456306677658697e-05,
+      "loss": 2.3572,
       "step": 1249
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.000846042868920033,
-      "loss": 5.6051,
+      "learning_rate": 8.454245671887881e-05,
+      "loss": 2.2594,
       "step": 1250
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008458367683429514,
-      "loss": 6.0235,
+      "learning_rate": 8.452184666117066e-05,
+      "loss": 3.8251,
       "step": 1251
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008456306677658697,
-      "loss": 5.8462,
+      "learning_rate": 8.45012366034625e-05,
+      "loss": 3.8697,
       "step": 1252
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008454245671887881,
-      "loss": 5.7775,
+      "learning_rate": 8.448062654575433e-05,
+      "loss": 3.5477,
       "step": 1253
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008452184666117065,
-      "loss": 5.587,
+      "learning_rate": 8.446001648804617e-05,
+      "loss": 3.2526,
       "step": 1254
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008450123660346249,
-      "loss": 6.2628,
+      "learning_rate": 8.4439406430338e-05,
+      "loss": 3.7286,
       "step": 1255
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008448062654575432,
-      "loss": 6.2243,
+      "learning_rate": 8.441879637262985e-05,
+      "loss": 3.5199,
       "step": 1256
     },
     {
       "epoch": 0.7,
-      "learning_rate": 0.0008446001648804616,
-      "loss": 5.3067,
+      "learning_rate": 8.439818631492169e-05,
+      "loss": 3.3646,
       "step": 1257
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.00084439406430338,
-      "loss": 5.3502,
+      "learning_rate": 8.437757625721352e-05,
+      "loss": 3.2961,
       "step": 1258
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008441879637262985,
-      "loss": 5.4845,
+      "learning_rate": 8.435696619950536e-05,
+      "loss": 3.3367,
       "step": 1259
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008439818631492169,
-      "loss": 5.8301,
+      "learning_rate": 8.43363561417972e-05,
+      "loss": 3.5792,
       "step": 1260
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008437757625721352,
-      "loss": 5.5284,
+      "learning_rate": 8.431574608408904e-05,
+      "loss": 3.2386,
       "step": 1261
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008435696619950536,
-      "loss": 5.519,
+      "learning_rate": 8.429513602638088e-05,
+      "loss": 3.571,
       "step": 1262
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.000843363561417972,
-      "loss": 5.3825,
+      "learning_rate": 8.427452596867271e-05,
+      "loss": 3.2391,
       "step": 1263
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008431574608408904,
-      "loss": 5.3299,
+      "learning_rate": 8.425391591096455e-05,
+      "loss": 3.0556,
       "step": 1264
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008429513602638088,
-      "loss": 7.3644,
+      "learning_rate": 8.42333058532564e-05,
+      "loss": 3.5949,
       "step": 1265
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008427452596867271,
-      "loss": 6.1926,
+      "learning_rate": 8.421269579554824e-05,
+      "loss": 3.5777,
       "step": 1266
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008425391591096455,
-      "loss": 5.5954,
+      "learning_rate": 8.419208573784007e-05,
+      "loss": 3.2968,
       "step": 1267
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008423330585325639,
-      "loss": 5.3388,
+      "learning_rate": 8.417147568013191e-05,
+      "loss": 3.0445,
       "step": 1268
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008421269579554823,
-      "loss": 5.4468,
+      "learning_rate": 8.415086562242374e-05,
+      "loss": 3.3897,
       "step": 1269
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008419208573784006,
-      "loss": 5.3038,
+      "learning_rate": 8.413025556471559e-05,
+      "loss": 3.127,
       "step": 1270
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.000841714756801319,
-      "loss": 5.9022,
+      "learning_rate": 8.410964550700743e-05,
+      "loss": 3.4141,
       "step": 1271
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008415086562242374,
-      "loss": 5.396,
+      "learning_rate": 8.408903544929926e-05,
+      "loss": 3.4752,
       "step": 1272
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008413025556471558,
-      "loss": 5.5037,
+      "learning_rate": 8.40684253915911e-05,
+      "loss": 3.2293,
       "step": 1273
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008410964550700742,
-      "loss": 5.4653,
+      "learning_rate": 8.404781533388295e-05,
+      "loss": 3.4242,
       "step": 1274
     },
     {
       "epoch": 0.71,
-      "learning_rate": 0.0008408903544929925,
-      "loss": 5.4283,
+      "learning_rate": 8.402720527617478e-05,
+      "loss": 3.1961,
       "step": 1275
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008406842539159109,
-      "loss": 5.8997,
+      "learning_rate": 8.40065952184666e-05,
+      "loss": 3.4521,
       "step": 1276
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008404781533388293,
-      "loss": 5.6983,
+      "learning_rate": 8.398598516075845e-05,
+      "loss": 3.3504,
       "step": 1277
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008402720527617478,
-      "loss": 5.5442,
+      "learning_rate": 8.396537510305029e-05,
+      "loss": 3.1691,
       "step": 1278
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.000840065952184666,
-      "loss": 5.775,
+      "learning_rate": 8.394476504534214e-05,
+      "loss": 3.6596,
       "step": 1279
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008398598516075845,
-      "loss": 5.9126,
+      "learning_rate": 8.392415498763398e-05,
+      "loss": 3.2845,
       "step": 1280
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008396537510305029,
-      "loss": 5.3171,
+      "learning_rate": 8.390354492992581e-05,
+      "loss": 3.7254,
       "step": 1281
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008394476504534213,
-      "loss": 5.3271,
+      "learning_rate": 8.388293487221764e-05,
+      "loss": 3.1872,
       "step": 1282
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008392415498763397,
-      "loss": 6.0536,
+      "learning_rate": 8.386232481450948e-05,
+      "loss": 3.3666,
       "step": 1283
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.000839035449299258,
-      "loss": 5.394,
+      "learning_rate": 8.384171475680132e-05,
+      "loss": 3.0749,
       "step": 1284
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008388293487221764,
-      "loss": 5.3376,
+      "learning_rate": 8.382110469909317e-05,
+      "loss": 2.9784,
       "step": 1285
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008386232481450948,
-      "loss": 5.5803,
+      "learning_rate": 8.3800494641385e-05,
+      "loss": 3.1085,
       "step": 1286
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008384171475680132,
-      "loss": 5.706,
+      "learning_rate": 8.377988458367684e-05,
+      "loss": 3.1536,
       "step": 1287
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008382110469909316,
-      "loss": 5.331,
+      "learning_rate": 8.375927452596868e-05,
+      "loss": 3.1804,
       "step": 1288
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008380049464138499,
-      "loss": 5.4349,
+      "learning_rate": 8.373866446826051e-05,
+      "loss": 2.8663,
       "step": 1289
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008377988458367683,
-      "loss": 5.5164,
+      "learning_rate": 8.371805441055234e-05,
+      "loss": 2.8371,
       "step": 1290
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008375927452596867,
-      "loss": 5.5294,
+      "learning_rate": 8.369744435284419e-05,
+      "loss": 3.1125,
       "step": 1291
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008373866446826051,
-      "loss": 7.0537,
+      "learning_rate": 8.367683429513603e-05,
+      "loss": 3.2103,
       "step": 1292
     },
     {
       "epoch": 0.72,
-      "learning_rate": 0.0008371805441055234,
-      "loss": 7.4283,
+      "learning_rate": 8.365622423742787e-05,
+      "loss": 3.7727,
       "step": 1293
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008369744435284418,
-      "loss": 5.5964,
+      "learning_rate": 8.363561417971972e-05,
+      "loss": 3.0589,
       "step": 1294
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008367683429513602,
-      "loss": 5.1711,
+      "learning_rate": 8.361500412201155e-05,
+      "loss": 2.7783,
       "step": 1295
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008365622423742787,
-      "loss": 5.1941,
+      "learning_rate": 8.359439406430338e-05,
+      "loss": 2.551,
       "step": 1296
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008363561417971972,
-      "loss": 5.1003,
+      "learning_rate": 8.357378400659522e-05,
+      "loss": 2.4329,
       "step": 1297
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008361500412201155,
-      "loss": 4.8134,
+      "learning_rate": 8.355317394888706e-05,
+      "loss": 1.9222,
       "step": 1298
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008359439406430339,
-      "loss": 7.1477,
+      "learning_rate": 8.35325638911789e-05,
+      "loss": 2.9008,
       "step": 1299
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008357378400659523,
-      "loss": 4.9195,
+      "learning_rate": 8.351195383347074e-05,
+      "loss": 2.198,
       "step": 1300
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008355317394888707,
-      "loss": 6.711,
+      "learning_rate": 8.349134377576258e-05,
+      "loss": 3.647,
       "step": 1301
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.000835325638911789,
-      "loss": 6.8379,
+      "learning_rate": 8.347073371805441e-05,
+      "loss": 4.0206,
       "step": 1302
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008351195383347074,
-      "loss": 5.6529,
+      "learning_rate": 8.345012366034625e-05,
+      "loss": 3.5826,
       "step": 1303
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008349134377576258,
-      "loss": 5.798,
+      "learning_rate": 8.342951360263808e-05,
+      "loss": 3.6467,
       "step": 1304
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008347073371805442,
-      "loss": 5.51,
+      "learning_rate": 8.340890354492993e-05,
+      "loss": 3.214,
       "step": 1305
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008345012366034626,
-      "loss": 5.6212,
+      "learning_rate": 8.338829348722177e-05,
+      "loss": 3.3422,
       "step": 1306
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008342951360263809,
-      "loss": 6.1351,
+      "learning_rate": 8.336768342951361e-05,
+      "loss": 3.7896,
       "step": 1307
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008340890354492993,
-      "loss": 5.3197,
+      "learning_rate": 8.334707337180544e-05,
+      "loss": 3.3511,
       "step": 1308
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008338829348722177,
-      "loss": 7.2568,
+      "learning_rate": 8.332646331409727e-05,
+      "loss": 3.5358,
       "step": 1309
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008336768342951361,
-      "loss": 5.3516,
+      "learning_rate": 8.330585325638912e-05,
+      "loss": 3.1521,
       "step": 1310
     },
     {
       "epoch": 0.73,
-      "learning_rate": 0.0008334707337180544,
-      "loss": 5.2981,
+      "learning_rate": 8.328524319868096e-05,
+      "loss": 3.1304,
       "step": 1311
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008332646331409728,
-      "loss": 5.6216,
+      "learning_rate": 8.32646331409728e-05,
+      "loss": 3.4395,
       "step": 1312
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008330585325638912,
-      "loss": 5.3826,
+      "learning_rate": 8.324402308326463e-05,
+      "loss": 3.1879,
       "step": 1313
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008328524319868096,
-      "loss": 5.1565,
+      "learning_rate": 8.322341302555648e-05,
+      "loss": 2.9113,
       "step": 1314
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.000832646331409728,
-      "loss": 5.8199,
+      "learning_rate": 8.320280296784832e-05,
+      "loss": 3.2182,
       "step": 1315
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008324402308326463,
-      "loss": 5.6888,
+      "learning_rate": 8.318219291014015e-05,
+      "loss": 2.947,
       "step": 1316
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008322341302555647,
-      "loss": 5.1787,
+      "learning_rate": 8.3161582852432e-05,
+      "loss": 2.938,
       "step": 1317
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008320280296784832,
-      "loss": 5.7905,
+      "learning_rate": 8.314097279472382e-05,
+      "loss": 3.5481,
       "step": 1318
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008318219291014016,
-      "loss": 5.3534,
+      "learning_rate": 8.312036273701567e-05,
+      "loss": 3.2249,
       "step": 1319
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.00083161582852432,
-      "loss": 5.2817,
+      "learning_rate": 8.309975267930751e-05,
+      "loss": 2.9289,
       "step": 1320
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008314097279472383,
-      "loss": 5.063,
+      "learning_rate": 8.307914262159935e-05,
+      "loss": 3.0008,
       "step": 1321
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008312036273701567,
-      "loss": 5.2679,
+      "learning_rate": 8.305853256389118e-05,
+      "loss": 3.1862,
       "step": 1322
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008309975267930751,
-      "loss": 5.5354,
+      "learning_rate": 8.303792250618301e-05,
+      "loss": 2.903,
       "step": 1323
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008307914262159935,
-      "loss": 5.7153,
+      "learning_rate": 8.301731244847486e-05,
+      "loss": 3.4732,
       "step": 1324
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008305853256389118,
-      "loss": 5.6569,
+      "learning_rate": 8.29967023907667e-05,
+      "loss": 3.4684,
       "step": 1325
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008303792250618302,
-      "loss": 5.5469,
+      "learning_rate": 8.297609233305854e-05,
+      "loss": 3.0824,
       "step": 1326
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008301731244847486,
-      "loss": 5.3349,
+      "learning_rate": 8.295548227535037e-05,
+      "loss": 2.8999,
       "step": 1327
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.000829967023907667,
-      "loss": 5.8347,
+      "learning_rate": 8.293487221764222e-05,
+      "loss": 3.2056,
       "step": 1328
     },
     {
       "epoch": 0.74,
-      "learning_rate": 0.0008297609233305854,
-      "loss": 5.6732,
+      "learning_rate": 8.291426215993405e-05,
+      "loss": 3.2012,
       "step": 1329
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008295548227535037,
-      "loss": 5.7834,
+      "learning_rate": 8.289365210222589e-05,
+      "loss": 3.4888,
       "step": 1330
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008293487221764221,
-      "loss": 5.1808,
+      "learning_rate": 8.287304204451772e-05,
+      "loss": 3.0497,
       "step": 1331
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008291426215993405,
-      "loss": 5.8839,
+      "learning_rate": 8.285243198680956e-05,
+      "loss": 3.1229,
       "step": 1332
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008289365210222589,
-      "loss": 5.8086,
+      "learning_rate": 8.28318219291014e-05,
+      "loss": 3.381,
       "step": 1333
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008287304204451772,
-      "loss": 5.5766,
+      "learning_rate": 8.281121187139325e-05,
+      "loss": 3.0136,
       "step": 1334
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008285243198680956,
-      "loss": 5.2663,
+      "learning_rate": 8.27906018136851e-05,
+      "loss": 2.9316,
       "step": 1335
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.000828318219291014,
-      "loss": 5.3895,
+      "learning_rate": 8.276999175597692e-05,
+      "loss": 2.9445,
       "step": 1336
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008281121187139324,
-      "loss": 5.5717,
+      "learning_rate": 8.274938169826875e-05,
+      "loss": 3.1081,
       "step": 1337
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008279060181368509,
-      "loss": 5.3612,
+      "learning_rate": 8.27287716405606e-05,
+      "loss": 2.9731,
       "step": 1338
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008276999175597691,
-      "loss": 5.6131,
+      "learning_rate": 8.270816158285244e-05,
+      "loss": 3.0567,
       "step": 1339
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008274938169826876,
-      "loss": 6.4087,
+      "learning_rate": 8.268755152514427e-05,
+      "loss": 3.196,
       "step": 1340
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.000827287716405606,
-      "loss": 5.7496,
+      "learning_rate": 8.266694146743611e-05,
+      "loss": 2.9826,
       "step": 1341
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008270816158285244,
-      "loss": 5.8622,
+      "learning_rate": 8.264633140972796e-05,
+      "loss": 3.1175,
       "step": 1342
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008268755152514427,
-      "loss": 5.7602,
+      "learning_rate": 8.262572135201979e-05,
+      "loss": 2.9858,
       "step": 1343
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008266694146743611,
-      "loss": 5.5818,
+      "learning_rate": 8.260511129431163e-05,
+      "loss": 3.0446,
       "step": 1344
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008264633140972795,
-      "loss": 6.1063,
+      "learning_rate": 8.258450123660346e-05,
+      "loss": 3.1262,
       "step": 1345
     },
     {
       "epoch": 0.75,
-      "learning_rate": 0.0008262572135201979,
-      "loss": 5.5154,
+      "learning_rate": 8.25638911788953e-05,
+      "loss": 2.5059,
       "step": 1346
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008260511129431163,
-      "loss": 6.4484,
+      "learning_rate": 8.254328112118715e-05,
+      "loss": 3.1915,
       "step": 1347
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008258450123660346,
-      "loss": 5.1804,
+      "learning_rate": 8.252267106347899e-05,
+      "loss": 2.3362,
       "step": 1348
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.000825638911788953,
-      "loss": 6.3442,
+      "learning_rate": 8.250206100577082e-05,
+      "loss": 2.3134,
       "step": 1349
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008254328112118714,
-      "loss": 5.1424,
+      "learning_rate": 8.248145094806265e-05,
+      "loss": 2.0572,
       "step": 1350
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008252267106347898,
-      "loss": 5.8901,
+      "learning_rate": 8.246084089035449e-05,
+      "loss": 3.6827,
       "step": 1351
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008250206100577082,
-      "loss": 5.666,
+      "learning_rate": 8.244023083264634e-05,
+      "loss": 3.549,
       "step": 1352
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008248145094806265,
-      "loss": 5.6649,
+      "learning_rate": 8.241962077493818e-05,
+      "loss": 3.3273,
       "step": 1353
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008246084089035449,
-      "loss": 5.9874,
+      "learning_rate": 8.239901071723001e-05,
+      "loss": 3.6194,
       "step": 1354
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008244023083264633,
-      "loss": 5.3731,
+      "learning_rate": 8.237840065952185e-05,
+      "loss": 3.4694,
       "step": 1355
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008241962077493817,
-      "loss": 5.7902,
+      "learning_rate": 8.23577906018137e-05,
+      "loss": 3.4872,
       "step": 1356
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008239901071723,
-      "loss": 5.1884,
+      "learning_rate": 8.233718054410553e-05,
+      "loss": 3.0536,
       "step": 1357
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008237840065952184,
-      "loss": 5.7171,
+      "learning_rate": 8.231657048639737e-05,
+      "loss": 3.5202,
       "step": 1358
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008235779060181368,
-      "loss": 5.5956,
+      "learning_rate": 8.22959604286892e-05,
+      "loss": 3.2232,
       "step": 1359
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008233718054410553,
-      "loss": 5.8232,
+      "learning_rate": 8.227535037098104e-05,
+      "loss": 3.7403,
       "step": 1360
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008231657048639737,
-      "loss": 5.4079,
+      "learning_rate": 8.225474031327289e-05,
+      "loss": 3.387,
       "step": 1361
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.000822959604286892,
-      "loss": 5.303,
+      "learning_rate": 8.223413025556473e-05,
+      "loss": 3.2031,
       "step": 1362
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008227535037098104,
-      "loss": 5.3391,
+      "learning_rate": 8.221352019785656e-05,
+      "loss": 3.5119,
       "step": 1363
     },
     {
       "epoch": 0.76,
-      "learning_rate": 0.0008225474031327288,
-      "loss": 5.9572,
+      "learning_rate": 8.219291014014839e-05,
+      "loss": 3.7749,
       "step": 1364
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008223413025556472,
-      "loss": 6.1778,
+      "learning_rate": 8.217230008244023e-05,
+      "loss": 3.4511,
       "step": 1365
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008221352019785655,
-      "loss": 5.4847,
+      "learning_rate": 8.215169002473208e-05,
+      "loss": 3.3164,
       "step": 1366
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008219291014014839,
-      "loss": 5.6996,
+      "learning_rate": 8.213107996702392e-05,
+      "loss": 3.2712,
       "step": 1367
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008217230008244023,
-      "loss": 5.3126,
+      "learning_rate": 8.211046990931575e-05,
+      "loss": 2.9809,
       "step": 1368
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008215169002473207,
-      "loss": 5.4157,
+      "learning_rate": 8.208985985160759e-05,
+      "loss": 3.0277,
       "step": 1369
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008213107996702391,
-      "loss": 5.516,
+      "learning_rate": 8.206924979389942e-05,
+      "loss": 2.9483,
       "step": 1370
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008211046990931574,
-      "loss": 5.6983,
+      "learning_rate": 8.204863973619127e-05,
+      "loss": 3.1291,
       "step": 1371
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008208985985160758,
-      "loss": 5.0904,
+      "learning_rate": 8.20280296784831e-05,
+      "loss": 2.8611,
       "step": 1372
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008206924979389942,
-      "loss": 5.8467,
+      "learning_rate": 8.200741962077494e-05,
+      "loss": 3.4791,
       "step": 1373
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008204863973619126,
-      "loss": 5.8424,
+      "learning_rate": 8.198680956306678e-05,
+      "loss": 3.4157,
       "step": 1374
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008202802967848309,
-      "loss": 5.6727,
+      "learning_rate": 8.196619950535863e-05,
+      "loss": 3.0638,
       "step": 1375
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008200741962077493,
-      "loss": 5.2052,
+      "learning_rate": 8.194558944765046e-05,
+      "loss": 3.1813,
       "step": 1376
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008198680956306677,
-      "loss": 6.0616,
+      "learning_rate": 8.192497938994228e-05,
+      "loss": 3.2391,
       "step": 1377
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008196619950535861,
-      "loss": 5.6108,
+      "learning_rate": 8.190436933223413e-05,
+      "loss": 3.3054,
       "step": 1378
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008194558944765046,
-      "loss": 5.5427,
+      "learning_rate": 8.188375927452597e-05,
+      "loss": 3.3955,
       "step": 1379
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008192497938994228,
-      "loss": 5.3388,
+      "learning_rate": 8.186314921681781e-05,
+      "loss": 3.0171,
       "step": 1380
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008190436933223413,
-      "loss": 5.0211,
+      "learning_rate": 8.184253915910966e-05,
+      "loss": 2.7733,
       "step": 1381
     },
     {
       "epoch": 0.77,
-      "learning_rate": 0.0008188375927452597,
-      "loss": 5.4836,
+      "learning_rate": 8.182192910140149e-05,
+      "loss": 3.059,
       "step": 1382
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008186314921681781,
-      "loss": 5.3669,
+      "learning_rate": 8.180131904369333e-05,
+      "loss": 2.9153,
       "step": 1383
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008184253915910965,
-      "loss": 5.8945,
+      "learning_rate": 8.178070898598516e-05,
+      "loss": 3.2317,
       "step": 1384
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008182192910140148,
-      "loss": 5.9555,
+      "learning_rate": 8.1760098928277e-05,
+      "loss": 3.0819,
       "step": 1385
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008180131904369332,
-      "loss": 5.4759,
+      "learning_rate": 8.173948887056883e-05,
+      "loss": 3.0949,
       "step": 1386
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008178070898598516,
-      "loss": 5.5654,
+      "learning_rate": 8.171887881286068e-05,
+      "loss": 3.0843,
       "step": 1387
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.00081760098928277,
-      "loss": 5.5622,
+      "learning_rate": 8.169826875515252e-05,
+      "loss": 3.4474,
       "step": 1388
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008173948887056883,
-      "loss": 5.2607,
+      "learning_rate": 8.167765869744436e-05,
+      "loss": 2.779,
       "step": 1389
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008171887881286068,
-      "loss": 5.6258,
+      "learning_rate": 8.16570486397362e-05,
+      "loss": 2.9017,
       "step": 1390
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008169826875515252,
-      "loss": 5.1475,
+      "learning_rate": 8.163643858202802e-05,
+      "loss": 2.884,
       "step": 1391
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008167765869744436,
-      "loss": 5.625,
+      "learning_rate": 8.161582852431987e-05,
+      "loss": 2.8445,
       "step": 1392
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.000816570486397362,
-      "loss": 5.2315,
+      "learning_rate": 8.159521846661171e-05,
+      "loss": 2.8437,
       "step": 1393
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008163643858202803,
-      "loss": 5.6026,
+      "learning_rate": 8.157460840890355e-05,
+      "loss": 2.9174,
       "step": 1394
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008161582852431987,
-      "loss": 7.0047,
+      "learning_rate": 8.155399835119538e-05,
+      "loss": 2.8333,
       "step": 1395
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008159521846661171,
-      "loss": 5.2876,
+      "learning_rate": 8.153338829348723e-05,
+      "loss": 2.5981,
       "step": 1396
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008157460840890355,
-      "loss": 5.3983,
+      "learning_rate": 8.151277823577906e-05,
+      "loss": 2.9061,
       "step": 1397
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008155399835119538,
-      "loss": 5.1521,
+      "learning_rate": 8.14921681780709e-05,
+      "loss": 2.0009,
       "step": 1398
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008153338829348723,
-      "loss": 5.3206,
+      "learning_rate": 8.147155812036274e-05,
+      "loss": 2.2981,
       "step": 1399
     },
     {
       "epoch": 0.78,
-      "learning_rate": 0.0008151277823577907,
-      "loss": 5.5063,
+      "learning_rate": 8.145094806265457e-05,
+      "loss": 2.4077,
       "step": 1400
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008149216817807091,
-      "loss": 5.3094,
+      "learning_rate": 8.143033800494642e-05,
+      "loss": 3.2399,
       "step": 1401
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008147155812036275,
-      "loss": 5.8986,
+      "learning_rate": 8.140972794723826e-05,
+      "loss": 3.5673,
       "step": 1402
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008145094806265458,
-      "loss": 5.4165,
+      "learning_rate": 8.13891178895301e-05,
+      "loss": 3.3888,
       "step": 1403
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008143033800494642,
-      "loss": 6.0671,
+      "learning_rate": 8.136850783182193e-05,
+      "loss": 3.5187,
       "step": 1404
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008140972794723826,
-      "loss": 5.5996,
+      "learning_rate": 8.134789777411376e-05,
+      "loss": 3.3562,
       "step": 1405
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.000813891178895301,
-      "loss": 6.4383,
+      "learning_rate": 8.132728771640561e-05,
+      "loss": 3.9135,
       "step": 1406
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008136850783182194,
-      "loss": 5.5775,
+      "learning_rate": 8.130667765869745e-05,
+      "loss": 3.6068,
       "step": 1407
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008134789777411377,
-      "loss": 6.0582,
+      "learning_rate": 8.12860676009893e-05,
+      "loss": 3.3813,
       "step": 1408
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008132728771640561,
-      "loss": 5.7177,
+      "learning_rate": 8.126545754328112e-05,
+      "loss": 3.5282,
       "step": 1409
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008130667765869745,
-      "loss": 5.3655,
+      "learning_rate": 8.124484748557297e-05,
+      "loss": 3.2901,
       "step": 1410
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008128606760098929,
-      "loss": 5.3407,
+      "learning_rate": 8.12242374278648e-05,
+      "loss": 3.1886,
       "step": 1411
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008126545754328112,
-      "loss": 5.442,
+      "learning_rate": 8.120362737015664e-05,
+      "loss": 3.3587,
       "step": 1412
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008124484748557296,
-      "loss": 5.6227,
+      "learning_rate": 8.118301731244848e-05,
+      "loss": 3.5233,
       "step": 1413
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.000812242374278648,
-      "loss": 5.6585,
+      "learning_rate": 8.116240725474031e-05,
+      "loss": 3.4683,
       "step": 1414
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008120362737015664,
-      "loss": 5.3374,
+      "learning_rate": 8.114179719703216e-05,
+      "loss": 3.1649,
       "step": 1415
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008118301731244848,
-      "loss": 5.7209,
+      "learning_rate": 8.1121187139324e-05,
+      "loss": 3.2084,
       "step": 1416
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008116240725474031,
-      "loss": 5.796,
+      "learning_rate": 8.110057708161583e-05,
+      "loss": 3.2155,
       "step": 1417
     },
     {
       "epoch": 0.79,
-      "learning_rate": 0.0008114179719703215,
-      "loss": 6.1732,
+      "learning_rate": 8.107996702390766e-05,
+      "loss": 3.4439,
       "step": 1418
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.00081121187139324,
-      "loss": 5.9102,
+      "learning_rate": 8.10593569661995e-05,
+      "loss": 3.4847,
       "step": 1419
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008110057708161584,
-      "loss": 5.3661,
+      "learning_rate": 8.103874690849135e-05,
+      "loss": 3.1257,
       "step": 1420
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008107996702390767,
-      "loss": 5.8076,
+      "learning_rate": 8.101813685078319e-05,
+      "loss": 3.7286,
       "step": 1421
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008105935696619951,
-      "loss": 6.0575,
+      "learning_rate": 8.099752679307503e-05,
+      "loss": 3.6998,
       "step": 1422
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008103874690849135,
-      "loss": 5.3093,
+      "learning_rate": 8.097691673536686e-05,
+      "loss": 3.2532,
       "step": 1423
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008101813685078319,
-      "loss": 5.4761,
+      "learning_rate": 8.09563066776587e-05,
+      "loss": 3.0665,
       "step": 1424
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008099752679307503,
-      "loss": 6.57,
+      "learning_rate": 8.093569661995054e-05,
+      "loss": 3.5392,
       "step": 1425
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008097691673536686,
-      "loss": 5.6883,
+      "learning_rate": 8.091508656224238e-05,
+      "loss": 3.2926,
       "step": 1426
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.000809563066776587,
-      "loss": 5.2033,
+      "learning_rate": 8.089447650453421e-05,
+      "loss": 2.9858,
       "step": 1427
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008093569661995054,
-      "loss": 5.4502,
+      "learning_rate": 8.087386644682605e-05,
+      "loss": 3.262,
       "step": 1428
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008091508656224238,
-      "loss": 6.3243,
+      "learning_rate": 8.08532563891179e-05,
+      "loss": 3.3486,
       "step": 1429
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008089447650453421,
-      "loss": 6.9092,
+      "learning_rate": 8.083264633140974e-05,
+      "loss": 3.6262,
       "step": 1430
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008087386644682605,
-      "loss": 6.8114,
+      "learning_rate": 8.081203627370157e-05,
+      "loss": 3.6905,
       "step": 1431
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008085325638911789,
-      "loss": 5.8481,
+      "learning_rate": 8.07914262159934e-05,
+      "loss": 3.2646,
       "step": 1432
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008083264633140973,
-      "loss": 6.0423,
+      "learning_rate": 8.077081615828524e-05,
+      "loss": 3.3016,
       "step": 1433
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008081203627370157,
-      "loss": 5.5656,
+      "learning_rate": 8.075020610057709e-05,
+      "loss": 3.2069,
       "step": 1434
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.000807914262159934,
-      "loss": 5.8115,
+      "learning_rate": 8.072959604286893e-05,
+      "loss": 3.3416,
       "step": 1435
     },
     {
       "epoch": 0.8,
-      "learning_rate": 0.0008077081615828524,
-      "loss": 5.156,
+      "learning_rate": 8.070898598516077e-05,
+      "loss": 3.1426,
       "step": 1436
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008075020610057708,
-      "loss": 5.7272,
+      "learning_rate": 8.06883759274526e-05,
+      "loss": 3.1571,
       "step": 1437
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008072959604286892,
-      "loss": 5.2911,
+      "learning_rate": 8.066776586974443e-05,
+      "loss": 2.8164,
       "step": 1438
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008070898598516077,
-      "loss": 5.3085,
+      "learning_rate": 8.064715581203628e-05,
+      "loss": 2.9681,
       "step": 1439
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.000806883759274526,
-      "loss": 5.7923,
+      "learning_rate": 8.062654575432812e-05,
+      "loss": 2.9661,
       "step": 1440
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008066776586974444,
-      "loss": 6.611,
+      "learning_rate": 8.060593569661995e-05,
+      "loss": 3.2934,
       "step": 1441
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008064715581203628,
-      "loss": 5.3446,
+      "learning_rate": 8.058532563891179e-05,
+      "loss": 2.7117,
       "step": 1442
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008062654575432812,
-      "loss": 5.2329,
+      "learning_rate": 8.056471558120364e-05,
+      "loss": 2.709,
       "step": 1443
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008060593569661995,
-      "loss": 5.231,
+      "learning_rate": 8.054410552349547e-05,
+      "loss": 2.3517,
       "step": 1444
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008058532563891179,
-      "loss": 5.3781,
+      "learning_rate": 8.052349546578731e-05,
+      "loss": 2.3933,
       "step": 1445
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008056471558120363,
-      "loss": 5.4795,
+      "learning_rate": 8.050288540807914e-05,
+      "loss": 2.4474,
       "step": 1446
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008054410552349547,
-      "loss": 5.35,
+      "learning_rate": 8.048227535037098e-05,
+      "loss": 2.2044,
       "step": 1447
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008052349546578731,
-      "loss": 5.2862,
+      "learning_rate": 8.046166529266283e-05,
+      "loss": 2.0805,
       "step": 1448
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008050288540807914,
-      "loss": 5.5892,
+      "learning_rate": 8.044105523495467e-05,
+      "loss": 2.1408,
       "step": 1449
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008048227535037098,
-      "loss": 5.3857,
+      "learning_rate": 8.04204451772465e-05,
+      "loss": 2.1655,
       "step": 1450
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008046166529266282,
-      "loss": 5.6694,
+      "learning_rate": 8.039983511953834e-05,
+      "loss": 3.7964,
       "step": 1451
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008044105523495466,
-      "loss": 5.9201,
+      "learning_rate": 8.037922506183017e-05,
+      "loss": 3.6718,
       "step": 1452
     },
     {
       "epoch": 0.81,
-      "learning_rate": 0.0008042044517724649,
-      "loss": 5.3134,
+      "learning_rate": 8.035861500412202e-05,
+      "loss": 2.9885,
       "step": 1453
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008039983511953833,
-      "loss": 5.3417,
+      "learning_rate": 8.033800494641386e-05,
+      "loss": 3.1486,
       "step": 1454
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008037922506183017,
-      "loss": 6.4734,
+      "learning_rate": 8.031739488870569e-05,
+      "loss": 3.3839,
       "step": 1455
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008035861500412201,
-      "loss": 5.8254,
+      "learning_rate": 8.029678483099753e-05,
+      "loss": 3.4785,
       "step": 1456
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008033800494641385,
-      "loss": 5.8066,
+      "learning_rate": 8.027617477328938e-05,
+      "loss": 3.4434,
       "step": 1457
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008031739488870568,
-      "loss": 6.0904,
+      "learning_rate": 8.02555647155812e-05,
+      "loss": 3.5938,
       "step": 1458
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008029678483099752,
-      "loss": 5.6002,
+      "learning_rate": 8.023495465787304e-05,
+      "loss": 3.6236,
       "step": 1459
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008027617477328936,
-      "loss": 6.4152,
+      "learning_rate": 8.021434460016488e-05,
+      "loss": 3.8806,
       "step": 1460
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008025556471558121,
-      "loss": 5.9509,
+      "learning_rate": 8.019373454245672e-05,
+      "loss": 3.5919,
       "step": 1461
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008023495465787304,
-      "loss": 5.572,
+      "learning_rate": 8.017312448474857e-05,
+      "loss": 3.4612,
       "step": 1462
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008021434460016488,
-      "loss": 5.3032,
+      "learning_rate": 8.015251442704041e-05,
+      "loss": 3.1167,
       "step": 1463
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008019373454245672,
-      "loss": 5.6387,
+      "learning_rate": 8.013190436933224e-05,
+      "loss": 3.1937,
       "step": 1464
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008017312448474856,
-      "loss": 6.2012,
+      "learning_rate": 8.011129431162407e-05,
+      "loss": 3.5218,
       "step": 1465
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.000801525144270404,
-      "loss": 5.7121,
+      "learning_rate": 8.009068425391591e-05,
+      "loss": 3.5791,
       "step": 1466
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008013190436933223,
-      "loss": 5.4505,
+      "learning_rate": 8.007007419620776e-05,
+      "loss": 3.2411,
       "step": 1467
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008011129431162407,
-      "loss": 5.7396,
+      "learning_rate": 8.00494641384996e-05,
+      "loss": 3.4008,
       "step": 1468
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008009068425391591,
-      "loss": 5.7421,
+      "learning_rate": 8.002885408079143e-05,
+      "loss": 3.357,
       "step": 1469
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008007007419620775,
-      "loss": 5.261,
+      "learning_rate": 8.000824402308327e-05,
+      "loss": 3.0316,
       "step": 1470
     },
     {
       "epoch": 0.82,
-      "learning_rate": 0.0008004946413849959,
-      "loss": 5.6178,
+      "learning_rate": 7.99876339653751e-05,
+      "loss": 3.2246,
       "step": 1471
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0008002885408079142,
-      "loss": 5.2159,
+      "learning_rate": 7.996702390766695e-05,
+      "loss": 3.1647,
       "step": 1472
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0008000824402308326,
-      "loss": 5.3448,
+      "learning_rate": 7.994641384995877e-05,
+      "loss": 2.8797,
       "step": 1473
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.000799876339653751,
-      "loss": 5.3446,
+      "learning_rate": 7.992580379225062e-05,
+      "loss": 3.2394,
       "step": 1474
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007996702390766694,
-      "loss": 5.7379,
+      "learning_rate": 7.990519373454246e-05,
+      "loss": 3.147,
       "step": 1475
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007994641384995877,
-      "loss": 5.9696,
+      "learning_rate": 7.98845836768343e-05,
+      "loss": 3.3335,
       "step": 1476
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007992580379225061,
-      "loss": 5.6844,
+      "learning_rate": 7.986397361912615e-05,
+      "loss": 3.077,
       "step": 1477
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007990519373454245,
-      "loss": 5.2966,
+      "learning_rate": 7.984336356141798e-05,
+      "loss": 3.1496,
       "step": 1478
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007988458367683429,
-      "loss": 5.544,
+      "learning_rate": 7.982275350370981e-05,
+      "loss": 3.351,
       "step": 1479
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007986397361912613,
-      "loss": 5.6466,
+      "learning_rate": 7.980214344600165e-05,
+      "loss": 3.4357,
       "step": 1480
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007984336356141796,
-      "loss": 5.4104,
+      "learning_rate": 7.97815333882935e-05,
+      "loss": 3.1288,
       "step": 1481
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.000798227535037098,
-      "loss": 6.1775,
+      "learning_rate": 7.976092333058532e-05,
+      "loss": 3.455,
       "step": 1482
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007980214344600166,
-      "loss": 5.6859,
+      "learning_rate": 7.974031327287717e-05,
+      "loss": 3.3903,
       "step": 1483
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.000797815333882935,
-      "loss": 5.0078,
+      "learning_rate": 7.971970321516901e-05,
+      "loss": 2.8385,
       "step": 1484
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007976092333058533,
-      "loss": 5.4872,
+      "learning_rate": 7.969909315746084e-05,
+      "loss": 3.1595,
       "step": 1485
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007974031327287717,
-      "loss": 5.7126,
+      "learning_rate": 7.967848309975268e-05,
+      "loss": 3.2279,
       "step": 1486
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007971970321516901,
-      "loss": 5.5299,
+      "learning_rate": 7.965787304204451e-05,
+      "loss": 2.9567,
       "step": 1487
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007969909315746085,
-      "loss": 5.5282,
+      "learning_rate": 7.963726298433636e-05,
+      "loss": 2.8166,
       "step": 1488
     },
     {
       "epoch": 0.83,
-      "learning_rate": 0.0007967848309975269,
-      "loss": 5.8606,
+      "learning_rate": 7.96166529266282e-05,
+      "loss": 3.0881,
       "step": 1489
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007965787304204452,
-      "loss": 5.1812,
+      "learning_rate": 7.959604286892004e-05,
+      "loss": 2.8277,
       "step": 1490
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007963726298433636,
-      "loss": 5.7487,
+      "learning_rate": 7.957543281121187e-05,
+      "loss": 2.9971,
       "step": 1491
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.000796166529266282,
-      "loss": 5.6274,
+      "learning_rate": 7.95548227535037e-05,
+      "loss": 3.4556,
       "step": 1492
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007959604286892004,
-      "loss": 4.9282,
+      "learning_rate": 7.953421269579555e-05,
+      "loss": 2.9807,
       "step": 1493
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007957543281121187,
-      "loss": 5.6627,
+      "learning_rate": 7.951360263808739e-05,
+      "loss": 2.9906,
       "step": 1494
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007955482275350371,
-      "loss": 5.3832,
+      "learning_rate": 7.949299258037923e-05,
+      "loss": 2.7308,
       "step": 1495
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007953421269579555,
-      "loss": 6.0631,
+      "learning_rate": 7.947238252267106e-05,
+      "loss": 3.5605,
       "step": 1496
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007951360263808739,
-      "loss": 5.4315,
+      "learning_rate": 7.945177246496291e-05,
+      "loss": 2.7374,
       "step": 1497
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007949299258037923,
-      "loss": 5.7649,
+      "learning_rate": 7.943116240725475e-05,
+      "loss": 2.5697,
       "step": 1498
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007947238252267106,
-      "loss": 4.9583,
+      "learning_rate": 7.941055234954658e-05,
+      "loss": 1.9264,
       "step": 1499
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.000794517724649629,
-      "loss": 5.5664,
+      "learning_rate": 7.938994229183842e-05,
+      "loss": 2.3312,
       "step": 1500
     },
     {
       "epoch": 0.84,
-      "eval_loss": 5.693754196166992,
-      "eval_runtime": 1243.8508,
-      "eval_samples_per_second": 2.124,
-      "eval_steps_per_second": 0.266,
-      "eval_wer": 2.006644188813963,
+      "eval_loss": 5.697879314422607,
+      "eval_runtime": 1235.7249,
+      "eval_samples_per_second": 2.138,
+      "eval_steps_per_second": 0.268,
+      "eval_wer": 1.7344902816342722,
       "step": 1500
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007943116240725475,
-      "loss": 5.2739,
+      "learning_rate": 7.936933223413025e-05,
+      "loss": 3.4538,
       "step": 1501
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007941055234954659,
-      "loss": 5.5035,
+      "learning_rate": 7.93487221764221e-05,
+      "loss": 3.4086,
       "step": 1502
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007938994229183843,
-      "loss": 5.8292,
+      "learning_rate": 7.932811211871394e-05,
+      "loss": 3.6732,
       "step": 1503
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007936933223413026,
-      "loss": 5.799,
+      "learning_rate": 7.930750206100578e-05,
+      "loss": 3.3481,
       "step": 1504
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.000793487221764221,
-      "loss": 5.5344,
+      "learning_rate": 7.928689200329761e-05,
+      "loss": 3.5253,
       "step": 1505
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007932811211871394,
-      "loss": 5.7577,
+      "learning_rate": 7.926628194558944e-05,
+      "loss": 3.3051,
       "step": 1506
     },
     {
       "epoch": 0.84,
-      "learning_rate": 0.0007930750206100578,
-      "loss": 5.273,
+      "learning_rate": 7.924567188788129e-05,
+      "loss": 3.2636,
       "step": 1507
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007928689200329761,
-      "loss": 5.3651,
+      "learning_rate": 7.922506183017313e-05,
+      "loss": 3.263,
       "step": 1508
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007926628194558945,
-      "loss": 5.6284,
+      "learning_rate": 7.920445177246497e-05,
+      "loss": 3.5113,
       "step": 1509
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007924567188788129,
-      "loss": 5.3762,
+      "learning_rate": 7.91838417147568e-05,
+      "loss": 3.4268,
       "step": 1510
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007922506183017313,
-      "loss": 5.2086,
+      "learning_rate": 7.916323165704865e-05,
+      "loss": 3.1727,
       "step": 1511
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007920445177246497,
-      "loss": 5.4511,
+      "learning_rate": 7.914262159934048e-05,
+      "loss": 3.6533,
       "step": 1512
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.000791838417147568,
-      "loss": 5.1656,
+      "learning_rate": 7.912201154163232e-05,
+      "loss": 2.9669,
       "step": 1513
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007916323165704864,
-      "loss": 5.3216,
+      "learning_rate": 7.910140148392415e-05,
+      "loss": 3.1002,
       "step": 1514
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007914262159934048,
-      "loss": 5.236,
+      "learning_rate": 7.9080791426216e-05,
+      "loss": 2.9691,
       "step": 1515
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007912201154163232,
-      "loss": 6.1776,
+      "learning_rate": 7.906018136850784e-05,
+      "loss": 3.5831,
       "step": 1516
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007910140148392415,
-      "loss": 5.2409,
+      "learning_rate": 7.903957131079968e-05,
+      "loss": 2.9848,
       "step": 1517
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007908079142621599,
-      "loss": 5.4533,
+      "learning_rate": 7.901896125309151e-05,
+      "loss": 3.1764,
       "step": 1518
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007906018136850783,
-      "loss": 6.0416,
+      "learning_rate": 7.899835119538334e-05,
+      "loss": 3.772,
       "step": 1519
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007903957131079967,
-      "loss": 5.7621,
+      "learning_rate": 7.897774113767518e-05,
+      "loss": 3.1518,
       "step": 1520
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007901896125309152,
-      "loss": 5.3385,
+      "learning_rate": 7.895713107996703e-05,
+      "loss": 2.9528,
       "step": 1521
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007899835119538335,
-      "loss": 5.5288,
+      "learning_rate": 7.893652102225887e-05,
+      "loss": 3.3242,
       "step": 1522
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007897774113767519,
-      "loss": 5.8208,
+      "learning_rate": 7.891591096455071e-05,
+      "loss": 3.5752,
       "step": 1523
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007895713107996703,
-      "loss": 5.2498,
+      "learning_rate": 7.889530090684254e-05,
+      "loss": 3.106,
       "step": 1524
     },
     {
       "epoch": 0.85,
-      "learning_rate": 0.0007893652102225887,
-      "loss": 5.1569,
+      "learning_rate": 7.887469084913439e-05,
+      "loss": 3.0383,
       "step": 1525
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007891591096455071,
-      "loss": 5.0932,
+      "learning_rate": 7.885408079142622e-05,
+      "loss": 2.9221,
       "step": 1526
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007889530090684254,
-      "loss": 5.3862,
+      "learning_rate": 7.883347073371806e-05,
+      "loss": 2.7561,
       "step": 1527
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007887469084913438,
-      "loss": 5.2179,
+      "learning_rate": 7.881286067600989e-05,
+      "loss": 2.8084,
       "step": 1528
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007885408079142622,
-      "loss": 5.2247,
+      "learning_rate": 7.879225061830173e-05,
+      "loss": 3.1413,
       "step": 1529
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007883347073371806,
-      "loss": 5.2149,
+      "learning_rate": 7.877164056059358e-05,
+      "loss": 2.8562,
       "step": 1530
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007881286067600989,
-      "loss": 5.2661,
+      "learning_rate": 7.875103050288542e-05,
+      "loss": 2.9085,
       "step": 1531
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007879225061830173,
-      "loss": 6.1815,
+      "learning_rate": 7.873042044517725e-05,
+      "loss": 3.2306,
       "step": 1532
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007877164056059357,
-      "loss": 5.6746,
+      "learning_rate": 7.870981038746908e-05,
+      "loss": 3.2903,
       "step": 1533
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007875103050288541,
-      "loss": 5.2847,
+      "learning_rate": 7.868920032976092e-05,
+      "loss": 2.915,
       "step": 1534
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007873042044517725,
-      "loss": 6.6166,
+      "learning_rate": 7.866859027205277e-05,
+      "loss": 3.529,
       "step": 1535
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007870981038746908,
-      "loss": 5.2027,
+      "learning_rate": 7.864798021434461e-05,
+      "loss": 3.0042,
       "step": 1536
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007868920032976092,
-      "loss": 6.5368,
+      "learning_rate": 7.862737015663644e-05,
+      "loss": 3.5163,
       "step": 1537
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007866859027205276,
-      "loss": 5.7218,
+      "learning_rate": 7.860676009892828e-05,
+      "loss": 2.8488,
       "step": 1538
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.000786479802143446,
-      "loss": 5.4336,
+      "learning_rate": 7.858615004122011e-05,
+      "loss": 2.8413,
       "step": 1539
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007862737015663643,
-      "loss": 5.4167,
+      "learning_rate": 7.856553998351196e-05,
+      "loss": 2.9219,
       "step": 1540
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007860676009892827,
-      "loss": 5.5607,
+      "learning_rate": 7.85449299258038e-05,
+      "loss": 2.9648,
       "step": 1541
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007858615004122012,
-      "loss": 5.5747,
+      "learning_rate": 7.852431986809563e-05,
+      "loss": 3.6074,
       "step": 1542
     },
     {
       "epoch": 0.86,
-      "learning_rate": 0.0007856553998351196,
-      "loss": 5.5386,
+      "learning_rate": 7.850370981038747e-05,
+      "loss": 2.4697,
       "step": 1543
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.000785449299258038,
-      "loss": 5.7116,
+      "learning_rate": 7.848309975267932e-05,
+      "loss": 2.9236,
       "step": 1544
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007852431986809563,
-      "loss": 6.2746,
+      "learning_rate": 7.846248969497116e-05,
+      "loss": 3.0705,
       "step": 1545
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007850370981038747,
-      "loss": 5.2856,
+      "learning_rate": 7.844187963726299e-05,
+      "loss": 2.5837,
       "step": 1546
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007848309975267931,
-      "loss": 5.4119,
+      "learning_rate": 7.842126957955482e-05,
+      "loss": 2.3122,
       "step": 1547
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007846248969497115,
-      "loss": 5.4726,
+      "learning_rate": 7.840065952184666e-05,
+      "loss": 2.4937,
       "step": 1548
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007844187963726298,
-      "loss": 5.7192,
+      "learning_rate": 7.83800494641385e-05,
+      "loss": 2.2501,
       "step": 1549
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007842126957955482,
-      "loss": 5.7007,
+      "learning_rate": 7.835943940643035e-05,
+      "loss": 2.749,
       "step": 1550
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007840065952184666,
-      "loss": 5.7659,
+      "learning_rate": 7.833882934872218e-05,
+      "loss": 3.9368,
       "step": 1551
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.000783800494641385,
-      "loss": 5.7072,
+      "learning_rate": 7.831821929101402e-05,
+      "loss": 4.0694,
       "step": 1552
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007835943940643034,
-      "loss": 5.3353,
+      "learning_rate": 7.829760923330585e-05,
+      "loss": 3.3222,
       "step": 1553
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007833882934872217,
-      "loss": 5.3389,
+      "learning_rate": 7.82769991755977e-05,
+      "loss": 3.1017,
       "step": 1554
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007831821929101401,
-      "loss": 5.7256,
+      "learning_rate": 7.825638911788954e-05,
+      "loss": 3.5761,
       "step": 1555
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007829760923330585,
-      "loss": 6.2199,
+      "learning_rate": 7.823577906018137e-05,
+      "loss": 4.0974,
       "step": 1556
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007827699917559769,
-      "loss": 5.7906,
+      "learning_rate": 7.821516900247321e-05,
+      "loss": 3.2618,
       "step": 1557
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007825638911788953,
-      "loss": 5.4976,
+      "learning_rate": 7.819455894476506e-05,
+      "loss": 3.3216,
       "step": 1558
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.0007823577906018136,
-      "loss": 5.2849,
+      "learning_rate": 7.817394888705689e-05,
+      "loss": 3.4782,
       "step": 1559
     },
     {
       "epoch": 0.87,
-      "learning_rate": 0.000782151690024732,
-      "loss": 5.7787,
+      "learning_rate": 7.815333882934872e-05,
+      "loss": 3.676,
       "step": 1560
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007819455894476504,
-      "loss": 5.7005,
+      "learning_rate": 7.813272877164056e-05,
+      "loss": 3.4738,
       "step": 1561
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007817394888705689,
-      "loss": 5.5949,
+      "learning_rate": 7.81121187139324e-05,
+      "loss": 3.1737,
       "step": 1562
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007815333882934872,
-      "loss": 5.8771,
+      "learning_rate": 7.809150865622425e-05,
+      "loss": 3.3426,
       "step": 1563
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007813272877164056,
-      "loss": 5.3363,
+      "learning_rate": 7.807089859851609e-05,
+      "loss": 2.9941,
       "step": 1564
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.000781121187139324,
-      "loss": 5.3779,
+      "learning_rate": 7.805028854080792e-05,
+      "loss": 3.0429,
       "step": 1565
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007809150865622424,
-      "loss": 5.6413,
+      "learning_rate": 7.802967848309975e-05,
+      "loss": 3.4133,
       "step": 1566
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007807089859851608,
-      "loss": 5.0905,
+      "learning_rate": 7.800906842539159e-05,
+      "loss": 2.8763,
       "step": 1567
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007805028854080791,
-      "loss": 5.4187,
+      "learning_rate": 7.798845836768344e-05,
+      "loss": 3.217,
       "step": 1568
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007802967848309975,
-      "loss": 5.4871,
+      "learning_rate": 7.796784830997527e-05,
+      "loss": 3.6656,
       "step": 1569
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007800906842539159,
-      "loss": 5.5974,
+      "learning_rate": 7.794723825226711e-05,
+      "loss": 3.6153,
       "step": 1570
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007798845836768343,
-      "loss": 5.3391,
+      "learning_rate": 7.792662819455895e-05,
+      "loss": 3.2866,
       "step": 1571
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007796784830997526,
-      "loss": 5.5727,
+      "learning_rate": 7.79060181368508e-05,
+      "loss": 3.4466,
       "step": 1572
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.000779472382522671,
-      "loss": 5.5227,
+      "learning_rate": 7.788540807914263e-05,
+      "loss": 3.3131,
       "step": 1573
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007792662819455894,
-      "loss": 5.0716,
+      "learning_rate": 7.786479802143445e-05,
+      "loss": 2.9055,
       "step": 1574
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007790601813685078,
-      "loss": 5.1044,
+      "learning_rate": 7.78441879637263e-05,
+      "loss": 2.9419,
       "step": 1575
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007788540807914263,
-      "loss": 5.5813,
+      "learning_rate": 7.782357790601814e-05,
+      "loss": 3.2531,
       "step": 1576
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.0007786479802143446,
-      "loss": 5.9337,
+      "learning_rate": 7.780296784830998e-05,
+      "loss": 3.2816,
       "step": 1577
     },
     {
       "epoch": 0.88,
-      "learning_rate": 0.000778441879637263,
-      "loss": 5.5009,
+      "learning_rate": 7.778235779060181e-05,
+      "loss": 3.0786,
       "step": 1578
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007782357790601814,
-      "loss": 5.4651,
+      "learning_rate": 7.776174773289366e-05,
+      "loss": 3.2706,
       "step": 1579
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007780296784830998,
-      "loss": 6.2407,
+      "learning_rate": 7.774113767518549e-05,
+      "loss": 3.431,
       "step": 1580
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007778235779060181,
-      "loss": 5.5154,
+      "learning_rate": 7.772052761747733e-05,
+      "loss": 3.3183,
       "step": 1581
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007776174773289366,
-      "loss": 5.5077,
+      "learning_rate": 7.769991755976917e-05,
+      "loss": 3.1281,
       "step": 1582
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.000777411376751855,
-      "loss": 5.4512,
+      "learning_rate": 7.7679307502061e-05,
+      "loss": 2.8294,
       "step": 1583
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007772052761747734,
-      "loss": 5.2939,
+      "learning_rate": 7.765869744435285e-05,
+      "loss": 2.9154,
       "step": 1584
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007769991755976918,
-      "loss": 5.384,
+      "learning_rate": 7.763808738664469e-05,
+      "loss": 3.202,
       "step": 1585
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007767930750206101,
-      "loss": 5.2142,
+      "learning_rate": 7.761747732893652e-05,
+      "loss": 2.8845,
       "step": 1586
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007765869744435285,
-      "loss": 5.6352,
+      "learning_rate": 7.759686727122836e-05,
+      "loss": 3.1867,
       "step": 1587
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007763808738664469,
-      "loss": 5.3955,
+      "learning_rate": 7.75762572135202e-05,
+      "loss": 2.9652,
       "step": 1588
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007761747732893653,
-      "loss": 5.6869,
+      "learning_rate": 7.755564715581204e-05,
+      "loss": 3.4287,
       "step": 1589
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007759686727122837,
-      "loss": 5.3159,
+      "learning_rate": 7.753503709810388e-05,
+      "loss": 2.9802,
       "step": 1590
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.000775762572135202,
-      "loss": 5.1971,
+      "learning_rate": 7.751442704039572e-05,
+      "loss": 2.5327,
       "step": 1591
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007755564715581204,
-      "loss": 5.1773,
+      "learning_rate": 7.749381698268755e-05,
+      "loss": 2.4904,
       "step": 1592
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007753503709810388,
-      "loss": 5.1063,
+      "learning_rate": 7.74732069249794e-05,
+      "loss": 2.7303,
       "step": 1593
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007751442704039572,
-      "loss": 5.1922,
+      "learning_rate": 7.745259686727123e-05,
+      "loss": 2.7308,
       "step": 1594
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007749381698268755,
-      "loss": 5.1557,
+      "learning_rate": 7.743198680956307e-05,
+      "loss": 2.4714,
       "step": 1595
     },
     {
       "epoch": 0.89,
-      "learning_rate": 0.0007747320692497939,
-      "loss": 5.214,
+      "learning_rate": 7.741137675185491e-05,
+      "loss": 2.0618,
       "step": 1596
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007745259686727123,
-      "loss": 5.5801,
+      "learning_rate": 7.739076669414674e-05,
+      "loss": 2.6978,
       "step": 1597
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007743198680956307,
-      "loss": 5.5975,
+      "learning_rate": 7.737015663643859e-05,
+      "loss": 2.5393,
       "step": 1598
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007741137675185491,
-      "loss": 6.3652,
+      "learning_rate": 7.734954657873043e-05,
+      "loss": 3.1506,
       "step": 1599
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007739076669414674,
-      "loss": 5.4861,
+      "learning_rate": 7.732893652102226e-05,
+      "loss": 1.8936,
       "step": 1600
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007737015663643858,
-      "loss": 5.6266,
+      "learning_rate": 7.730832646331409e-05,
+      "loss": 3.3482,
       "step": 1601
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007734954657873043,
-      "loss": 5.2021,
+      "learning_rate": 7.728771640560593e-05,
+      "loss": 3.019,
       "step": 1602
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007732893652102227,
-      "loss": 5.7848,
+      "learning_rate": 7.726710634789778e-05,
+      "loss": 3.3261,
       "step": 1603
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.000773083264633141,
-      "loss": 5.6969,
+      "learning_rate": 7.724649629018962e-05,
+      "loss": 3.3755,
       "step": 1604
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007728771640560594,
-      "loss": 5.2357,
+      "learning_rate": 7.722588623248146e-05,
+      "loss": 2.9082,
       "step": 1605
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007726710634789778,
-      "loss": 5.402,
+      "learning_rate": 7.72052761747733e-05,
+      "loss": 3.3312,
       "step": 1606
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007724649629018962,
-      "loss": 5.5232,
+      "learning_rate": 7.718466611706512e-05,
+      "loss": 3.317,
       "step": 1607
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007722588623248146,
-      "loss": 5.867,
+      "learning_rate": 7.716405605935697e-05,
+      "loss": 3.4738,
       "step": 1608
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007720527617477329,
-      "loss": 5.7561,
+      "learning_rate": 7.714344600164881e-05,
+      "loss": 3.7665,
       "step": 1609
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007718466611706513,
-      "loss": 5.5532,
+      "learning_rate": 7.712283594394065e-05,
+      "loss": 3.4377,
       "step": 1610
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007716405605935697,
-      "loss": 5.456,
+      "learning_rate": 7.710222588623248e-05,
+      "loss": 3.162,
       "step": 1611
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007714344600164881,
-      "loss": 5.6274,
+      "learning_rate": 7.708161582852433e-05,
+      "loss": 3.5091,
       "step": 1612
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007712283594394065,
-      "loss": 5.9085,
+      "learning_rate": 7.706100577081616e-05,
+      "loss": 3.6477,
       "step": 1613
     },
     {
       "epoch": 0.9,
-      "learning_rate": 0.0007710222588623248,
-      "loss": 5.1706,
+      "learning_rate": 7.7040395713108e-05,
+      "loss": 3.1953,
       "step": 1614
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007708161582852432,
-      "loss": 5.3254,
+      "learning_rate": 7.701978565539983e-05,
+      "loss": 3.1082,
       "step": 1615
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007706100577081616,
-      "loss": 5.4561,
+      "learning_rate": 7.699917559769167e-05,
+      "loss": 2.9454,
       "step": 1616
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.00077040395713108,
-      "loss": 5.5489,
+      "learning_rate": 7.697856553998352e-05,
+      "loss": 3.2647,
       "step": 1617
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007701978565539983,
-      "loss": 5.6313,
+      "learning_rate": 7.695795548227536e-05,
+      "loss": 3.2443,
       "step": 1618
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007699917559769167,
-      "loss": 5.306,
+      "learning_rate": 7.69373454245672e-05,
+      "loss": 3.0801,
       "step": 1619
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007697856553998351,
-      "loss": 5.2442,
+      "learning_rate": 7.691673536685903e-05,
+      "loss": 2.7553,
       "step": 1620
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007695795548227535,
-      "loss": 6.3999,
+      "learning_rate": 7.689612530915086e-05,
+      "loss": 3.3552,
       "step": 1621
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.000769373454245672,
-      "loss": 5.1559,
+      "learning_rate": 7.68755152514427e-05,
+      "loss": 2.9602,
       "step": 1622
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007691673536685903,
-      "loss": 5.4703,
+      "learning_rate": 7.685490519373455e-05,
+      "loss": 3.157,
       "step": 1623
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007689612530915087,
-      "loss": 5.5496,
+      "learning_rate": 7.683429513602638e-05,
+      "loss": 3.0363,
       "step": 1624
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007687551525144271,
-      "loss": 5.4868,
+      "learning_rate": 7.681368507831822e-05,
+      "loss": 3.2368,
       "step": 1625
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007685490519373455,
-      "loss": 5.6641,
+      "learning_rate": 7.679307502061007e-05,
+      "loss": 3.4525,
       "step": 1626
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007683429513602638,
-      "loss": 5.5336,
+      "learning_rate": 7.67724649629019e-05,
+      "loss": 2.8124,
       "step": 1627
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007681368507831822,
-      "loss": 5.2135,
+      "learning_rate": 7.675185490519374e-05,
+      "loss": 2.8334,
       "step": 1628
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007679307502061006,
-      "loss": 5.4933,
+      "learning_rate": 7.673124484748557e-05,
+      "loss": 3.4957,
       "step": 1629
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.000767724649629019,
-      "loss": 5.1853,
+      "learning_rate": 7.671063478977741e-05,
+      "loss": 3.0883,
       "step": 1630
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007675185490519374,
-      "loss": 5.1833,
+      "learning_rate": 7.669002473206926e-05,
+      "loss": 2.8393,
       "step": 1631
     },
     {
       "epoch": 0.91,
-      "learning_rate": 0.0007673124484748557,
-      "loss": 5.5151,
+      "learning_rate": 7.66694146743611e-05,
+      "loss": 2.8064,
       "step": 1632
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007671063478977741,
-      "loss": 6.6036,
+      "learning_rate": 7.664880461665293e-05,
+      "loss": 3.579,
       "step": 1633
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007669002473206925,
-      "loss": 5.2396,
+      "learning_rate": 7.662819455894476e-05,
+      "loss": 2.955,
       "step": 1634
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007666941467436109,
-      "loss": 5.3314,
+      "learning_rate": 7.66075845012366e-05,
+      "loss": 2.918,
       "step": 1635
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007664880461665292,
-      "loss": 6.0022,
+      "learning_rate": 7.658697444352845e-05,
+      "loss": 3.193,
       "step": 1636
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007662819455894476,
-      "loss": 5.1715,
+      "learning_rate": 7.656636438582029e-05,
+      "loss": 2.8457,
       "step": 1637
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.000766075845012366,
-      "loss": 6.3325,
+      "learning_rate": 7.654575432811212e-05,
+      "loss": 3.3749,
       "step": 1638
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007658697444352844,
-      "loss": 5.1207,
+      "learning_rate": 7.652514427040396e-05,
+      "loss": 3.0209,
       "step": 1639
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007656636438582028,
-      "loss": 5.6787,
+      "learning_rate": 7.65045342126958e-05,
+      "loss": 3.3038,
       "step": 1640
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007654575432811211,
-      "loss": 5.6104,
+      "learning_rate": 7.648392415498764e-05,
+      "loss": 3.0982,
       "step": 1641
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007652514427040395,
-      "loss": 5.3238,
+      "learning_rate": 7.646331409727948e-05,
+      "loss": 2.7396,
       "step": 1642
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.000765045342126958,
-      "loss": 5.3629,
+      "learning_rate": 7.644270403957131e-05,
+      "loss": 2.2752,
       "step": 1643
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007648392415498764,
-      "loss": 5.4534,
+      "learning_rate": 7.642209398186315e-05,
+      "loss": 2.509,
       "step": 1644
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007646331409727948,
-      "loss": 5.6934,
+      "learning_rate": 7.6401483924155e-05,
+      "loss": 2.2568,
       "step": 1645
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007644270403957131,
-      "loss": 5.7245,
+      "learning_rate": 7.638087386644684e-05,
+      "loss": 2.4189,
       "step": 1646
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007642209398186315,
-      "loss": 5.4302,
+      "learning_rate": 7.636026380873867e-05,
+      "loss": 2.6111,
       "step": 1647
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007640148392415499,
-      "loss": 5.5526,
+      "learning_rate": 7.63396537510305e-05,
+      "loss": 2.0193,
       "step": 1648
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007638087386644683,
-      "loss": 5.6798,
+      "learning_rate": 7.631904369332234e-05,
+      "loss": 1.8882,
       "step": 1649
     },
     {
       "epoch": 0.92,
-      "learning_rate": 0.0007636026380873866,
-      "loss": 5.3074,
+      "learning_rate": 7.629843363561419e-05,
+      "loss": 2.3659,
       "step": 1650
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.000763396537510305,
-      "loss": 6.9425,
+      "learning_rate": 7.627782357790603e-05,
+      "loss": 4.0651,
       "step": 1651
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007631904369332234,
-      "loss": 5.4832,
+      "learning_rate": 7.625721352019786e-05,
+      "loss": 3.5652,
       "step": 1652
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007629843363561418,
-      "loss": 5.6279,
+      "learning_rate": 7.62366034624897e-05,
+      "loss": 3.3533,
       "step": 1653
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007627782357790602,
-      "loss": 5.4551,
+      "learning_rate": 7.621599340478153e-05,
+      "loss": 3.3332,
       "step": 1654
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007625721352019785,
-      "loss": 5.146,
+      "learning_rate": 7.619538334707338e-05,
+      "loss": 2.9756,
       "step": 1655
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007623660346248969,
-      "loss": 5.4204,
+      "learning_rate": 7.61747732893652e-05,
+      "loss": 3.1764,
       "step": 1656
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007621599340478153,
-      "loss": 6.1318,
+      "learning_rate": 7.615416323165705e-05,
+      "loss": 3.3174,
       "step": 1657
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007619538334707337,
-      "loss": 5.4465,
+      "learning_rate": 7.613355317394889e-05,
+      "loss": 3.2898,
       "step": 1658
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.000761747732893652,
-      "loss": 5.6286,
+      "learning_rate": 7.611294311624074e-05,
+      "loss": 3.1073,
       "step": 1659
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007615416323165704,
-      "loss": 5.2668,
+      "learning_rate": 7.609233305853258e-05,
+      "loss": 3.1394,
       "step": 1660
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007613355317394888,
-      "loss": 5.377,
+      "learning_rate": 7.60717230008244e-05,
+      "loss": 3.2144,
       "step": 1661
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007611294311624072,
-      "loss": 5.299,
+      "learning_rate": 7.605111294311624e-05,
+      "loss": 3.5789,
       "step": 1662
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007609233305853257,
-      "loss": 5.1236,
+      "learning_rate": 7.603050288540808e-05,
+      "loss": 3.1249,
       "step": 1663
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.000760717230008244,
-      "loss": 5.4572,
+      "learning_rate": 7.600989282769993e-05,
+      "loss": 3.0774,
       "step": 1664
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007605111294311624,
-      "loss": 5.1316,
+      "learning_rate": 7.598928276999176e-05,
+      "loss": 3.0476,
       "step": 1665
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007603050288540808,
-      "loss": 5.3337,
+      "learning_rate": 7.59686727122836e-05,
+      "loss": 2.9519,
       "step": 1666
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007600989282769992,
-      "loss": 5.2521,
+      "learning_rate": 7.594806265457544e-05,
+      "loss": 2.9839,
       "step": 1667
     },
     {
       "epoch": 0.93,
-      "learning_rate": 0.0007598928276999175,
-      "loss": 5.8702,
+      "learning_rate": 7.592745259686727e-05,
+      "loss": 3.2761,
       "step": 1668
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007596867271228359,
-      "loss": 5.0664,
+      "learning_rate": 7.590684253915912e-05,
+      "loss": 3.0993,
       "step": 1669
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007594806265457544,
-      "loss": 5.5366,
+      "learning_rate": 7.588623248145094e-05,
+      "loss": 3.2316,
       "step": 1670
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007592745259686728,
-      "loss": 5.7692,
+      "learning_rate": 7.586562242374279e-05,
+      "loss": 3.4019,
       "step": 1671
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007590684253915912,
-      "loss": 5.7895,
+      "learning_rate": 7.584501236603463e-05,
+      "loss": 3.2851,
       "step": 1672
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007588623248145095,
-      "loss": 5.4638,
+      "learning_rate": 7.582440230832648e-05,
+      "loss": 3.3973,
       "step": 1673
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007586562242374279,
-      "loss": 5.4016,
+      "learning_rate": 7.58037922506183e-05,
+      "loss": 3.3503,
       "step": 1674
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007584501236603463,
-      "loss": 5.6734,
+      "learning_rate": 7.578318219291013e-05,
+      "loss": 3.0901,
       "step": 1675
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007582440230832647,
-      "loss": 5.2043,
+      "learning_rate": 7.576257213520198e-05,
+      "loss": 3.3132,
       "step": 1676
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007580379225061831,
-      "loss": 5.0709,
+      "learning_rate": 7.574196207749382e-05,
+      "loss": 2.7159,
       "step": 1677
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007578318219291014,
-      "loss": 5.1183,
+      "learning_rate": 7.572135201978566e-05,
+      "loss": 2.9766,
       "step": 1678
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007576257213520198,
-      "loss": 5.7026,
+      "learning_rate": 7.57007419620775e-05,
+      "loss": 3.1191,
       "step": 1679
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007574196207749382,
-      "loss": 5.6478,
+      "learning_rate": 7.568013190436934e-05,
+      "loss": 3.1617,
       "step": 1680
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007572135201978566,
-      "loss": 4.9254,
+      "learning_rate": 7.565952184666117e-05,
+      "loss": 2.9758,
       "step": 1681
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.000757007419620775,
-      "loss": 5.1116,
+      "learning_rate": 7.563891178895301e-05,
+      "loss": 2.9542,
       "step": 1682
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007568013190436934,
-      "loss": 5.3446,
+      "learning_rate": 7.561830173124485e-05,
+      "loss": 3.2752,
       "step": 1683
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007565952184666118,
-      "loss": 5.5533,
+      "learning_rate": 7.559769167353668e-05,
+      "loss": 3.2459,
       "step": 1684
     },
     {
       "epoch": 0.94,
-      "learning_rate": 0.0007563891178895302,
-      "loss": 5.3844,
+      "learning_rate": 7.557708161582853e-05,
+      "loss": 2.9652,
       "step": 1685
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007561830173124486,
-      "loss": 5.2712,
+      "learning_rate": 7.555647155812037e-05,
+      "loss": 3.0522,
       "step": 1686
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007559769167353669,
-      "loss": 5.4178,
+      "learning_rate": 7.553586150041221e-05,
+      "loss": 2.967,
       "step": 1687
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007557708161582853,
-      "loss": 5.7833,
+      "learning_rate": 7.551525144270404e-05,
+      "loss": 3.1136,
       "step": 1688
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007555647155812037,
-      "loss": 5.5157,
+      "learning_rate": 7.549464138499587e-05,
+      "loss": 2.9319,
       "step": 1689
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007553586150041221,
-      "loss": 5.03,
+      "learning_rate": 7.547403132728772e-05,
+      "loss": 3.3485,
       "step": 1690
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007551525144270404,
-      "loss": 5.6973,
+      "learning_rate": 7.545342126957956e-05,
+      "loss": 2.9551,
       "step": 1691
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007549464138499588,
-      "loss": 5.3835,
+      "learning_rate": 7.54328112118714e-05,
+      "loss": 2.5999,
       "step": 1692
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007547403132728772,
-      "loss": 5.5059,
+      "learning_rate": 7.541220115416323e-05,
+      "loss": 2.5568,
       "step": 1693
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007545342126957956,
-      "loss": 5.5242,
+      "learning_rate": 7.539159109645508e-05,
+      "loss": 3.0656,
       "step": 1694
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.000754328112118714,
-      "loss": 5.5533,
+      "learning_rate": 7.537098103874691e-05,
+      "loss": 3.0891,
       "step": 1695
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007541220115416323,
-      "loss": 7.746,
+      "learning_rate": 7.535037098103875e-05,
+      "loss": 2.8748,
       "step": 1696
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007539159109645507,
-      "loss": 5.1658,
+      "learning_rate": 7.532976092333058e-05,
+      "loss": 1.9678,
       "step": 1697
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007537098103874691,
-      "loss": 5.7214,
+      "learning_rate": 7.530915086562242e-05,
+      "loss": 2.2389,
       "step": 1698
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007535037098103875,
-      "loss": 5.5684,
+      "learning_rate": 7.528854080791427e-05,
+      "loss": 2.5172,
       "step": 1699
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007532976092333058,
-      "loss": 5.3929,
+      "learning_rate": 7.526793075020611e-05,
+      "loss": 1.9408,
       "step": 1700
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007530915086562242,
-      "loss": 6.357,
+      "learning_rate": 7.524732069249794e-05,
+      "loss": 4.031,
       "step": 1701
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.0007528854080791426,
-      "loss": 5.4245,
+      "learning_rate": 7.522671063478977e-05,
+      "loss": 3.3732,
       "step": 1702
     },
     {
       "epoch": 0.95,
-      "learning_rate": 0.000752679307502061,
-      "loss": 5.6498,
+      "learning_rate": 7.520610057708161e-05,
+      "loss": 3.5464,
       "step": 1703
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007524732069249795,
-      "loss": 5.4691,
+      "learning_rate": 7.518549051937346e-05,
+      "loss": 3.0412,
       "step": 1704
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007522671063478978,
-      "loss": 5.4461,
+      "learning_rate": 7.51648804616653e-05,
+      "loss": 3.2815,
       "step": 1705
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007520610057708162,
-      "loss": 5.2842,
+      "learning_rate": 7.514427040395714e-05,
+      "loss": 3.6983,
       "step": 1706
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007518549051937346,
-      "loss": 5.3931,
+      "learning_rate": 7.512366034624897e-05,
+      "loss": 3.313,
       "step": 1707
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.000751648804616653,
-      "loss": 5.5527,
+      "learning_rate": 7.510305028854082e-05,
+      "loss": 3.4348,
       "step": 1708
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007514427040395714,
-      "loss": 5.9373,
+      "learning_rate": 7.508244023083265e-05,
+      "loss": 3.4614,
       "step": 1709
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007512366034624897,
-      "loss": 5.7114,
+      "learning_rate": 7.506183017312449e-05,
+      "loss": 3.1853,
       "step": 1710
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007510305028854081,
-      "loss": 5.0905,
+      "learning_rate": 7.504122011541632e-05,
+      "loss": 3.0735,
       "step": 1711
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007508244023083265,
-      "loss": 5.7914,
+      "learning_rate": 7.502061005770816e-05,
+      "loss": 3.3023,
       "step": 1712
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007506183017312449,
-      "loss": 5.8535,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 3.7714,
       "step": 1713
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007504122011541632,
-      "loss": 5.4343,
+      "learning_rate": 7.497938994229185e-05,
+      "loss": 2.9373,
       "step": 1714
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007502061005770816,
-      "loss": 5.2519,
+      "learning_rate": 7.495877988458368e-05,
+      "loss": 3.1659,
       "step": 1715
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.00075,
-      "loss": 5.8742,
+      "learning_rate": 7.493816982687551e-05,
+      "loss": 3.3905,
       "step": 1716
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007497938994229184,
-      "loss": 6.315,
+      "learning_rate": 7.491755976916735e-05,
+      "loss": 3.7101,
       "step": 1717
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007495877988458368,
-      "loss": 6.4051,
+      "learning_rate": 7.48969497114592e-05,
+      "loss": 3.7732,
       "step": 1718
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007493816982687551,
-      "loss": 5.6608,
+      "learning_rate": 7.487633965375104e-05,
+      "loss": 2.9104,
       "step": 1719
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007491755976916735,
-      "loss": 5.3271,
+      "learning_rate": 7.485572959604287e-05,
+      "loss": 3.2561,
       "step": 1720
     },
     {
       "epoch": 0.96,
-      "learning_rate": 0.0007489694971145919,
-      "loss": 5.3743,
+      "learning_rate": 7.483511953833471e-05,
+      "loss": 3.2719,
       "step": 1721
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007487633965375103,
-      "loss": 5.8151,
+      "learning_rate": 7.481450948062654e-05,
+      "loss": 3.1006,
       "step": 1722
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007485572959604286,
-      "loss": 5.7325,
+      "learning_rate": 7.479389942291839e-05,
+      "loss": 3.3313,
       "step": 1723
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.000748351195383347,
-      "loss": 5.299,
+      "learning_rate": 7.477328936521023e-05,
+      "loss": 3.2769,
       "step": 1724
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007481450948062655,
-      "loss": 5.4698,
+      "learning_rate": 7.475267930750206e-05,
+      "loss": 3.202,
       "step": 1725
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007479389942291839,
-      "loss": 5.2479,
+      "learning_rate": 7.47320692497939e-05,
+      "loss": 3.1616,
       "step": 1726
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007477328936521023,
-      "loss": 5.463,
+      "learning_rate": 7.471145919208575e-05,
+      "loss": 3.2184,
       "step": 1727
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007475267930750206,
-      "loss": 5.4187,
+      "learning_rate": 7.469084913437758e-05,
+      "loss": 3.3517,
       "step": 1728
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.000747320692497939,
-      "loss": 5.8931,
+      "learning_rate": 7.467023907666942e-05,
+      "loss": 3.737,
       "step": 1729
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007471145919208574,
-      "loss": 5.362,
+      "learning_rate": 7.464962901896125e-05,
+      "loss": 2.9768,
       "step": 1730
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007469084913437758,
-      "loss": 4.8648,
+      "learning_rate": 7.462901896125309e-05,
+      "loss": 2.9849,
       "step": 1731
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007467023907666942,
-      "loss": 5.1222,
+      "learning_rate": 7.460840890354494e-05,
+      "loss": 2.9426,
       "step": 1732
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007464962901896125,
-      "loss": 5.3404,
+      "learning_rate": 7.458779884583678e-05,
+      "loss": 2.9584,
       "step": 1733
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007462901896125309,
-      "loss": 5.6978,
+      "learning_rate": 7.456718878812861e-05,
+      "loss": 3.1584,
       "step": 1734
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007460840890354493,
-      "loss": 6.3196,
+      "learning_rate": 7.454657873042045e-05,
+      "loss": 3.5259,
       "step": 1735
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007458779884583677,
-      "loss": 6.147,
+      "learning_rate": 7.452596867271228e-05,
+      "loss": 3.7954,
       "step": 1736
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.000745671887881286,
-      "loss": 5.0887,
+      "learning_rate": 7.450535861500413e-05,
+      "loss": 2.8787,
       "step": 1737
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007454657873042044,
-      "loss": 5.7822,
+      "learning_rate": 7.448474855729597e-05,
+      "loss": 3.5244,
       "step": 1738
     },
     {
       "epoch": 0.97,
-      "learning_rate": 0.0007452596867271228,
-      "loss": 5.5801,
+      "learning_rate": 7.44641384995878e-05,
+      "loss": 3.1715,
       "step": 1739
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007450535861500412,
-      "loss": 5.4264,
+      "learning_rate": 7.444352844187964e-05,
+      "loss": 2.9337,
       "step": 1740
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007448474855729596,
-      "loss": 5.5353,
+      "learning_rate": 7.442291838417149e-05,
+      "loss": 3.217,
       "step": 1741
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007446413849958779,
-      "loss": 5.3254,
+      "learning_rate": 7.440230832646332e-05,
+      "loss": 2.8567,
       "step": 1742
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007444352844187963,
-      "loss": 5.1066,
+      "learning_rate": 7.438169826875515e-05,
+      "loss": 2.6575,
       "step": 1743
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007442291838417148,
-      "loss": 5.354,
+      "learning_rate": 7.436108821104699e-05,
+      "loss": 2.7415,
       "step": 1744
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007440230832646332,
-      "loss": 5.5587,
+      "learning_rate": 7.434047815333883e-05,
+      "loss": 2.996,
       "step": 1745
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007438169826875515,
-      "loss": 5.246,
+      "learning_rate": 7.431986809563068e-05,
+      "loss": 2.4628,
       "step": 1746
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007436108821104699,
-      "loss": 5.4677,
+      "learning_rate": 7.429925803792252e-05,
+      "loss": 2.3106,
       "step": 1747
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007434047815333883,
-      "loss": 5.0217,
+      "learning_rate": 7.427864798021435e-05,
+      "loss": 1.9535,
       "step": 1748
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007431986809563067,
-      "loss": 5.1716,
+      "learning_rate": 7.425803792250618e-05,
+      "loss": 2.0529,
       "step": 1749
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007429925803792251,
-      "loss": 5.1908,
+      "learning_rate": 7.423742786479802e-05,
+      "loss": 2.0671,
       "step": 1750
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007427864798021434,
-      "loss": 5.039,
+      "learning_rate": 7.421681780708987e-05,
+      "loss": 3.2709,
       "step": 1751
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007425803792250618,
-      "loss": 5.1832,
+      "learning_rate": 7.41962077493817e-05,
+      "loss": 3.3108,
       "step": 1752
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007423742786479802,
-      "loss": 5.5627,
+      "learning_rate": 7.417559769167354e-05,
+      "loss": 3.0689,
       "step": 1753
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007421681780708986,
-      "loss": 5.8069,
+      "learning_rate": 7.415498763396538e-05,
+      "loss": 3.5592,
       "step": 1754
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007419620774938169,
-      "loss": 5.5625,
+      "learning_rate": 7.413437757625723e-05,
+      "loss": 3.2694,
       "step": 1755
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007417559769167353,
-      "loss": 5.3716,
+      "learning_rate": 7.411376751854906e-05,
+      "loss": 3.2416,
       "step": 1756
     },
     {
       "epoch": 0.98,
-      "learning_rate": 0.0007415498763396537,
-      "loss": 5.1685,
+      "learning_rate": 7.409315746084089e-05,
+      "loss": 3.1403,
       "step": 1757
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007413437757625721,
-      "loss": 5.0729,
+      "learning_rate": 7.407254740313273e-05,
+      "loss": 3.0076,
       "step": 1758
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007411376751854905,
-      "loss": 5.6822,
+      "learning_rate": 7.405193734542457e-05,
+      "loss": 3.5737,
       "step": 1759
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007409315746084088,
-      "loss": 5.5346,
+      "learning_rate": 7.403132728771642e-05,
+      "loss": 3.3459,
       "step": 1760
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007407254740313272,
-      "loss": 5.764,
+      "learning_rate": 7.401071723000826e-05,
+      "loss": 3.2012,
       "step": 1761
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007405193734542456,
-      "loss": 5.0768,
+      "learning_rate": 7.399010717230009e-05,
+      "loss": 3.0604,
       "step": 1762
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007403132728771642,
-      "loss": 5.4564,
+      "learning_rate": 7.396949711459192e-05,
+      "loss": 3.0589,
       "step": 1763
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007401071723000826,
-      "loss": 5.6846,
+      "learning_rate": 7.394888705688376e-05,
+      "loss": 3.4432,
       "step": 1764
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007399010717230009,
-      "loss": 5.3839,
+      "learning_rate": 7.39282769991756e-05,
+      "loss": 3.2637,
       "step": 1765
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007396949711459193,
-      "loss": 5.5391,
+      "learning_rate": 7.390766694146744e-05,
+      "loss": 3.0606,
       "step": 1766
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007394888705688377,
-      "loss": 5.1325,
+      "learning_rate": 7.388705688375928e-05,
+      "loss": 2.7566,
       "step": 1767
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007392827699917561,
-      "loss": 5.3311,
+      "learning_rate": 7.386644682605112e-05,
+      "loss": 3.5617,
       "step": 1768
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007390766694146744,
-      "loss": 5.3466,
+      "learning_rate": 7.384583676834295e-05,
+      "loss": 3.4273,
       "step": 1769
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007388705688375928,
-      "loss": 5.223,
+      "learning_rate": 7.38252267106348e-05,
+      "loss": 3.1839,
       "step": 1770
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007386644682605112,
-      "loss": 5.2403,
+      "learning_rate": 7.380461665292662e-05,
+      "loss": 2.9087,
       "step": 1771
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007384583676834296,
-      "loss": 5.1288,
+      "learning_rate": 7.378400659521847e-05,
+      "loss": 2.6594,
       "step": 1772
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.000738252267106348,
-      "loss": 5.2811,
+      "learning_rate": 7.376339653751031e-05,
+      "loss": 2.8594,
       "step": 1773
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007380461665292663,
-      "loss": 5.3473,
+      "learning_rate": 7.374278647980215e-05,
+      "loss": 2.989,
       "step": 1774
     },
     {
       "epoch": 0.99,
-      "learning_rate": 0.0007378400659521847,
-      "loss": 5.0818,
+      "learning_rate": 7.372217642209398e-05,
+      "loss": 2.489,
       "step": 1775
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007376339653751031,
-      "loss": 5.2832,
+      "learning_rate": 7.370156636438581e-05,
+      "loss": 2.7637,
       "step": 1776
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007374278647980215,
-      "loss": 5.9785,
+      "learning_rate": 7.368095630667766e-05,
+      "loss": 3.3548,
       "step": 1777
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007372217642209398,
-      "loss": 5.2032,
+      "learning_rate": 7.36603462489695e-05,
+      "loss": 2.6006,
       "step": 1778
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007370156636438582,
-      "loss": 6.1154,
+      "learning_rate": 7.363973619126134e-05,
+      "loss": 3.0656,
       "step": 1779
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007368095630667766,
-      "loss": 5.2006,
+      "learning_rate": 7.361912613355317e-05,
+      "loss": 2.4469,
       "step": 1780
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.000736603462489695,
-      "loss": 6.3009,
+      "learning_rate": 7.359851607584502e-05,
+      "loss": 3.1273,
       "step": 1781
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007363973619126134,
-      "loss": 5.3086,
+      "learning_rate": 7.357790601813686e-05,
+      "loss": 1.9515,
       "step": 1782
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007361912613355317,
-      "loss": 5.3209,
+      "learning_rate": 7.355729596042869e-05,
+      "loss": 1.874,
       "step": 1783
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007359851607584502,
-      "loss": 4.4866,
+      "learning_rate": 7.353668590272052e-05,
+      "loss": 1.6989,
       "step": 1784
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007357790601813686,
-      "loss": 6.7161,
+      "learning_rate": 7.351607584501236e-05,
+      "loss": 4.1816,
       "step": 1785
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.000735572959604287,
-      "loss": 6.2056,
+      "learning_rate": 7.349546578730421e-05,
+      "loss": 3.7863,
       "step": 1786
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007353668590272053,
-      "loss": 5.1341,
+      "learning_rate": 7.347485572959605e-05,
+      "loss": 3.2488,
       "step": 1787
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007351607584501237,
-      "loss": 5.0733,
+      "learning_rate": 7.34542456718879e-05,
+      "loss": 3.0428,
       "step": 1788
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007349546578730421,
-      "loss": 5.3742,
+      "learning_rate": 7.343363561417972e-05,
+      "loss": 3.4786,
       "step": 1789
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007347485572959605,
-      "loss": 5.6744,
+      "learning_rate": 7.341302555647155e-05,
+      "loss": 3.4836,
       "step": 1790
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007345424567188789,
-      "loss": 5.0179,
+      "learning_rate": 7.33924154987634e-05,
+      "loss": 3.1326,
       "step": 1791
     },
     {
       "epoch": 1.0,
-      "learning_rate": 0.0007343363561417972,
-      "loss": 4.9713,
+      "learning_rate": 7.337180544105524e-05,
+      "loss": 3.0629,
       "step": 1792
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007341302555647156,
-      "loss": 5.4606,
+      "learning_rate": 7.335119538334708e-05,
+      "loss": 3.3974,
       "step": 1793
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.000733924154987634,
-      "loss": 4.9745,
+      "learning_rate": 7.333058532563891e-05,
+      "loss": 2.9951,
       "step": 1794
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007337180544105524,
-      "loss": 5.1346,
+      "learning_rate": 7.330997526793076e-05,
+      "loss": 3.1711,
       "step": 1795
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007335119538334708,
-      "loss": 5.2538,
+      "learning_rate": 7.328936521022259e-05,
+      "loss": 3.3147,
       "step": 1796
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007333058532563891,
-      "loss": 4.751,
+      "learning_rate": 7.326875515251443e-05,
+      "loss": 2.6349,
       "step": 1797
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007330997526793075,
-      "loss": 5.1878,
+      "learning_rate": 7.324814509480626e-05,
+      "loss": 3.0513,
       "step": 1798
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007328936521022259,
-      "loss": 4.9865,
+      "learning_rate": 7.32275350370981e-05,
+      "loss": 3.0436,
       "step": 1799
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007326875515251443,
-      "loss": 5.3912,
+      "learning_rate": 7.320692497938995e-05,
+      "loss": 3.402,
       "step": 1800
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007324814509480626,
-      "loss": 5.1692,
+      "learning_rate": 7.318631492168179e-05,
+      "loss": 3.1258,
       "step": 1801
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.000732275350370981,
-      "loss": 4.8574,
+      "learning_rate": 7.316570486397363e-05,
+      "loss": 2.7646,
       "step": 1802
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007320692497938994,
-      "loss": 5.0755,
+      "learning_rate": 7.314509480626546e-05,
+      "loss": 2.9772,
       "step": 1803
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007318631492168179,
-      "loss": 5.7155,
+      "learning_rate": 7.31244847485573e-05,
+      "loss": 3.1625,
       "step": 1804
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007316570486397363,
-      "loss": 4.696,
+      "learning_rate": 7.310387469084914e-05,
+      "loss": 2.7461,
       "step": 1805
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007314509480626546,
-      "loss": 5.0782,
+      "learning_rate": 7.308326463314098e-05,
+      "loss": 2.9568,
       "step": 1806
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.000731244847485573,
-      "loss": 4.7626,
+      "learning_rate": 7.306265457543281e-05,
+      "loss": 2.702,
       "step": 1807
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007310387469084914,
-      "loss": 5.4785,
+      "learning_rate": 7.304204451772465e-05,
+      "loss": 3.4841,
       "step": 1808
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007308326463314098,
-      "loss": 5.4731,
+      "learning_rate": 7.30214344600165e-05,
+      "loss": 2.9658,
       "step": 1809
     },
     {
       "epoch": 1.01,
-      "learning_rate": 0.0007306265457543281,
-      "loss": 4.8763,
+      "learning_rate": 7.300082440230833e-05,
+      "loss": 2.9049,
       "step": 1810
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007304204451772465,
-      "loss": 5.1314,
+      "learning_rate": 7.298021434460017e-05,
+      "loss": 2.9054,
       "step": 1811
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007302143446001649,
-      "loss": 6.0777,
+      "learning_rate": 7.2959604286892e-05,
+      "loss": 3.5595,
       "step": 1812
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007300082440230833,
-      "loss": 4.9886,
+      "learning_rate": 7.293899422918384e-05,
+      "loss": 3.0327,
       "step": 1813
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007298021434460017,
-      "loss": 4.5431,
+      "learning_rate": 7.291838417147569e-05,
+      "loss": 2.4342,
       "step": 1814
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.00072959604286892,
-      "loss": 5.2027,
+      "learning_rate": 7.289777411376753e-05,
+      "loss": 2.8203,
       "step": 1815
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007293899422918384,
-      "loss": 5.7955,
+      "learning_rate": 7.287716405605936e-05,
+      "loss": 3.0827,
       "step": 1816
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007291838417147568,
-      "loss": 5.5344,
+      "learning_rate": 7.285655399835119e-05,
+      "loss": 3.3496,
       "step": 1817
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007289777411376752,
-      "loss": 5.1074,
+      "learning_rate": 7.283594394064303e-05,
+      "loss": 2.674,
       "step": 1818
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007287716405605935,
-      "loss": 5.263,
+      "learning_rate": 7.281533388293488e-05,
+      "loss": 2.9444,
       "step": 1819
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007285655399835119,
-      "loss": 5.099,
+      "learning_rate": 7.279472382522672e-05,
+      "loss": 2.5668,
       "step": 1820
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007283594394064303,
-      "loss": 5.2997,
+      "learning_rate": 7.277411376751855e-05,
+      "loss": 2.9464,
       "step": 1821
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007281533388293487,
-      "loss": 4.9187,
+      "learning_rate": 7.27535037098104e-05,
+      "loss": 2.5676,
       "step": 1822
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007279472382522671,
-      "loss": 4.8737,
+      "learning_rate": 7.273289365210222e-05,
+      "loss": 2.6317,
       "step": 1823
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007277411376751854,
-      "loss": 5.8766,
+      "learning_rate": 7.271228359439407e-05,
+      "loss": 3.5075,
       "step": 1824
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007275350370981039,
-      "loss": 4.8703,
+      "learning_rate": 7.269167353668591e-05,
+      "loss": 2.7555,
       "step": 1825
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007273289365210223,
-      "loss": 5.8658,
+      "learning_rate": 7.267106347897774e-05,
+      "loss": 2.8276,
       "step": 1826
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007271228359439407,
-      "loss": 4.8,
+      "learning_rate": 7.265045342126958e-05,
+      "loss": 2.8721,
       "step": 1827
     },
     {
       "epoch": 1.02,
-      "learning_rate": 0.0007269167353668591,
-      "loss": 5.0083,
+      "learning_rate": 7.262984336356143e-05,
+      "loss": 2.516,
       "step": 1828
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007267106347897774,
-      "loss": 5.0207,
+      "learning_rate": 7.260923330585327e-05,
+      "loss": 2.7029,
       "step": 1829
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007265045342126958,
-      "loss": 5.5562,
+      "learning_rate": 7.25886232481451e-05,
+      "loss": 2.7159,
       "step": 1830
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007262984336356142,
-      "loss": 4.4522,
+      "learning_rate": 7.256801319043693e-05,
+      "loss": 1.6248,
       "step": 1831
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007260923330585326,
-      "loss": 4.9096,
+      "learning_rate": 7.254740313272877e-05,
+      "loss": 2.0523,
       "step": 1832
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007258862324814509,
-      "loss": 8.2927,
+      "learning_rate": 7.252679307502062e-05,
+      "loss": 2.6233,
       "step": 1833
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007256801319043693,
-      "loss": 4.8999,
+      "learning_rate": 7.250618301731246e-05,
+      "loss": 1.8069,
       "step": 1834
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007254740313272877,
-      "loss": 5.2236,
+      "learning_rate": 7.248557295960429e-05,
+      "loss": 3.2784,
       "step": 1835
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007252679307502061,
-      "loss": 5.3805,
+      "learning_rate": 7.246496290189613e-05,
+      "loss": 3.2294,
       "step": 1836
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007250618301731245,
-      "loss": 5.0225,
+      "learning_rate": 7.244435284418796e-05,
+      "loss": 3.1282,
       "step": 1837
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007248557295960428,
-      "loss": 5.0855,
+      "learning_rate": 7.24237427864798e-05,
+      "loss": 2.9912,
       "step": 1838
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007246496290189612,
-      "loss": 4.7451,
+      "learning_rate": 7.240313272877164e-05,
+      "loss": 2.8058,
       "step": 1839
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007244435284418796,
-      "loss": 5.0579,
+      "learning_rate": 7.238252267106348e-05,
+      "loss": 2.9596,
       "step": 1840
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.000724237427864798,
-      "loss": 5.1009,
+      "learning_rate": 7.236191261335532e-05,
+      "loss": 3.0251,
       "step": 1841
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007240313272877163,
-      "loss": 5.0356,
+      "learning_rate": 7.234130255564717e-05,
+      "loss": 3.2251,
       "step": 1842
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007238252267106347,
-      "loss": 5.2618,
+      "learning_rate": 7.2320692497939e-05,
+      "loss": 3.0343,
       "step": 1843
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007236191261335531,
-      "loss": 5.8649,
+      "learning_rate": 7.230008244023083e-05,
+      "loss": 3.7132,
       "step": 1844
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.0007234130255564716,
-      "loss": 5.5624,
+      "learning_rate": 7.227947238252267e-05,
+      "loss": 3.4502,
       "step": 1845
     },
     {
       "epoch": 1.03,
-      "learning_rate": 0.00072320692497939,
-      "loss": 4.8541,
+      "learning_rate": 7.225886232481451e-05,
+      "loss": 2.5899,
       "step": 1846
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007230008244023083,
-      "loss": 5.504,
+      "learning_rate": 7.223825226710636e-05,
+      "loss": 3.237,
       "step": 1847
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007227947238252267,
-      "loss": 4.9489,
+      "learning_rate": 7.22176422093982e-05,
+      "loss": 3.061,
       "step": 1848
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007225886232481451,
-      "loss": 5.0252,
+      "learning_rate": 7.219703215169003e-05,
+      "loss": 2.9103,
       "step": 1849
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007223825226710635,
-      "loss": 5.3229,
+      "learning_rate": 7.217642209398187e-05,
+      "loss": 3.3738,
       "step": 1850
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007221764220939819,
-      "loss": 4.7599,
+      "learning_rate": 7.21558120362737e-05,
+      "loss": 2.3419,
       "step": 1851
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007219703215169002,
-      "loss": 5.318,
+      "learning_rate": 7.213520197856555e-05,
+      "loss": 3.3845,
       "step": 1852
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007217642209398186,
-      "loss": 4.7916,
+      "learning_rate": 7.211459192085738e-05,
+      "loss": 2.7207,
       "step": 1853
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.000721558120362737,
-      "loss": 4.9588,
+      "learning_rate": 7.209398186314922e-05,
+      "loss": 3.1099,
       "step": 1854
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007213520197856554,
-      "loss": 5.0499,
+      "learning_rate": 7.207337180544106e-05,
+      "loss": 2.8808,
       "step": 1855
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007211459192085737,
-      "loss": 4.9607,
+      "learning_rate": 7.20527617477329e-05,
+      "loss": 2.9221,
       "step": 1856
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007209398186314922,
-      "loss": 4.879,
+      "learning_rate": 7.203215169002474e-05,
+      "loss": 2.5701,
       "step": 1857
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007207337180544106,
-      "loss": 4.9978,
+      "learning_rate": 7.201154163231657e-05,
+      "loss": 2.8489,
       "step": 1858
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.000720527617477329,
-      "loss": 4.8292,
+      "learning_rate": 7.199093157460841e-05,
+      "loss": 3.0572,
       "step": 1859
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007203215169002474,
-      "loss": 4.8893,
+      "learning_rate": 7.197032151690025e-05,
+      "loss": 2.4509,
       "step": 1860
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007201154163231657,
-      "loss": 4.6447,
+      "learning_rate": 7.19497114591921e-05,
+      "loss": 2.4266,
       "step": 1861
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007199093157460841,
-      "loss": 5.1184,
+      "learning_rate": 7.192910140148393e-05,
+      "loss": 2.8464,
       "step": 1862
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.0007197032151690025,
-      "loss": 4.8806,
+      "learning_rate": 7.190849134377577e-05,
+      "loss": 2.5182,
       "step": 1863
     },
     {
       "epoch": 1.04,
-      "learning_rate": 0.000719497114591921,
-      "loss": 5.1212,
+      "learning_rate": 7.18878812860676e-05,
+      "loss": 2.9924,
       "step": 1864
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007192910140148393,
-      "loss": 5.2627,
+      "learning_rate": 7.186727122835944e-05,
+      "loss": 2.8226,
       "step": 1865
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007190849134377577,
-      "loss": 4.5504,
+      "learning_rate": 7.184666117065129e-05,
+      "loss": 2.2502,
       "step": 1866
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007188788128606761,
-      "loss": 5.1228,
+      "learning_rate": 7.182605111294311e-05,
+      "loss": 2.5304,
       "step": 1867
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007186727122835945,
-      "loss": 4.8928,
+      "learning_rate": 7.180544105523496e-05,
+      "loss": 2.4761,
       "step": 1868
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007184666117065129,
-      "loss": 4.8402,
+      "learning_rate": 7.17848309975268e-05,
+      "loss": 3.0403,
       "step": 1869
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007182605111294312,
-      "loss": 5.755,
+      "learning_rate": 7.176422093981863e-05,
+      "loss": 3.178,
       "step": 1870
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007180544105523496,
-      "loss": 5.0116,
+      "learning_rate": 7.174361088211046e-05,
+      "loss": 2.6465,
       "step": 1871
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.000717848309975268,
-      "loss": 5.0736,
+      "learning_rate": 7.17230008244023e-05,
+      "loss": 2.6443,
       "step": 1872
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007176422093981864,
-      "loss": 4.6837,
+      "learning_rate": 7.170239076669415e-05,
+      "loss": 2.3674,
       "step": 1873
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007174361088211047,
-      "loss": 4.7889,
+      "learning_rate": 7.168178070898599e-05,
+      "loss": 2.3123,
       "step": 1874
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007172300082440231,
-      "loss": 5.0305,
+      "learning_rate": 7.166117065127783e-05,
+      "loss": 2.8213,
       "step": 1875
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007170239076669415,
-      "loss": 4.941,
+      "learning_rate": 7.164056059356966e-05,
+      "loss": 2.6576,
       "step": 1876
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007168178070898599,
-      "loss": 5.0131,
+      "learning_rate": 7.161995053586151e-05,
+      "loss": 2.5649,
       "step": 1877
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007166117065127783,
-      "loss": 5.0236,
+      "learning_rate": 7.159934047815334e-05,
+      "loss": 2.3321,
       "step": 1878
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007164056059356966,
-      "loss": 5.0443,
+      "learning_rate": 7.157873042044518e-05,
+      "loss": 2.0716,
       "step": 1879
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.000716199505358615,
-      "loss": 4.3573,
+      "learning_rate": 7.155812036273702e-05,
+      "loss": 1.5413,
       "step": 1880
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007159934047815334,
-      "loss": 4.7753,
+      "learning_rate": 7.153751030502885e-05,
+      "loss": 1.7893,
       "step": 1881
     },
     {
       "epoch": 1.05,
-      "learning_rate": 0.0007157873042044518,
-      "loss": 5.1045,
+      "learning_rate": 7.15169002473207e-05,
+      "loss": 2.1404,
       "step": 1882
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007155812036273702,
-      "loss": 4.3955,
+      "learning_rate": 7.149629018961254e-05,
+      "loss": 2.3528,
       "step": 1883
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007153751030502885,
-      "loss": 4.8774,
+      "learning_rate": 7.147568013190437e-05,
+      "loss": 1.8947,
       "step": 1884
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.000715169002473207,
-      "loss": 4.8209,
+      "learning_rate": 7.14550700741962e-05,
+      "loss": 3.2267,
       "step": 1885
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007149629018961254,
-      "loss": 4.8963,
+      "learning_rate": 7.143446001648804e-05,
+      "loss": 3.0389,
       "step": 1886
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007147568013190438,
-      "loss": 5.3112,
+      "learning_rate": 7.141384995877989e-05,
+      "loss": 3.2545,
       "step": 1887
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007145507007419621,
-      "loss": 5.2001,
+      "learning_rate": 7.139323990107173e-05,
+      "loss": 3.4465,
       "step": 1888
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007143446001648805,
-      "loss": 4.7737,
+      "learning_rate": 7.137262984336357e-05,
+      "loss": 2.6626,
       "step": 1889
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007141384995877989,
-      "loss": 5.4572,
+      "learning_rate": 7.13520197856554e-05,
+      "loss": 3.4539,
       "step": 1890
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007139323990107173,
-      "loss": 4.9596,
+      "learning_rate": 7.133140972794723e-05,
+      "loss": 2.717,
       "step": 1891
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007137262984336357,
-      "loss": 5.1131,
+      "learning_rate": 7.131079967023908e-05,
+      "loss": 2.7626,
       "step": 1892
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.000713520197856554,
-      "loss": 5.2216,
+      "learning_rate": 7.129018961253092e-05,
+      "loss": 3.0697,
       "step": 1893
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007133140972794724,
-      "loss": 4.8685,
+      "learning_rate": 7.126957955482275e-05,
+      "loss": 3.0628,
       "step": 1894
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007131079967023908,
-      "loss": 5.2497,
+      "learning_rate": 7.12489694971146e-05,
+      "loss": 3.1298,
       "step": 1895
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007129018961253092,
-      "loss": 4.9546,
+      "learning_rate": 7.122835943940644e-05,
+      "loss": 2.7858,
       "step": 1896
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007126957955482275,
-      "loss": 4.9165,
+      "learning_rate": 7.120774938169828e-05,
+      "loss": 2.5771,
       "step": 1897
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007124896949711459,
-      "loss": 5.1838,
+      "learning_rate": 7.118713932399011e-05,
+      "loss": 3.1384,
       "step": 1898
     },
     {
       "epoch": 1.06,
-      "learning_rate": 0.0007122835943940643,
-      "loss": 4.8545,
+      "learning_rate": 7.116652926628194e-05,
+      "loss": 2.5621,
       "step": 1899
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007120774938169827,
-      "loss": 4.8862,
+      "learning_rate": 7.114591920857378e-05,
+      "loss": 2.7812,
       "step": 1900
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007118713932399011,
-      "loss": 5.5258,
+      "learning_rate": 7.112530915086563e-05,
+      "loss": 3.311,
       "step": 1901
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007116652926628194,
-      "loss": 4.909,
+      "learning_rate": 7.110469909315747e-05,
+      "loss": 2.8178,
       "step": 1902
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007114591920857378,
-      "loss": 5.1176,
+      "learning_rate": 7.10840890354493e-05,
+      "loss": 2.7889,
       "step": 1903
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007112530915086562,
-      "loss": 5.3091,
+      "learning_rate": 7.106347897774114e-05,
+      "loss": 2.9363,
       "step": 1904
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007110469909315747,
-      "loss": 5.2653,
+      "learning_rate": 7.104286892003297e-05,
+      "loss": 2.7588,
       "step": 1905
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.000710840890354493,
-      "loss": 5.7553,
+      "learning_rate": 7.102225886232482e-05,
+      "loss": 3.2579,
       "step": 1906
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007106347897774114,
-      "loss": 5.3612,
+      "learning_rate": 7.100164880461666e-05,
+      "loss": 2.7745,
       "step": 1907
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007104286892003298,
-      "loss": 5.1875,
+      "learning_rate": 7.098103874690849e-05,
+      "loss": 3.0173,
       "step": 1908
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007102225886232482,
-      "loss": 4.9192,
+      "learning_rate": 7.096042868920033e-05,
+      "loss": 2.4572,
       "step": 1909
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007100164880461666,
-      "loss": 5.0839,
+      "learning_rate": 7.093981863149218e-05,
+      "loss": 3.1077,
       "step": 1910
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007098103874690849,
-      "loss": 5.1741,
+      "learning_rate": 7.091920857378401e-05,
+      "loss": 3.1584,
       "step": 1911
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007096042868920033,
-      "loss": 5.1112,
+      "learning_rate": 7.089859851607585e-05,
+      "loss": 2.9753,
       "step": 1912
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007093981863149217,
-      "loss": 4.9319,
+      "learning_rate": 7.087798845836768e-05,
+      "loss": 3.0491,
       "step": 1913
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007091920857378401,
-      "loss": 5.1723,
+      "learning_rate": 7.085737840065952e-05,
+      "loss": 3.0448,
       "step": 1914
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007089859851607585,
-      "loss": 5.0431,
+      "learning_rate": 7.083676834295137e-05,
+      "loss": 2.5982,
       "step": 1915
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007087798845836768,
-      "loss": 5.134,
+      "learning_rate": 7.081615828524321e-05,
+      "loss": 2.9252,
       "step": 1916
     },
     {
       "epoch": 1.07,
-      "learning_rate": 0.0007085737840065952,
-      "loss": 4.9249,
+      "learning_rate": 7.079554822753504e-05,
+      "loss": 2.5183,
       "step": 1917
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007083676834295136,
-      "loss": 5.2112,
+      "learning_rate": 7.077493816982687e-05,
+      "loss": 2.7719,
       "step": 1918
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.000708161582852432,
-      "loss": 5.1078,
+      "learning_rate": 7.075432811211871e-05,
+      "loss": 2.9904,
       "step": 1919
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007079554822753503,
-      "loss": 4.7082,
+      "learning_rate": 7.073371805441056e-05,
+      "loss": 2.5978,
       "step": 1920
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007077493816982687,
-      "loss": 4.741,
+      "learning_rate": 7.07131079967024e-05,
+      "loss": 2.5047,
       "step": 1921
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007075432811211871,
-      "loss": 5.6072,
+      "learning_rate": 7.069249793899423e-05,
+      "loss": 3.0642,
       "step": 1922
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007073371805441055,
-      "loss": 5.1473,
+      "learning_rate": 7.067188788128607e-05,
+      "loss": 2.6266,
       "step": 1923
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007071310799670239,
-      "loss": 4.9798,
+      "learning_rate": 7.065127782357792e-05,
+      "loss": 2.9404,
       "step": 1924
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007069249793899422,
-      "loss": 5.5453,
+      "learning_rate": 7.063066776586975e-05,
+      "loss": 2.977,
       "step": 1925
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007067188788128607,
-      "loss": 4.8624,
+      "learning_rate": 7.061005770816158e-05,
+      "loss": 2.2247,
       "step": 1926
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007065127782357791,
-      "loss": 4.7663,
+      "learning_rate": 7.058944765045342e-05,
+      "loss": 2.0434,
       "step": 1927
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007063066776586975,
-      "loss": 4.9708,
+      "learning_rate": 7.056883759274526e-05,
+      "loss": 2.1386,
       "step": 1928
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007061005770816158,
-      "loss": 4.6775,
+      "learning_rate": 7.05482275350371e-05,
+      "loss": 1.9126,
       "step": 1929
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007058944765045342,
-      "loss": 4.3418,
+      "learning_rate": 7.052761747732895e-05,
+      "loss": 1.7426,
       "step": 1930
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007056883759274526,
-      "loss": 4.2575,
+      "learning_rate": 7.050700741962078e-05,
+      "loss": 1.6179,
       "step": 1931
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.000705482275350371,
-      "loss": 6.3119,
+      "learning_rate": 7.048639736191261e-05,
+      "loss": 2.6271,
       "step": 1932
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007052761747732894,
-      "loss": 4.4464,
+      "learning_rate": 7.046578730420445e-05,
+      "loss": 1.7184,
       "step": 1933
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007050700741962077,
-      "loss": 6.336,
+      "learning_rate": 7.04451772464963e-05,
+      "loss": 2.1103,
       "step": 1934
     },
     {
       "epoch": 1.08,
-      "learning_rate": 0.0007048639736191261,
-      "loss": 5.2734,
+      "learning_rate": 7.042456718878814e-05,
+      "loss": 3.5248,
       "step": 1935
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007046578730420445,
-      "loss": 4.9526,
+      "learning_rate": 7.040395713107997e-05,
+      "loss": 3.1503,
       "step": 1936
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007044517724649629,
-      "loss": 5.1503,
+      "learning_rate": 7.038334707337181e-05,
+      "loss": 2.916,
       "step": 1937
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007042456718878813,
-      "loss": 6.6474,
+      "learning_rate": 7.036273701566364e-05,
+      "loss": 3.4541,
       "step": 1938
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007040395713107996,
-      "loss": 5.0745,
+      "learning_rate": 7.034212695795549e-05,
+      "loss": 2.9266,
       "step": 1939
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.000703833470733718,
-      "loss": 5.6552,
+      "learning_rate": 7.032151690024732e-05,
+      "loss": 3.2562,
       "step": 1940
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007036273701566364,
-      "loss": 4.8594,
+      "learning_rate": 7.030090684253916e-05,
+      "loss": 2.8126,
       "step": 1941
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007034212695795548,
-      "loss": 5.5089,
+      "learning_rate": 7.0280296784831e-05,
+      "loss": 3.0183,
       "step": 1942
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007032151690024731,
-      "loss": 5.8453,
+      "learning_rate": 7.025968672712285e-05,
+      "loss": 3.5855,
       "step": 1943
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007030090684253915,
-      "loss": 5.1221,
+      "learning_rate": 7.023907666941469e-05,
+      "loss": 2.9036,
       "step": 1944
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007028029678483099,
-      "loss": 5.1654,
+      "learning_rate": 7.021846661170652e-05,
+      "loss": 3.0927,
       "step": 1945
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007025968672712284,
-      "loss": 4.8595,
+      "learning_rate": 7.019785655399835e-05,
+      "loss": 3.0026,
       "step": 1946
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007023907666941468,
-      "loss": 5.4633,
+      "learning_rate": 7.017724649629019e-05,
+      "loss": 3.2426,
       "step": 1947
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007021846661170651,
-      "loss": 5.1242,
+      "learning_rate": 7.015663643858204e-05,
+      "loss": 2.8074,
       "step": 1948
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007019785655399835,
-      "loss": 5.2278,
+      "learning_rate": 7.013602638087387e-05,
+      "loss": 3.5624,
       "step": 1949
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.000701772464962902,
-      "loss": 4.8614,
+      "learning_rate": 7.011541632316571e-05,
+      "loss": 3.0746,
       "step": 1950
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007015663643858204,
-      "loss": 5.3163,
+      "learning_rate": 7.009480626545755e-05,
+      "loss": 3.371,
       "step": 1951
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007013602638087387,
-      "loss": 4.778,
+      "learning_rate": 7.007419620774938e-05,
+      "loss": 2.8209,
       "step": 1952
     },
     {
       "epoch": 1.09,
-      "learning_rate": 0.0007011541632316571,
-      "loss": 5.4249,
+      "learning_rate": 7.005358615004123e-05,
+      "loss": 3.2446,
       "step": 1953
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0007009480626545755,
-      "loss": 5.0991,
+      "learning_rate": 7.003297609233306e-05,
+      "loss": 2.9616,
       "step": 1954
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0007007419620774939,
-      "loss": 5.1253,
+      "learning_rate": 7.00123660346249e-05,
+      "loss": 3.0677,
       "step": 1955
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0007005358615004123,
-      "loss": 5.0108,
+      "learning_rate": 6.999175597691674e-05,
+      "loss": 3.0189,
       "step": 1956
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0007003297609233306,
-      "loss": 5.2935,
+      "learning_rate": 6.997114591920859e-05,
+      "loss": 2.906,
       "step": 1957
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.000700123660346249,
-      "loss": 5.1025,
+      "learning_rate": 6.995053586150042e-05,
+      "loss": 2.7705,
       "step": 1958
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006999175597691674,
-      "loss": 4.9619,
+      "learning_rate": 6.992992580379225e-05,
+      "loss": 2.4526,
       "step": 1959
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006997114591920858,
-      "loss": 4.8111,
+      "learning_rate": 6.990931574608409e-05,
+      "loss": 2.6455,
       "step": 1960
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006995053586150041,
-      "loss": 4.9499,
+      "learning_rate": 6.988870568837593e-05,
+      "loss": 2.4016,
       "step": 1961
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006992992580379225,
-      "loss": 4.6626,
+      "learning_rate": 6.986809563066778e-05,
+      "loss": 2.3033,
       "step": 1962
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006990931574608409,
-      "loss": 4.9205,
+      "learning_rate": 6.98474855729596e-05,
+      "loss": 2.6714,
       "step": 1963
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006988870568837593,
-      "loss": 5.0247,
+      "learning_rate": 6.982687551525145e-05,
+      "loss": 2.6984,
       "step": 1964
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006986809563066778,
-      "loss": 5.6156,
+      "learning_rate": 6.980626545754329e-05,
+      "loss": 3.3267,
       "step": 1965
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.000698474855729596,
-      "loss": 4.6375,
+      "learning_rate": 6.978565539983512e-05,
+      "loss": 2.3201,
       "step": 1966
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006982687551525145,
-      "loss": 4.965,
+      "learning_rate": 6.976504534212697e-05,
+      "loss": 2.7997,
       "step": 1967
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006980626545754329,
-      "loss": 4.5958,
+      "learning_rate": 6.97444352844188e-05,
+      "loss": 2.2409,
       "step": 1968
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006978565539983513,
-      "loss": 5.2138,
+      "learning_rate": 6.972382522671064e-05,
+      "loss": 2.6801,
       "step": 1969
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.0006976504534212697,
-      "loss": 4.931,
+      "learning_rate": 6.970321516900248e-05,
+      "loss": 2.3878,
       "step": 1970
     },
     {
       "epoch": 1.1,
-      "learning_rate": 0.000697444352844188,
-      "loss": 5.1836,
+      "learning_rate": 6.968260511129432e-05,
+      "loss": 2.5204,
       "step": 1971
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006972382522671064,
-      "loss": 4.9763,
+      "learning_rate": 6.966199505358615e-05,
+      "loss": 2.3114,
       "step": 1972
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006970321516900248,
-      "loss": 4.8877,
+      "learning_rate": 6.964138499587798e-05,
+      "loss": 2.1272,
       "step": 1973
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006968260511129432,
-      "loss": 4.745,
+      "learning_rate": 6.962077493816983e-05,
+      "loss": 2.4082,
       "step": 1974
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006966199505358615,
-      "loss": 5.3999,
+      "learning_rate": 6.960016488046167e-05,
+      "loss": 2.727,
       "step": 1975
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006964138499587799,
-      "loss": 4.8253,
+      "learning_rate": 6.957955482275351e-05,
+      "loss": 2.4629,
       "step": 1976
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006962077493816983,
-      "loss": 4.8059,
+      "learning_rate": 6.955894476504534e-05,
+      "loss": 2.2048,
       "step": 1977
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006960016488046167,
-      "loss": 4.557,
+      "learning_rate": 6.953833470733719e-05,
+      "loss": 1.7574,
       "step": 1978
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006957955482275351,
-      "loss": 4.3461,
+      "learning_rate": 6.951772464962902e-05,
+      "loss": 1.7625,
       "step": 1979
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006955894476504534,
-      "loss": 5.3581,
+      "learning_rate": 6.949711459192086e-05,
+      "loss": 2.1594,
       "step": 1980
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006953833470733718,
-      "loss": 5.1943,
+      "learning_rate": 6.947650453421269e-05,
+      "loss": 2.1918,
       "step": 1981
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006951772464962902,
-      "loss": 5.4875,
+      "learning_rate": 6.945589447650453e-05,
+      "loss": 2.4692,
       "step": 1982
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006949711459192086,
-      "loss": 4.9124,
+      "learning_rate": 6.943528441879638e-05,
+      "loss": 2.001,
       "step": 1983
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006947650453421269,
-      "loss": 4.2013,
+      "learning_rate": 6.941467436108822e-05,
+      "loss": 1.6058,
       "step": 1984
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006945589447650453,
-      "loss": 5.0194,
+      "learning_rate": 6.939406430338005e-05,
+      "loss": 3.312,
       "step": 1985
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006943528441879638,
-      "loss": 5.304,
+      "learning_rate": 6.937345424567188e-05,
+      "loss": 3.2338,
       "step": 1986
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006941467436108822,
-      "loss": 5.1742,
+      "learning_rate": 6.935284418796372e-05,
+      "loss": 3.21,
       "step": 1987
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006939406430338006,
-      "loss": 5.8334,
+      "learning_rate": 6.933223413025557e-05,
+      "loss": 3.3209,
       "step": 1988
     },
     {
       "epoch": 1.11,
-      "learning_rate": 0.0006937345424567189,
-      "loss": 5.0403,
+      "learning_rate": 6.931162407254741e-05,
+      "loss": 3.2838,
       "step": 1989
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006935284418796373,
-      "loss": 5.1716,
+      "learning_rate": 6.929101401483924e-05,
+      "loss": 3.1977,
       "step": 1990
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006933223413025557,
-      "loss": 5.2278,
+      "learning_rate": 6.927040395713108e-05,
+      "loss": 3.1661,
       "step": 1991
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006931162407254741,
-      "loss": 4.8494,
+      "learning_rate": 6.924979389942293e-05,
+      "loss": 2.97,
       "step": 1992
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006929101401483924,
-      "loss": 4.968,
+      "learning_rate": 6.922918384171476e-05,
+      "loss": 2.7911,
       "step": 1993
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006927040395713108,
-      "loss": 5.0556,
+      "learning_rate": 6.92085737840066e-05,
+      "loss": 3.2571,
       "step": 1994
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006924979389942292,
-      "loss": 5.0608,
+      "learning_rate": 6.918796372629843e-05,
+      "loss": 2.6035,
       "step": 1995
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006922918384171476,
-      "loss": 5.3535,
+      "learning_rate": 6.916735366859027e-05,
+      "loss": 2.91,
       "step": 1996
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.000692085737840066,
-      "loss": 5.0817,
+      "learning_rate": 6.914674361088212e-05,
+      "loss": 3.0103,
       "step": 1997
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006918796372629843,
-      "loss": 5.3395,
+      "learning_rate": 6.912613355317396e-05,
+      "loss": 3.7088,
       "step": 1998
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006916735366859027,
-      "loss": 5.0693,
+      "learning_rate": 6.910552349546579e-05,
+      "loss": 2.9388,
       "step": 1999
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006914674361088211,
-      "loss": 4.9162,
+      "learning_rate": 6.908491343775762e-05,
+      "loss": 2.8004,
       "step": 2000
     },
     {
       "epoch": 1.12,
-      "eval_loss": 5.641247272491455,
-      "eval_runtime": 1246.1483,
-      "eval_samples_per_second": 2.12,
-      "eval_steps_per_second": 0.266,
-      "eval_wer": 1.9732844109480365,
+      "eval_loss": 6.043630599975586,
+      "eval_runtime": 1238.531,
+      "eval_samples_per_second": 2.133,
+      "eval_steps_per_second": 0.267,
+      "eval_wer": 1.67869892899643,
       "step": 2000
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006912613355317395,
-      "loss": 5.1554,
+      "learning_rate": 6.906430338004946e-05,
+      "loss": 3.3267,
       "step": 2001
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006910552349546579,
-      "loss": 4.8891,
+      "learning_rate": 6.904369332234131e-05,
+      "loss": 3.0503,
       "step": 2002
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006908491343775762,
-      "loss": 4.9826,
+      "learning_rate": 6.902308326463315e-05,
+      "loss": 2.6245,
       "step": 2003
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006906430338004946,
-      "loss": 5.1483,
+      "learning_rate": 6.900247320692498e-05,
+      "loss": 3.284,
       "step": 2004
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.000690436933223413,
-      "loss": 4.7722,
+      "learning_rate": 6.898186314921682e-05,
+      "loss": 2.7719,
       "step": 2005
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006902308326463315,
-      "loss": 4.915,
+      "learning_rate": 6.896125309150865e-05,
+      "loss": 3.0409,
       "step": 2006
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0006900247320692497,
-      "loss": 4.7808,
+      "learning_rate": 6.89406430338005e-05,
+      "loss": 2.5627,
       "step": 2007
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006898186314921682,
-      "loss": 5.1847,
+      "learning_rate": 6.892003297609234e-05,
+      "loss": 3.0582,
       "step": 2008
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006896125309150866,
-      "loss": 4.9068,
+      "learning_rate": 6.889942291838417e-05,
+      "loss": 2.6723,
       "step": 2009
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.000689406430338005,
-      "loss": 4.7811,
+      "learning_rate": 6.887881286067601e-05,
+      "loss": 2.6391,
       "step": 2010
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006892003297609234,
-      "loss": 4.8156,
+      "learning_rate": 6.885820280296786e-05,
+      "loss": 2.6617,
       "step": 2011
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006889942291838417,
-      "loss": 5.0743,
+      "learning_rate": 6.88375927452597e-05,
+      "loss": 2.6289,
       "step": 2012
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006887881286067601,
-      "loss": 5.2864,
+      "learning_rate": 6.881698268755153e-05,
+      "loss": 2.6601,
       "step": 2013
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006885820280296785,
-      "loss": 5.6686,
+      "learning_rate": 6.879637262984336e-05,
+      "loss": 2.9384,
       "step": 2014
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006883759274525969,
-      "loss": 5.014,
+      "learning_rate": 6.87757625721352e-05,
+      "loss": 2.5677,
       "step": 2015
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006881698268755152,
-      "loss": 4.8173,
+      "learning_rate": 6.875515251442705e-05,
+      "loss": 2.4531,
       "step": 2016
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006879637262984336,
-      "loss": 5.0305,
+      "learning_rate": 6.873454245671889e-05,
+      "loss": 2.8312,
       "step": 2017
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.000687757625721352,
-      "loss": 5.1696,
+      "learning_rate": 6.871393239901072e-05,
+      "loss": 2.765,
       "step": 2018
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006875515251442704,
-      "loss": 4.9071,
+      "learning_rate": 6.869332234130256e-05,
+      "loss": 2.438,
       "step": 2019
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006873454245671888,
-      "loss": 5.0102,
+      "learning_rate": 6.867271228359439e-05,
+      "loss": 2.7921,
       "step": 2020
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006871393239901071,
-      "loss": 5.1051,
+      "learning_rate": 6.865210222588624e-05,
+      "loss": 2.7142,
       "step": 2021
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006869332234130255,
-      "loss": 4.6792,
+      "learning_rate": 6.863149216817807e-05,
+      "loss": 2.1624,
       "step": 2022
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006867271228359439,
-      "loss": 4.7521,
+      "learning_rate": 6.861088211046991e-05,
+      "loss": 2.7758,
       "step": 2023
     },
     {
       "epoch": 1.13,
-      "learning_rate": 0.0006865210222588623,
-      "loss": 5.682,
+      "learning_rate": 6.859027205276175e-05,
+      "loss": 2.9296,
       "step": 2024
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006863149216817806,
-      "loss": 5.6318,
+      "learning_rate": 6.85696619950536e-05,
+      "loss": 3.1204,
       "step": 2025
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.000686108821104699,
-      "loss": 5.7699,
+      "learning_rate": 6.854905193734543e-05,
+      "loss": 2.9648,
       "step": 2026
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006859027205276174,
-      "loss": 5.8706,
+      "learning_rate": 6.852844187963726e-05,
+      "loss": 2.9712,
       "step": 2027
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006856966199505359,
-      "loss": 4.7844,
+      "learning_rate": 6.85078318219291e-05,
+      "loss": 2.0791,
       "step": 2028
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006854905193734543,
-      "loss": 6.2101,
+      "learning_rate": 6.848722176422094e-05,
+      "loss": 2.6485,
       "step": 2029
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006852844187963726,
-      "loss": 5.4375,
+      "learning_rate": 6.846661170651279e-05,
+      "loss": 2.1881,
       "step": 2030
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.000685078318219291,
-      "loss": 6.7288,
+      "learning_rate": 6.844600164880463e-05,
+      "loss": 2.9141,
       "step": 2031
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006848722176422094,
-      "loss": 4.4745,
+      "learning_rate": 6.842539159109646e-05,
+      "loss": 1.6852,
       "step": 2032
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006846661170651278,
-      "loss": 5.2756,
+      "learning_rate": 6.840478153338829e-05,
+      "loss": 1.7769,
       "step": 2033
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006844600164880462,
-      "loss": 4.9741,
+      "learning_rate": 6.838417147568013e-05,
+      "loss": 1.7211,
       "step": 2034
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006842539159109645,
-      "loss": 5.3267,
+      "learning_rate": 6.836356141797198e-05,
+      "loss": 3.5578,
       "step": 2035
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006840478153338829,
-      "loss": 5.0628,
+      "learning_rate": 6.83429513602638e-05,
+      "loss": 3.3666,
       "step": 2036
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006838417147568013,
-      "loss": 5.4387,
+      "learning_rate": 6.832234130255565e-05,
+      "loss": 3.7375,
       "step": 2037
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006836356141797197,
-      "loss": 4.8829,
+      "learning_rate": 6.830173124484749e-05,
+      "loss": 2.8597,
       "step": 2038
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.000683429513602638,
-      "loss": 5.2361,
+      "learning_rate": 6.828112118713934e-05,
+      "loss": 3.2079,
       "step": 2039
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006832234130255564,
-      "loss": 5.438,
+      "learning_rate": 6.826051112943117e-05,
+      "loss": 3.409,
       "step": 2040
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006830173124484748,
-      "loss": 4.7598,
+      "learning_rate": 6.8239901071723e-05,
+      "loss": 2.5015,
       "step": 2041
     },
     {
       "epoch": 1.14,
-      "learning_rate": 0.0006828112118713932,
-      "loss": 5.3378,
+      "learning_rate": 6.821929101401484e-05,
+      "loss": 3.4328,
       "step": 2042
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006826051112943117,
-      "loss": 4.9224,
+      "learning_rate": 6.819868095630668e-05,
+      "loss": 2.7759,
       "step": 2043
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.00068239901071723,
-      "loss": 4.8368,
+      "learning_rate": 6.817807089859853e-05,
+      "loss": 2.6777,
       "step": 2044
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006821929101401484,
-      "loss": 5.5691,
+      "learning_rate": 6.815746084089036e-05,
+      "loss": 3.2365,
       "step": 2045
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006819868095630669,
-      "loss": 5.0673,
+      "learning_rate": 6.81368507831822e-05,
+      "loss": 2.6651,
       "step": 2046
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006817807089859853,
-      "loss": 4.8679,
+      "learning_rate": 6.811624072547403e-05,
+      "loss": 2.496,
       "step": 2047
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006815746084089036,
-      "loss": 5.7343,
+      "learning_rate": 6.809563066776587e-05,
+      "loss": 3.2131,
       "step": 2048
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.000681368507831822,
-      "loss": 5.1696,
+      "learning_rate": 6.807502061005772e-05,
+      "loss": 2.778,
       "step": 2049
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006811624072547404,
-      "loss": 5.115,
+      "learning_rate": 6.805441055234955e-05,
+      "loss": 2.9473,
       "step": 2050
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006809563066776588,
-      "loss": 4.8737,
+      "learning_rate": 6.803380049464139e-05,
+      "loss": 2.9041,
       "step": 2051
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006807502061005772,
-      "loss": 4.8922,
+      "learning_rate": 6.801319043693323e-05,
+      "loss": 2.929,
       "step": 2052
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006805441055234955,
-      "loss": 5.2827,
+      "learning_rate": 6.799258037922506e-05,
+      "loss": 2.6159,
       "step": 2053
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006803380049464139,
-      "loss": 4.832,
+      "learning_rate": 6.79719703215169e-05,
+      "loss": 2.5476,
       "step": 2054
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006801319043693323,
-      "loss": 5.1816,
+      "learning_rate": 6.795136026380874e-05,
+      "loss": 2.8988,
       "step": 2055
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006799258037922507,
-      "loss": 4.8719,
+      "learning_rate": 6.793075020610058e-05,
+      "loss": 2.5264,
       "step": 2056
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006797197032151691,
-      "loss": 4.848,
+      "learning_rate": 6.791014014839242e-05,
+      "loss": 2.5588,
       "step": 2057
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006795136026380874,
-      "loss": 4.8054,
+      "learning_rate": 6.788953009068427e-05,
+      "loss": 2.2639,
       "step": 2058
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006793075020610058,
-      "loss": 5.3394,
+      "learning_rate": 6.78689200329761e-05,
+      "loss": 3.2257,
       "step": 2059
     },
     {
       "epoch": 1.15,
-      "learning_rate": 0.0006791014014839242,
-      "loss": 4.8862,
+      "learning_rate": 6.784830997526794e-05,
+      "loss": 2.5711,
       "step": 2060
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006788953009068426,
-      "loss": 5.4533,
+      "learning_rate": 6.782769991755977e-05,
+      "loss": 2.9893,
       "step": 2061
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006786892003297609,
-      "loss": 5.2488,
+      "learning_rate": 6.780708985985161e-05,
+      "loss": 3.0382,
       "step": 2062
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006784830997526793,
-      "loss": 4.9514,
+      "learning_rate": 6.778647980214346e-05,
+      "loss": 2.6238,
       "step": 2063
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006782769991755977,
-      "loss": 4.6766,
+      "learning_rate": 6.776586974443528e-05,
+      "loss": 2.2181,
       "step": 2064
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006780708985985161,
-      "loss": 5.0576,
+      "learning_rate": 6.774525968672713e-05,
+      "loss": 2.4961,
       "step": 2065
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006778647980214346,
-      "loss": 5.5898,
+      "learning_rate": 6.772464962901897e-05,
+      "loss": 3.0997,
       "step": 2066
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006776586974443528,
-      "loss": 4.7287,
+      "learning_rate": 6.77040395713108e-05,
+      "loss": 2.3704,
       "step": 2067
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006774525968672713,
-      "loss": 5.0363,
+      "learning_rate": 6.768342951360263e-05,
+      "loss": 2.3248,
       "step": 2068
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006772464962901897,
-      "loss": 5.0711,
+      "learning_rate": 6.766281945589447e-05,
+      "loss": 2.7154,
       "step": 2069
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006770403957131081,
-      "loss": 4.9878,
+      "learning_rate": 6.764220939818632e-05,
+      "loss": 2.8005,
       "step": 2070
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006768342951360264,
-      "loss": 4.9903,
+      "learning_rate": 6.762159934047816e-05,
+      "loss": 2.8152,
       "step": 2071
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006766281945589448,
-      "loss": 5.771,
+      "learning_rate": 6.760098928277e-05,
+      "loss": 3.2808,
       "step": 2072
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006764220939818632,
-      "loss": 4.6962,
+      "learning_rate": 6.758037922506183e-05,
+      "loss": 2.5624,
       "step": 2073
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006762159934047816,
-      "loss": 4.6339,
+      "learning_rate": 6.755976916735366e-05,
+      "loss": 2.3283,
       "step": 2074
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006760098928277,
-      "loss": 4.9984,
+      "learning_rate": 6.753915910964551e-05,
+      "loss": 2.4426,
       "step": 2075
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006758037922506183,
-      "loss": 5.1957,
+      "learning_rate": 6.751854905193735e-05,
+      "loss": 3.2038,
       "step": 2076
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006755976916735367,
-      "loss": 5.4708,
+      "learning_rate": 6.749793899422918e-05,
+      "loss": 2.4949,
       "step": 2077
     },
     {
       "epoch": 1.16,
-      "learning_rate": 0.0006753915910964551,
-      "loss": 5.0499,
+      "learning_rate": 6.747732893652102e-05,
+      "loss": 2.1817,
       "step": 2078
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006751854905193735,
-      "loss": 4.6799,
+      "learning_rate": 6.745671887881287e-05,
+      "loss": 2.3183,
       "step": 2079
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006749793899422918,
-      "loss": 6.6565,
+      "learning_rate": 6.74361088211047e-05,
+      "loss": 2.9743,
       "step": 2080
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006747732893652102,
-      "loss": 5.3778,
+      "learning_rate": 6.741549876339654e-05,
+      "loss": 2.7138,
       "step": 2081
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006745671887881286,
-      "loss": 4.9116,
+      "learning_rate": 6.739488870568837e-05,
+      "loss": 1.7181,
       "step": 2082
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.000674361088211047,
-      "loss": 5.4526,
+      "learning_rate": 6.737427864798021e-05,
+      "loss": 2.4202,
       "step": 2083
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006741549876339654,
-      "loss": 7.7358,
+      "learning_rate": 6.735366859027206e-05,
+      "loss": 2.6782,
       "step": 2084
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006739488870568837,
-      "loss": 5.1698,
+      "learning_rate": 6.73330585325639e-05,
+      "loss": 3.8997,
       "step": 2085
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006737427864798021,
-      "loss": 4.6365,
+      "learning_rate": 6.731244847485574e-05,
+      "loss": 2.627,
       "step": 2086
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006735366859027205,
-      "loss": 5.1635,
+      "learning_rate": 6.729183841714757e-05,
+      "loss": 2.9701,
       "step": 2087
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.000673330585325639,
-      "loss": 5.0961,
+      "learning_rate": 6.72712283594394e-05,
+      "loss": 2.7573,
       "step": 2088
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006731244847485574,
-      "loss": 4.9722,
+      "learning_rate": 6.725061830173125e-05,
+      "loss": 2.744,
       "step": 2089
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006729183841714757,
-      "loss": 5.4982,
+      "learning_rate": 6.723000824402309e-05,
+      "loss": 2.9582,
       "step": 2090
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006727122835943941,
-      "loss": 5.1925,
+      "learning_rate": 6.720939818631492e-05,
+      "loss": 3.0215,
       "step": 2091
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006725061830173125,
-      "loss": 5.236,
+      "learning_rate": 6.718878812860676e-05,
+      "loss": 2.8779,
       "step": 2092
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006723000824402309,
-      "loss": 5.0232,
+      "learning_rate": 6.716817807089861e-05,
+      "loss": 2.849,
       "step": 2093
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006720939818631492,
-      "loss": 5.0245,
+      "learning_rate": 6.714756801319044e-05,
+      "loss": 2.6383,
       "step": 2094
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.0006718878812860676,
-      "loss": 5.1335,
+      "learning_rate": 6.712695795548228e-05,
+      "loss": 3.0522,
       "step": 2095
     },
     {
       "epoch": 1.17,
-      "learning_rate": 0.000671681780708986,
-      "loss": 4.9398,
+      "learning_rate": 6.710634789777411e-05,
+      "loss": 3.3823,
       "step": 2096
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006714756801319044,
-      "loss": 5.2128,
+      "learning_rate": 6.708573784006595e-05,
+      "loss": 3.1955,
       "step": 2097
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006712695795548228,
-      "loss": 5.4953,
+      "learning_rate": 6.70651277823578e-05,
+      "loss": 3.1535,
       "step": 2098
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006710634789777411,
-      "loss": 5.675,
+      "learning_rate": 6.704451772464964e-05,
+      "loss": 3.5491,
       "step": 2099
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006708573784006595,
-      "loss": 4.7282,
+      "learning_rate": 6.702390766694147e-05,
+      "loss": 2.8054,
       "step": 2100
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006706512778235779,
-      "loss": 5.0345,
+      "learning_rate": 6.70032976092333e-05,
+      "loss": 2.8118,
       "step": 2101
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006704451772464963,
-      "loss": 4.8078,
+      "learning_rate": 6.698268755152514e-05,
+      "loss": 2.7673,
       "step": 2102
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006702390766694146,
-      "loss": 4.7974,
+      "learning_rate": 6.696207749381699e-05,
+      "loss": 2.286,
       "step": 2103
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.000670032976092333,
-      "loss": 5.9797,
+      "learning_rate": 6.694146743610883e-05,
+      "loss": 3.4553,
       "step": 2104
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006698268755152514,
-      "loss": 5.2685,
+      "learning_rate": 6.692085737840066e-05,
+      "loss": 3.3762,
       "step": 2105
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006696207749381698,
-      "loss": 5.1761,
+      "learning_rate": 6.69002473206925e-05,
+      "loss": 2.9303,
       "step": 2106
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006694146743610882,
-      "loss": 5.2984,
+      "learning_rate": 6.687963726298435e-05,
+      "loss": 2.9424,
       "step": 2107
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006692085737840065,
-      "loss": 5.0798,
+      "learning_rate": 6.685902720527618e-05,
+      "loss": 2.9382,
       "step": 2108
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.000669002473206925,
-      "loss": 5.2807,
+      "learning_rate": 6.6838417147568e-05,
+      "loss": 3.0751,
       "step": 2109
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006687963726298434,
-      "loss": 4.8416,
+      "learning_rate": 6.681780708985985e-05,
+      "loss": 2.4973,
       "step": 2110
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006685902720527618,
-      "loss": 5.1076,
+      "learning_rate": 6.67971970321517e-05,
+      "loss": 2.665,
       "step": 2111
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006683841714756801,
-      "loss": 5.5306,
+      "learning_rate": 6.677658697444354e-05,
+      "loss": 3.4869,
       "step": 2112
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006681780708985985,
-      "loss": 5.152,
+      "learning_rate": 6.675597691673538e-05,
+      "loss": 3.2595,
       "step": 2113
     },
     {
       "epoch": 1.18,
-      "learning_rate": 0.0006679719703215169,
-      "loss": 4.957,
+      "learning_rate": 6.673536685902721e-05,
+      "loss": 2.6042,
       "step": 2114
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006677658697444353,
-      "loss": 5.2095,
+      "learning_rate": 6.671475680131904e-05,
+      "loss": 2.8262,
       "step": 2115
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006675597691673537,
-      "loss": 4.7728,
+      "learning_rate": 6.669414674361088e-05,
+      "loss": 2.7536,
       "step": 2116
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.000667353668590272,
-      "loss": 4.6956,
+      "learning_rate": 6.667353668590273e-05,
+      "loss": 2.6872,
       "step": 2117
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006671475680131904,
-      "loss": 5.2243,
+      "learning_rate": 6.665292662819457e-05,
+      "loss": 2.9678,
       "step": 2118
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006669414674361088,
-      "loss": 4.7681,
+      "learning_rate": 6.66323165704864e-05,
+      "loss": 2.2239,
       "step": 2119
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006667353668590272,
-      "loss": 5.1002,
+      "learning_rate": 6.661170651277824e-05,
+      "loss": 2.7768,
       "step": 2120
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006665292662819456,
-      "loss": 5.2388,
+      "learning_rate": 6.659109645507007e-05,
+      "loss": 2.9819,
       "step": 2121
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006663231657048639,
-      "loss": 4.9552,
+      "learning_rate": 6.657048639736192e-05,
+      "loss": 2.716,
       "step": 2122
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006661170651277823,
-      "loss": 4.9023,
+      "learning_rate": 6.654987633965375e-05,
+      "loss": 2.3425,
       "step": 2123
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006659109645507007,
-      "loss": 4.7282,
+      "learning_rate": 6.652926628194559e-05,
+      "loss": 2.1124,
       "step": 2124
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006657048639736191,
-      "loss": 5.5284,
+      "learning_rate": 6.650865622423743e-05,
+      "loss": 2.9368,
       "step": 2125
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006654987633965374,
-      "loss": 5.4734,
+      "learning_rate": 6.648804616652928e-05,
+      "loss": 2.6032,
       "step": 2126
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006652926628194558,
-      "loss": 4.7063,
+      "learning_rate": 6.64674361088211e-05,
+      "loss": 1.939,
       "step": 2127
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006650865622423742,
-      "loss": 5.0588,
+      "learning_rate": 6.644682605111294e-05,
+      "loss": 2.2763,
       "step": 2128
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006648804616652927,
-      "loss": 4.8522,
+      "learning_rate": 6.642621599340478e-05,
+      "loss": 2.048,
       "step": 2129
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006646743610882111,
-      "loss": 6.3917,
+      "learning_rate": 6.640560593569662e-05,
+      "loss": 3.1694,
       "step": 2130
     },
     {
       "epoch": 1.19,
-      "learning_rate": 0.0006644682605111294,
-      "loss": 4.5742,
+      "learning_rate": 6.638499587798847e-05,
+      "loss": 1.595,
       "step": 2131
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006642621599340478,
-      "loss": 5.0439,
+      "learning_rate": 6.63643858202803e-05,
+      "loss": 2.3933,
       "step": 2132
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006640560593569662,
-      "loss": 5.0811,
+      "learning_rate": 6.634377576257214e-05,
+      "loss": 1.7518,
       "step": 2133
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006638499587798846,
-      "loss": 4.8239,
+      "learning_rate": 6.632316570486398e-05,
+      "loss": 2.0221,
       "step": 2134
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006636438582028029,
-      "loss": 5.4555,
+      "learning_rate": 6.630255564715581e-05,
+      "loss": 3.6729,
       "step": 2135
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006634377576257213,
-      "loss": 5.2008,
+      "learning_rate": 6.628194558944766e-05,
+      "loss": 3.173,
       "step": 2136
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006632316570486398,
-      "loss": 4.9969,
+      "learning_rate": 6.626133553173949e-05,
+      "loss": 2.8937,
       "step": 2137
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006630255564715582,
-      "loss": 4.7945,
+      "learning_rate": 6.624072547403133e-05,
+      "loss": 2.6084,
       "step": 2138
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006628194558944766,
-      "loss": 4.7548,
+      "learning_rate": 6.622011541632317e-05,
+      "loss": 2.6746,
       "step": 2139
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006626133553173949,
-      "loss": 5.4747,
+      "learning_rate": 6.619950535861502e-05,
+      "loss": 3.2476,
       "step": 2140
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006624072547403133,
-      "loss": 5.8072,
+      "learning_rate": 6.617889530090685e-05,
+      "loss": 3.519,
       "step": 2141
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006622011541632317,
-      "loss": 4.7993,
+      "learning_rate": 6.615828524319868e-05,
+      "loss": 2.9599,
       "step": 2142
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006619950535861501,
-      "loss": 5.079,
+      "learning_rate": 6.613767518549052e-05,
+      "loss": 2.9583,
       "step": 2143
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006617889530090684,
-      "loss": 5.0109,
+      "learning_rate": 6.611706512778236e-05,
+      "loss": 2.9064,
       "step": 2144
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006615828524319868,
-      "loss": 5.3951,
+      "learning_rate": 6.60964550700742e-05,
+      "loss": 2.9808,
       "step": 2145
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006613767518549052,
-      "loss": 5.3109,
+      "learning_rate": 6.607584501236604e-05,
+      "loss": 3.407,
       "step": 2146
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006611706512778237,
-      "loss": 4.9196,
+      "learning_rate": 6.605523495465788e-05,
+      "loss": 2.7415,
       "step": 2147
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006609645507007421,
-      "loss": 4.9524,
+      "learning_rate": 6.603462489694971e-05,
+      "loss": 2.8192,
       "step": 2148
     },
     {
       "epoch": 1.2,
-      "learning_rate": 0.0006607584501236604,
-      "loss": 4.9983,
+      "learning_rate": 6.601401483924155e-05,
+      "loss": 2.4052,
       "step": 2149
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006605523495465788,
-      "loss": 5.2353,
+      "learning_rate": 6.59934047815334e-05,
+      "loss": 2.9803,
       "step": 2150
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006603462489694972,
-      "loss": 5.2597,
+      "learning_rate": 6.597279472382523e-05,
+      "loss": 3.3818,
       "step": 2151
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006601401483924156,
-      "loss": 5.698,
+      "learning_rate": 6.595218466611707e-05,
+      "loss": 3.5399,
       "step": 2152
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.000659934047815334,
-      "loss": 5.1041,
+      "learning_rate": 6.593157460840891e-05,
+      "loss": 2.8021,
       "step": 2153
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006597279472382523,
-      "loss": 5.1161,
+      "learning_rate": 6.591096455070076e-05,
+      "loss": 3.4408,
       "step": 2154
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006595218466611707,
-      "loss": 5.1219,
+      "learning_rate": 6.589035449299259e-05,
+      "loss": 3.3177,
       "step": 2155
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006593157460840891,
-      "loss": 4.8318,
+      "learning_rate": 6.586974443528442e-05,
+      "loss": 2.7792,
       "step": 2156
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006591096455070075,
-      "loss": 5.456,
+      "learning_rate": 6.584913437757626e-05,
+      "loss": 2.8305,
       "step": 2157
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006589035449299258,
-      "loss": 5.0469,
+      "learning_rate": 6.58285243198681e-05,
+      "loss": 2.634,
       "step": 2158
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006586974443528442,
-      "loss": 5.1517,
+      "learning_rate": 6.580791426215995e-05,
+      "loss": 2.8523,
       "step": 2159
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006584913437757626,
-      "loss": 5.2521,
+      "learning_rate": 6.578730420445178e-05,
+      "loss": 2.792,
       "step": 2160
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.000658285243198681,
-      "loss": 4.861,
+      "learning_rate": 6.576669414674362e-05,
+      "loss": 2.6333,
       "step": 2161
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006580791426215994,
-      "loss": 4.9641,
+      "learning_rate": 6.574608408903545e-05,
+      "loss": 2.4692,
       "step": 2162
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006578730420445177,
-      "loss": 5.1267,
+      "learning_rate": 6.572547403132729e-05,
+      "loss": 3.2218,
       "step": 2163
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006576669414674361,
-      "loss": 4.8712,
+      "learning_rate": 6.570486397361912e-05,
+      "loss": 2.4505,
       "step": 2164
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006574608408903545,
-      "loss": 5.0347,
+      "learning_rate": 6.568425391591096e-05,
+      "loss": 2.4814,
       "step": 2165
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006572547403132729,
-      "loss": 5.0449,
+      "learning_rate": 6.566364385820281e-05,
+      "loss": 2.5408,
       "step": 2166
     },
     {
       "epoch": 1.21,
-      "learning_rate": 0.0006570486397361912,
-      "loss": 5.3153,
+      "learning_rate": 6.564303380049465e-05,
+      "loss": 3.0781,
       "step": 2167
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006568425391591096,
-      "loss": 4.8321,
+      "learning_rate": 6.562242374278648e-05,
+      "loss": 2.3667,
       "step": 2168
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006566364385820281,
-      "loss": 5.093,
+      "learning_rate": 6.560181368507831e-05,
+      "loss": 2.6961,
       "step": 2169
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006564303380049465,
-      "loss": 5.1645,
+      "learning_rate": 6.558120362737015e-05,
+      "loss": 2.3929,
       "step": 2170
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006562242374278649,
-      "loss": 4.7657,
+      "learning_rate": 6.5560593569662e-05,
+      "loss": 2.5079,
       "step": 2171
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006560181368507832,
-      "loss": 5.6444,
+      "learning_rate": 6.553998351195384e-05,
+      "loss": 3.0234,
       "step": 2172
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006558120362737016,
-      "loss": 4.9064,
+      "learning_rate": 6.551937345424568e-05,
+      "loss": 2.607,
       "step": 2173
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.00065560593569662,
-      "loss": 4.8151,
+      "learning_rate": 6.549876339653751e-05,
+      "loss": 2.36,
       "step": 2174
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006553998351195384,
-      "loss": 4.9157,
+      "learning_rate": 6.547815333882934e-05,
+      "loss": 2.1065,
       "step": 2175
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006551937345424568,
-      "loss": 4.7286,
+      "learning_rate": 6.545754328112119e-05,
+      "loss": 2.3123,
       "step": 2176
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006549876339653751,
-      "loss": 4.511,
+      "learning_rate": 6.543693322341303e-05,
+      "loss": 1.6656,
       "step": 2177
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006547815333882935,
-      "loss": 5.6233,
+      "learning_rate": 6.541632316570486e-05,
+      "loss": 2.7355,
       "step": 2178
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006545754328112119,
-      "loss": 4.5108,
+      "learning_rate": 6.53957131079967e-05,
+      "loss": 1.7348,
       "step": 2179
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006543693322341303,
-      "loss": 5.164,
+      "learning_rate": 6.537510305028855e-05,
+      "loss": 1.9474,
       "step": 2180
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006541632316570486,
-      "loss": 4.7171,
+      "learning_rate": 6.535449299258039e-05,
+      "loss": 1.398,
       "step": 2181
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.000653957131079967,
-      "loss": 4.5693,
+      "learning_rate": 6.533388293487222e-05,
+      "loss": 1.9651,
       "step": 2182
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006537510305028854,
-      "loss": 6.1144,
+      "learning_rate": 6.531327287716405e-05,
+      "loss": 1.9611,
       "step": 2183
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006535449299258038,
-      "loss": 4.3996,
+      "learning_rate": 6.52926628194559e-05,
+      "loss": 1.4461,
       "step": 2184
     },
     {
       "epoch": 1.22,
-      "learning_rate": 0.0006533388293487222,
-      "loss": 5.2977,
+      "learning_rate": 6.527205276174774e-05,
+      "loss": 3.4731,
       "step": 2185
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006531327287716405,
-      "loss": 5.2022,
+      "learning_rate": 6.525144270403958e-05,
+      "loss": 3.3538,
       "step": 2186
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006529266281945589,
-      "loss": 4.8768,
+      "learning_rate": 6.523083264633141e-05,
+      "loss": 2.6889,
       "step": 2187
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006527205276174773,
-      "loss": 5.2843,
+      "learning_rate": 6.521022258862325e-05,
+      "loss": 3.4026,
       "step": 2188
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006525144270403958,
-      "loss": 5.3472,
+      "learning_rate": 6.518961253091508e-05,
+      "loss": 2.8535,
       "step": 2189
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.000652308326463314,
-      "loss": 5.0712,
+      "learning_rate": 6.516900247320693e-05,
+      "loss": 3.2634,
       "step": 2190
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006521022258862325,
-      "loss": 5.0313,
+      "learning_rate": 6.514839241549877e-05,
+      "loss": 2.8816,
       "step": 2191
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006518961253091509,
-      "loss": 5.3047,
+      "learning_rate": 6.51277823577906e-05,
+      "loss": 3.4527,
       "step": 2192
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006516900247320693,
-      "loss": 5.5777,
+      "learning_rate": 6.510717230008244e-05,
+      "loss": 3.3671,
       "step": 2193
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006514839241549877,
-      "loss": 5.1147,
+      "learning_rate": 6.508656224237429e-05,
+      "loss": 3.0464,
       "step": 2194
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.000651277823577906,
-      "loss": 5.0325,
+      "learning_rate": 6.506595218466612e-05,
+      "loss": 2.9931,
       "step": 2195
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006510717230008244,
-      "loss": 5.1492,
+      "learning_rate": 6.504534212695795e-05,
+      "loss": 3.5519,
       "step": 2196
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006508656224237428,
-      "loss": 5.1345,
+      "learning_rate": 6.502473206924979e-05,
+      "loss": 3.1822,
       "step": 2197
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006506595218466612,
-      "loss": 5.3183,
+      "learning_rate": 6.500412201154163e-05,
+      "loss": 3.2072,
       "step": 2198
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006504534212695795,
-      "loss": 4.9002,
+      "learning_rate": 6.498351195383348e-05,
+      "loss": 2.845,
       "step": 2199
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006502473206924979,
-      "loss": 5.5162,
+      "learning_rate": 6.496290189612532e-05,
+      "loss": 3.0673,
       "step": 2200
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006500412201154163,
-      "loss": 5.4818,
+      "learning_rate": 6.494229183841715e-05,
+      "loss": 3.2635,
       "step": 2201
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006498351195383347,
-      "loss": 5.1126,
+      "learning_rate": 6.4921681780709e-05,
+      "loss": 3.4378,
       "step": 2202
     },
     {
       "epoch": 1.23,
-      "learning_rate": 0.0006496290189612531,
-      "loss": 4.833,
+      "learning_rate": 6.490107172300082e-05,
+      "loss": 2.576,
       "step": 2203
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006494229183841714,
-      "loss": 5.3174,
+      "learning_rate": 6.488046166529267e-05,
+      "loss": 3.2772,
       "step": 2204
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006492168178070898,
-      "loss": 5.2388,
+      "learning_rate": 6.485985160758451e-05,
+      "loss": 2.8987,
       "step": 2205
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006490107172300082,
-      "loss": 4.9963,
+      "learning_rate": 6.483924154987634e-05,
+      "loss": 2.6454,
       "step": 2206
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006488046166529266,
-      "loss": 5.2855,
+      "learning_rate": 6.481863149216818e-05,
+      "loss": 3.3792,
       "step": 2207
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.000648598516075845,
-      "loss": 5.0522,
+      "learning_rate": 6.479802143446003e-05,
+      "loss": 2.992,
       "step": 2208
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006483924154987633,
-      "loss": 4.8046,
+      "learning_rate": 6.477741137675186e-05,
+      "loss": 2.647,
       "step": 2209
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006481863149216818,
-      "loss": 5.0021,
+      "learning_rate": 6.475680131904369e-05,
+      "loss": 2.5442,
       "step": 2210
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006479802143446002,
-      "loss": 4.8862,
+      "learning_rate": 6.473619126133553e-05,
+      "loss": 2.5099,
       "step": 2211
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006477741137675186,
-      "loss": 4.723,
+      "learning_rate": 6.471558120362737e-05,
+      "loss": 2.5015,
       "step": 2212
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006475680131904369,
-      "loss": 6.2146,
+      "learning_rate": 6.469497114591922e-05,
+      "loss": 3.298,
       "step": 2213
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006473619126133553,
-      "loss": 5.2786,
+      "learning_rate": 6.467436108821106e-05,
+      "loss": 2.6763,
       "step": 2214
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006471558120362737,
-      "loss": 4.8828,
+      "learning_rate": 6.465375103050289e-05,
+      "loss": 2.4564,
       "step": 2215
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006469497114591921,
-      "loss": 5.0016,
+      "learning_rate": 6.463314097279472e-05,
+      "loss": 2.8676,
       "step": 2216
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006467436108821105,
-      "loss": 5.7356,
+      "learning_rate": 6.461253091508656e-05,
+      "loss": 3.0212,
       "step": 2217
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006465375103050288,
-      "loss": 5.269,
+      "learning_rate": 6.45919208573784e-05,
+      "loss": 3.0672,
       "step": 2218
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006463314097279472,
-      "loss": 5.1999,
+      "learning_rate": 6.457131079967024e-05,
+      "loss": 2.9126,
       "step": 2219
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.0006461253091508656,
-      "loss": 5.0834,
+      "learning_rate": 6.455070074196208e-05,
+      "loss": 2.6322,
       "step": 2220
     },
     {
       "epoch": 1.24,
-      "learning_rate": 0.000645919208573784,
-      "loss": 5.1478,
+      "learning_rate": 6.453009068425392e-05,
+      "loss": 2.8674,
       "step": 2221
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006457131079967023,
-      "loss": 4.9197,
+      "learning_rate": 6.450948062654575e-05,
+      "loss": 2.2294,
       "step": 2222
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006455070074196207,
-      "loss": 4.6722,
+      "learning_rate": 6.44888705688376e-05,
+      "loss": 2.0877,
       "step": 2223
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006453009068425391,
-      "loss": 7.0552,
+      "learning_rate": 6.446826051112943e-05,
+      "loss": 3.0758,
       "step": 2224
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006450948062654575,
-      "loss": 5.1267,
+      "learning_rate": 6.444765045342127e-05,
+      "loss": 2.6462,
       "step": 2225
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006448887056883759,
-      "loss": 5.0003,
+      "learning_rate": 6.442704039571311e-05,
+      "loss": 2.2029,
       "step": 2226
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006446826051112942,
-      "loss": 4.9738,
+      "learning_rate": 6.440643033800496e-05,
+      "loss": 2.1649,
       "step": 2227
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006444765045342126,
-      "loss": 5.3063,
+      "learning_rate": 6.438582028029679e-05,
+      "loss": 2.6128,
       "step": 2228
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.000644270403957131,
-      "loss": 5.1288,
+      "learning_rate": 6.436521022258863e-05,
+      "loss": 1.8058,
       "step": 2229
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006440643033800496,
-      "loss": 5.8976,
+      "learning_rate": 6.434460016488046e-05,
+      "loss": 2.2301,
       "step": 2230
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006438582028029679,
-      "loss": 4.8955,
+      "learning_rate": 6.43239901071723e-05,
+      "loss": 2.2326,
       "step": 2231
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006436521022258863,
-      "loss": 4.9153,
+      "learning_rate": 6.430338004946415e-05,
+      "loss": 1.556,
       "step": 2232
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006434460016488047,
-      "loss": 5.3267,
+      "learning_rate": 6.428276999175598e-05,
+      "loss": 2.5114,
       "step": 2233
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006432399010717231,
-      "loss": 4.9063,
+      "learning_rate": 6.426215993404782e-05,
+      "loss": 1.8808,
       "step": 2234
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006430338004946415,
-      "loss": 5.3899,
+      "learning_rate": 6.424154987633966e-05,
+      "loss": 3.5655,
       "step": 2235
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006428276999175598,
-      "loss": 4.8636,
+      "learning_rate": 6.422093981863149e-05,
+      "loss": 3.0745,
       "step": 2236
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006426215993404782,
-      "loss": 4.9403,
+      "learning_rate": 6.420032976092334e-05,
+      "loss": 3.0403,
       "step": 2237
     },
     {
       "epoch": 1.25,
-      "learning_rate": 0.0006424154987633966,
-      "loss": 5.3336,
+      "learning_rate": 6.417971970321517e-05,
+      "loss": 3.3698,
       "step": 2238
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.000642209398186315,
-      "loss": 5.1489,
+      "learning_rate": 6.415910964550701e-05,
+      "loss": 3.2757,
       "step": 2239
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006420032976092334,
-      "loss": 6.1086,
+      "learning_rate": 6.413849958779885e-05,
+      "loss": 3.5646,
       "step": 2240
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006417971970321517,
-      "loss": 5.4432,
+      "learning_rate": 6.41178895300907e-05,
+      "loss": 3.5319,
       "step": 2241
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006415910964550701,
-      "loss": 6.203,
+      "learning_rate": 6.409727947238253e-05,
+      "loss": 3.5392,
       "step": 2242
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006413849958779885,
-      "loss": 4.9286,
+      "learning_rate": 6.407666941467436e-05,
+      "loss": 2.9863,
       "step": 2243
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006411788953009069,
-      "loss": 4.9594,
+      "learning_rate": 6.40560593569662e-05,
+      "loss": 3.1839,
       "step": 2244
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006409727947238252,
-      "loss": 5.2662,
+      "learning_rate": 6.403544929925804e-05,
+      "loss": 3.3591,
       "step": 2245
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006407666941467436,
-      "loss": 4.9979,
+      "learning_rate": 6.401483924154989e-05,
+      "loss": 3.0742,
       "step": 2246
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.000640560593569662,
-      "loss": 4.9016,
+      "learning_rate": 6.399422918384172e-05,
+      "loss": 2.9052,
       "step": 2247
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006403544929925804,
-      "loss": 4.9496,
+      "learning_rate": 6.397361912613356e-05,
+      "loss": 2.7001,
       "step": 2248
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006401483924154989,
-      "loss": 4.8467,
+      "learning_rate": 6.39530090684254e-05,
+      "loss": 2.9376,
       "step": 2249
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006399422918384172,
-      "loss": 4.7107,
+      "learning_rate": 6.393239901071723e-05,
+      "loss": 2.5735,
       "step": 2250
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006397361912613356,
-      "loss": 5.3177,
+      "learning_rate": 6.391178895300906e-05,
+      "loss": 3.2369,
       "step": 2251
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.000639530090684254,
-      "loss": 5.1966,
+      "learning_rate": 6.38911788953009e-05,
+      "loss": 2.8486,
       "step": 2252
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006393239901071724,
-      "loss": 5.0696,
+      "learning_rate": 6.387056883759275e-05,
+      "loss": 2.65,
       "step": 2253
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006391178895300907,
-      "loss": 5.2792,
+      "learning_rate": 6.384995877988459e-05,
+      "loss": 2.9555,
       "step": 2254
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006389117889530091,
-      "loss": 5.3309,
+      "learning_rate": 6.382934872217644e-05,
+      "loss": 2.9868,
       "step": 2255
     },
     {
       "epoch": 1.26,
-      "learning_rate": 0.0006387056883759275,
-      "loss": 5.1494,
+      "learning_rate": 6.380873866446827e-05,
+      "loss": 2.8657,
       "step": 2256
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006384995877988459,
-      "loss": 4.8451,
+      "learning_rate": 6.37881286067601e-05,
+      "loss": 2.9111,
       "step": 2257
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006382934872217643,
-      "loss": 4.8265,
+      "learning_rate": 6.376751854905194e-05,
+      "loss": 2.8897,
       "step": 2258
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006380873866446826,
-      "loss": 5.15,
+      "learning_rate": 6.374690849134378e-05,
+      "loss": 2.9883,
       "step": 2259
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.000637881286067601,
-      "loss": 5.5345,
+      "learning_rate": 6.372629843363561e-05,
+      "loss": 3.1661,
       "step": 2260
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006376751854905194,
-      "loss": 5.0338,
+      "learning_rate": 6.370568837592745e-05,
+      "loss": 2.8187,
       "step": 2261
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006374690849134378,
-      "loss": 4.7857,
+      "learning_rate": 6.36850783182193e-05,
+      "loss": 2.9576,
       "step": 2262
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006372629843363561,
-      "loss": 5.3663,
+      "learning_rate": 6.366446826051113e-05,
+      "loss": 3.1198,
       "step": 2263
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006370568837592745,
-      "loss": 5.3876,
+      "learning_rate": 6.364385820280297e-05,
+      "loss": 3.3025,
       "step": 2264
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006368507831821929,
-      "loss": 5.333,
+      "learning_rate": 6.36232481450948e-05,
+      "loss": 3.3242,
       "step": 2265
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006366446826051113,
-      "loss": 5.4174,
+      "learning_rate": 6.360263808738664e-05,
+      "loss": 3.0775,
       "step": 2266
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006364385820280297,
-      "loss": 6.1951,
+      "learning_rate": 6.358202802967849e-05,
+      "loss": 3.1937,
       "step": 2267
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.000636232481450948,
-      "loss": 4.9606,
+      "learning_rate": 6.356141797197033e-05,
+      "loss": 2.7131,
       "step": 2268
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006360263808738664,
-      "loss": 5.1086,
+      "learning_rate": 6.354080791426217e-05,
+      "loss": 2.6765,
       "step": 2269
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006358202802967849,
-      "loss": 4.8038,
+      "learning_rate": 6.352019785655399e-05,
+      "loss": 2.6875,
       "step": 2270
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006356141797197033,
-      "loss": 5.4244,
+      "learning_rate": 6.349958779884583e-05,
+      "loss": 2.9657,
       "step": 2271
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006354080791426217,
-      "loss": 4.8888,
+      "learning_rate": 6.347897774113768e-05,
+      "loss": 2.4627,
       "step": 2272
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.00063520197856554,
-      "loss": 4.9327,
+      "learning_rate": 6.345836768342952e-05,
+      "loss": 2.3331,
       "step": 2273
     },
     {
       "epoch": 1.27,
-      "learning_rate": 0.0006349958779884584,
-      "loss": 4.4509,
+      "learning_rate": 6.343775762572135e-05,
+      "loss": 1.861,
       "step": 2274
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006347897774113768,
-      "loss": 5.3272,
+      "learning_rate": 6.34171475680132e-05,
+      "loss": 2.7239,
       "step": 2275
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006345836768342952,
-      "loss": 4.8705,
+      "learning_rate": 6.339653751030504e-05,
+      "loss": 2.3654,
       "step": 2276
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006343775762572135,
-      "loss": 4.7159,
+      "learning_rate": 6.337592745259687e-05,
+      "loss": 2.729,
       "step": 2277
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006341714756801319,
-      "loss": 4.6203,
+      "learning_rate": 6.335531739488871e-05,
+      "loss": 2.68,
       "step": 2278
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006339653751030503,
-      "loss": 4.8268,
+      "learning_rate": 6.333470733718054e-05,
+      "loss": 1.8634,
       "step": 2279
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006337592745259687,
-      "loss": 5.08,
+      "learning_rate": 6.331409727947238e-05,
+      "loss": 2.309,
       "step": 2280
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006335531739488871,
-      "loss": 4.9203,
+      "learning_rate": 6.329348722176423e-05,
+      "loss": 1.6964,
       "step": 2281
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006333470733718054,
-      "loss": 4.8121,
+      "learning_rate": 6.327287716405607e-05,
+      "loss": 1.6795,
       "step": 2282
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006331409727947238,
-      "loss": 5.0517,
+      "learning_rate": 6.32522671063479e-05,
+      "loss": 1.4838,
       "step": 2283
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006329348722176422,
-      "loss": 5.0679,
+      "learning_rate": 6.323165704863973e-05,
+      "loss": 2.2671,
       "step": 2284
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006327287716405606,
-      "loss": 5.8816,
+      "learning_rate": 6.321104699093157e-05,
+      "loss": 3.2518,
       "step": 2285
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006325226710634789,
-      "loss": 5.5661,
+      "learning_rate": 6.319043693322342e-05,
+      "loss": 3.4412,
       "step": 2286
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006323165704863973,
-      "loss": 5.0868,
+      "learning_rate": 6.316982687551526e-05,
+      "loss": 2.97,
       "step": 2287
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006321104699093157,
-      "loss": 5.1378,
+      "learning_rate": 6.314921681780709e-05,
+      "loss": 2.8984,
       "step": 2288
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006319043693322341,
-      "loss": 5.0546,
+      "learning_rate": 6.312860676009893e-05,
+      "loss": 3.003,
       "step": 2289
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006316982687551526,
-      "loss": 4.9491,
+      "learning_rate": 6.310799670239076e-05,
+      "loss": 2.595,
       "step": 2290
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006314921681780709,
-      "loss": 5.3176,
+      "learning_rate": 6.308738664468261e-05,
+      "loss": 3.2433,
       "step": 2291
     },
     {
       "epoch": 1.28,
-      "learning_rate": 0.0006312860676009893,
-      "loss": 5.2094,
+      "learning_rate": 6.306677658697445e-05,
+      "loss": 2.9818,
       "step": 2292
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006310799670239077,
-      "loss": 4.9723,
+      "learning_rate": 6.304616652926628e-05,
+      "loss": 2.7172,
       "step": 2293
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006308738664468261,
-      "loss": 4.9423,
+      "learning_rate": 6.302555647155812e-05,
+      "loss": 3.0789,
       "step": 2294
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006306677658697445,
-      "loss": 5.0661,
+      "learning_rate": 6.300494641384997e-05,
+      "loss": 2.6657,
       "step": 2295
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006304616652926628,
-      "loss": 4.937,
+      "learning_rate": 6.298433635614181e-05,
+      "loss": 2.9746,
       "step": 2296
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006302555647155812,
-      "loss": 4.8061,
+      "learning_rate": 6.296372629843364e-05,
+      "loss": 2.5641,
       "step": 2297
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006300494641384996,
-      "loss": 4.9952,
+      "learning_rate": 6.294311624072547e-05,
+      "loss": 2.6686,
       "step": 2298
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.000629843363561418,
-      "loss": 5.7981,
+      "learning_rate": 6.292250618301731e-05,
+      "loss": 3.8755,
       "step": 2299
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006296372629843363,
-      "loss": 4.8712,
+      "learning_rate": 6.290189612530916e-05,
+      "loss": 2.5296,
       "step": 2300
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006294311624072547,
-      "loss": 4.8221,
+      "learning_rate": 6.2881286067601e-05,
+      "loss": 2.2911,
       "step": 2301
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006292250618301731,
-      "loss": 4.6996,
+      "learning_rate": 6.286067600989283e-05,
+      "loss": 2.4122,
       "step": 2302
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006290189612530915,
-      "loss": 5.3258,
+      "learning_rate": 6.284006595218467e-05,
+      "loss": 3.2026,
       "step": 2303
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006288128606760099,
-      "loss": 4.818,
+      "learning_rate": 6.28194558944765e-05,
+      "loss": 2.4396,
       "step": 2304
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006286067600989282,
-      "loss": 4.9076,
+      "learning_rate": 6.279884583676835e-05,
+      "loss": 2.981,
       "step": 2305
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006284006595218466,
-      "loss": 5.7065,
+      "learning_rate": 6.277823577906018e-05,
+      "loss": 3.5439,
       "step": 2306
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.000628194558944765,
-      "loss": 5.3776,
+      "learning_rate": 6.275762572135202e-05,
+      "loss": 3.0902,
       "step": 2307
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006279884583676834,
-      "loss": 4.9641,
+      "learning_rate": 6.273701566364386e-05,
+      "loss": 2.8989,
       "step": 2308
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006277823577906017,
-      "loss": 4.869,
+      "learning_rate": 6.271640560593571e-05,
+      "loss": 3.1462,
       "step": 2309
     },
     {
       "epoch": 1.29,
-      "learning_rate": 0.0006275762572135201,
-      "loss": 4.9031,
+      "learning_rate": 6.269579554822754e-05,
+      "loss": 2.643,
       "step": 2310
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006273701566364386,
-      "loss": 5.7918,
+      "learning_rate": 6.267518549051937e-05,
+      "loss": 3.3682,
       "step": 2311
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.000627164056059357,
-      "loss": 4.7717,
+      "learning_rate": 6.265457543281121e-05,
+      "loss": 2.7246,
       "step": 2312
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006269579554822754,
-      "loss": 5.0949,
+      "learning_rate": 6.263396537510305e-05,
+      "loss": 3.4536,
       "step": 2313
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006267518549051937,
-      "loss": 5.2501,
+      "learning_rate": 6.26133553173949e-05,
+      "loss": 3.2443,
       "step": 2314
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006265457543281121,
-      "loss": 5.0696,
+      "learning_rate": 6.259274525968673e-05,
+      "loss": 2.8029,
       "step": 2315
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006263396537510305,
-      "loss": 5.3677,
+      "learning_rate": 6.257213520197857e-05,
+      "loss": 2.6596,
       "step": 2316
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006261335531739489,
-      "loss": 5.3111,
+      "learning_rate": 6.255152514427041e-05,
+      "loss": 3.1485,
       "step": 2317
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006259274525968672,
-      "loss": 5.3194,
+      "learning_rate": 6.253091508656224e-05,
+      "loss": 3.4216,
       "step": 2318
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006257213520197856,
-      "loss": 5.0015,
+      "learning_rate": 6.251030502885409e-05,
+      "loss": 2.4332,
       "step": 2319
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.000625515251442704,
-      "loss": 5.27,
+      "learning_rate": 6.248969497114592e-05,
+      "loss": 3.5072,
       "step": 2320
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006253091508656224,
-      "loss": 4.9523,
+      "learning_rate": 6.246908491343776e-05,
+      "loss": 2.551,
       "step": 2321
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006251030502885409,
-      "loss": 4.7454,
+      "learning_rate": 6.24484748557296e-05,
+      "loss": 2.2697,
       "step": 2322
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006248969497114591,
-      "loss": 5.177,
+      "learning_rate": 6.242786479802145e-05,
+      "loss": 2.3461,
       "step": 2323
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006246908491343776,
-      "loss": 5.2175,
+      "learning_rate": 6.240725474031328e-05,
+      "loss": 2.6204,
       "step": 2324
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.000624484748557296,
-      "loss": 5.1673,
+      "learning_rate": 6.23866446826051e-05,
+      "loss": 2.9287,
       "step": 2325
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006242786479802144,
-      "loss": 4.8122,
+      "learning_rate": 6.236603462489695e-05,
+      "loss": 2.3004,
       "step": 2326
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006240725474031328,
-      "loss": 4.4473,
+      "learning_rate": 6.234542456718879e-05,
+      "loss": 1.8396,
       "step": 2327
     },
     {
       "epoch": 1.3,
-      "learning_rate": 0.0006238664468260511,
-      "loss": 4.5248,
+      "learning_rate": 6.232481450948064e-05,
+      "loss": 1.7976,
       "step": 2328
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006236603462489695,
-      "loss": 5.195,
+      "learning_rate": 6.230420445177247e-05,
+      "loss": 2.5259,
       "step": 2329
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.000623454245671888,
-      "loss": 4.9204,
+      "learning_rate": 6.228359439406431e-05,
+      "loss": 1.6817,
       "step": 2330
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006232481450948064,
-      "loss": 4.4068,
+      "learning_rate": 6.226298433635614e-05,
+      "loss": 1.5872,
       "step": 2331
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006230420445177247,
-      "loss": 4.611,
+      "learning_rate": 6.224237427864798e-05,
+      "loss": 1.3839,
       "step": 2332
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006228359439406431,
-      "loss": 5.0655,
+      "learning_rate": 6.222176422093983e-05,
+      "loss": 1.6329,
       "step": 2333
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006226298433635615,
-      "loss": 6.2401,
+      "learning_rate": 6.220115416323166e-05,
+      "loss": 2.3015,
       "step": 2334
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006224237427864799,
-      "loss": 5.3271,
+      "learning_rate": 6.21805441055235e-05,
+      "loss": 3.5212,
       "step": 2335
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006222176422093983,
-      "loss": 4.8161,
+      "learning_rate": 6.215993404781534e-05,
+      "loss": 2.4389,
       "step": 2336
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006220115416323166,
-      "loss": 5.0242,
+      "learning_rate": 6.213932399010717e-05,
+      "loss": 2.9761,
       "step": 2337
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.000621805441055235,
-      "loss": 5.0262,
+      "learning_rate": 6.2118713932399e-05,
+      "loss": 2.5737,
       "step": 2338
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006215993404781534,
-      "loss": 4.9182,
+      "learning_rate": 6.209810387469085e-05,
+      "loss": 3.0437,
       "step": 2339
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006213932399010718,
-      "loss": 4.965,
+      "learning_rate": 6.207749381698269e-05,
+      "loss": 2.6278,
       "step": 2340
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006211871393239901,
-      "loss": 5.4312,
+      "learning_rate": 6.205688375927453e-05,
+      "loss": 3.377,
       "step": 2341
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006209810387469085,
-      "loss": 5.0733,
+      "learning_rate": 6.203627370156638e-05,
+      "loss": 2.7381,
       "step": 2342
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006207749381698269,
-      "loss": 5.4481,
+      "learning_rate": 6.20156636438582e-05,
+      "loss": 3.3325,
       "step": 2343
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006205688375927453,
-      "loss": 4.9876,
+      "learning_rate": 6.199505358615005e-05,
+      "loss": 3.1118,
       "step": 2344
     },
     {
       "epoch": 1.31,
-      "learning_rate": 0.0006203627370156637,
-      "loss": 4.9378,
+      "learning_rate": 6.197444352844188e-05,
+      "loss": 3.4315,
       "step": 2345
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.000620156636438582,
-      "loss": 4.8935,
+      "learning_rate": 6.195383347073372e-05,
+      "loss": 2.6958,
       "step": 2346
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006199505358615004,
-      "loss": 5.1521,
+      "learning_rate": 6.193322341302555e-05,
+      "loss": 2.9481,
       "step": 2347
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006197444352844188,
-      "loss": 4.9547,
+      "learning_rate": 6.19126133553174e-05,
+      "loss": 2.8504,
       "step": 2348
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006195383347073372,
-      "loss": 5.0906,
+      "learning_rate": 6.189200329760924e-05,
+      "loss": 2.9636,
       "step": 2349
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006193322341302555,
-      "loss": 5.3036,
+      "learning_rate": 6.187139323990108e-05,
+      "loss": 3.0127,
       "step": 2350
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.000619126133553174,
-      "loss": 4.9321,
+      "learning_rate": 6.185078318219291e-05,
+      "loss": 3.1621,
       "step": 2351
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006189200329760924,
-      "loss": 4.6555,
+      "learning_rate": 6.183017312448474e-05,
+      "loss": 2.4832,
       "step": 2352
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006187139323990108,
-      "loss": 4.9486,
+      "learning_rate": 6.180956306677659e-05,
+      "loss": 2.6178,
       "step": 2353
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006185078318219292,
-      "loss": 4.6891,
+      "learning_rate": 6.178895300906843e-05,
+      "loss": 2.6115,
       "step": 2354
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006183017312448475,
-      "loss": 4.8177,
+      "learning_rate": 6.176834295136027e-05,
+      "loss": 2.6961,
       "step": 2355
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006180956306677659,
-      "loss": 4.6584,
+      "learning_rate": 6.174773289365212e-05,
+      "loss": 2.4615,
       "step": 2356
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006178895300906843,
-      "loss": 5.3163,
+      "learning_rate": 6.172712283594395e-05,
+      "loss": 3.0262,
       "step": 2357
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006176834295136027,
-      "loss": 4.9132,
+      "learning_rate": 6.170651277823577e-05,
+      "loss": 2.5202,
       "step": 2358
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006174773289365211,
-      "loss": 5.0863,
+      "learning_rate": 6.168590272052762e-05,
+      "loss": 2.8254,
       "step": 2359
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006172712283594394,
-      "loss": 5.227,
+      "learning_rate": 6.166529266281946e-05,
+      "loss": 3.0831,
       "step": 2360
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006170651277823578,
-      "loss": 5.248,
+      "learning_rate": 6.164468260511129e-05,
+      "loss": 2.8089,
       "step": 2361
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006168590272052762,
-      "loss": 5.0642,
+      "learning_rate": 6.162407254740313e-05,
+      "loss": 2.2836,
       "step": 2362
     },
     {
       "epoch": 1.32,
-      "learning_rate": 0.0006166529266281946,
-      "loss": 5.5344,
+      "learning_rate": 6.160346248969498e-05,
+      "loss": 3.1613,
       "step": 2363
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006164468260511129,
-      "loss": 5.3228,
+      "learning_rate": 6.158285243198682e-05,
+      "loss": 3.1121,
       "step": 2364
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006162407254740313,
-      "loss": 5.1179,
+      "learning_rate": 6.156224237427865e-05,
+      "loss": 2.7352,
       "step": 2365
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006160346248969497,
-      "loss": 4.8827,
+      "learning_rate": 6.154163231657048e-05,
+      "loss": 2.9198,
       "step": 2366
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006158285243198681,
-      "loss": 5.3349,
+      "learning_rate": 6.152102225886232e-05,
+      "loss": 3.1005,
       "step": 2367
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006156224237427865,
-      "loss": 5.5565,
+      "learning_rate": 6.150041220115417e-05,
+      "loss": 3.4746,
       "step": 2368
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006154163231657048,
-      "loss": 5.3051,
+      "learning_rate": 6.147980214344601e-05,
+      "loss": 3.0507,
       "step": 2369
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006152102225886232,
-      "loss": 5.0246,
+      "learning_rate": 6.145919208573784e-05,
+      "loss": 2.9319,
       "step": 2370
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006150041220115417,
-      "loss": 4.8395,
+      "learning_rate": 6.143858202802968e-05,
+      "loss": 2.5166,
       "step": 2371
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006147980214344601,
-      "loss": 5.2386,
+      "learning_rate": 6.141797197032151e-05,
+      "loss": 2.6892,
       "step": 2372
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006145919208573784,
-      "loss": 4.8263,
+      "learning_rate": 6.139736191261336e-05,
+      "loss": 2.7951,
       "step": 2373
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006143858202802968,
-      "loss": 5.0549,
+      "learning_rate": 6.13767518549052e-05,
+      "loss": 2.6189,
       "step": 2374
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006141797197032152,
-      "loss": 5.4882,
+      "learning_rate": 6.135614179719703e-05,
+      "loss": 2.6658,
       "step": 2375
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006139736191261336,
-      "loss": 5.1165,
+      "learning_rate": 6.133553173948887e-05,
+      "loss": 2.4371,
       "step": 2376
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.000613767518549052,
-      "loss": 4.8328,
+      "learning_rate": 6.131492168178072e-05,
+      "loss": 2.0395,
       "step": 2377
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006135614179719703,
-      "loss": 4.9474,
+      "learning_rate": 6.129431162407255e-05,
+      "loss": 1.9036,
       "step": 2378
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006133553173948887,
-      "loss": 4.8292,
+      "learning_rate": 6.127370156636439e-05,
+      "loss": 1.9264,
       "step": 2379
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006131492168178071,
-      "loss": 4.9562,
+      "learning_rate": 6.125309150865622e-05,
+      "loss": 2.4247,
       "step": 2380
     },
     {
       "epoch": 1.33,
-      "learning_rate": 0.0006129431162407255,
-      "loss": 4.9211,
+      "learning_rate": 6.123248145094806e-05,
+      "loss": 2.4179,
       "step": 2381
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006127370156636439,
-      "loss": 4.9979,
+      "learning_rate": 6.121187139323991e-05,
+      "loss": 1.8373,
       "step": 2382
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006125309150865622,
-      "loss": 4.5683,
+      "learning_rate": 6.119126133553175e-05,
+      "loss": 1.4626,
       "step": 2383
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006123248145094806,
-      "loss": 4.862,
+      "learning_rate": 6.117065127782358e-05,
+      "loss": 1.6543,
       "step": 2384
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.000612118713932399,
-      "loss": 5.1538,
+      "learning_rate": 6.115004122011541e-05,
+      "loss": 2.9581,
       "step": 2385
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006119126133553174,
-      "loss": 5.3313,
+      "learning_rate": 6.112943116240725e-05,
+      "loss": 3.0596,
       "step": 2386
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006117065127782357,
-      "loss": 5.1489,
+      "learning_rate": 6.11088211046991e-05,
+      "loss": 2.8197,
       "step": 2387
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006115004122011541,
-      "loss": 4.7654,
+      "learning_rate": 6.108821104699094e-05,
+      "loss": 2.6404,
       "step": 2388
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006112943116240725,
-      "loss": 4.9831,
+      "learning_rate": 6.106760098928277e-05,
+      "loss": 2.9391,
       "step": 2389
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.000611088211046991,
-      "loss": 5.0932,
+      "learning_rate": 6.104699093157461e-05,
+      "loss": 3.0716,
       "step": 2390
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006108821104699094,
-      "loss": 5.1229,
+      "learning_rate": 6.102638087386645e-05,
+      "loss": 3.0091,
       "step": 2391
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006106760098928277,
-      "loss": 5.5162,
+      "learning_rate": 6.1005770816158294e-05,
+      "loss": 3.4403,
       "step": 2392
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006104699093157461,
-      "loss": 5.3291,
+      "learning_rate": 6.0985160758450124e-05,
+      "loss": 3.1348,
       "step": 2393
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006102638087386645,
-      "loss": 5.0907,
+      "learning_rate": 6.096455070074196e-05,
+      "loss": 2.5814,
       "step": 2394
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006100577081615829,
-      "loss": 4.4309,
+      "learning_rate": 6.0943940643033804e-05,
+      "loss": 2.3418,
       "step": 2395
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006098516075845012,
-      "loss": 4.7308,
+      "learning_rate": 6.092333058532565e-05,
+      "loss": 2.537,
       "step": 2396
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006096455070074196,
-      "loss": 4.9618,
+      "learning_rate": 6.0902720527617484e-05,
+      "loss": 2.8993,
       "step": 2397
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.000609439406430338,
-      "loss": 4.7887,
+      "learning_rate": 6.0882110469909314e-05,
+      "loss": 2.4182,
       "step": 2398
     },
     {
       "epoch": 1.34,
-      "learning_rate": 0.0006092333058532564,
-      "loss": 5.6544,
+      "learning_rate": 6.086150041220116e-05,
+      "loss": 3.6892,
       "step": 2399
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006090272052761748,
-      "loss": 5.1565,
+      "learning_rate": 6.0840890354492994e-05,
+      "loss": 3.3522,
       "step": 2400
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006088211046990931,
-      "loss": 5.6548,
+      "learning_rate": 6.082028029678484e-05,
+      "loss": 3.6666,
       "step": 2401
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006086150041220115,
-      "loss": 4.8998,
+      "learning_rate": 6.079967023907667e-05,
+      "loss": 2.4575,
       "step": 2402
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006084089035449299,
-      "loss": 5.5053,
+      "learning_rate": 6.077906018136851e-05,
+      "loss": 3.1409,
       "step": 2403
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006082028029678483,
-      "loss": 5.9069,
+      "learning_rate": 6.075845012366035e-05,
+      "loss": 3.1828,
       "step": 2404
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006079967023907666,
-      "loss": 5.4749,
+      "learning_rate": 6.073784006595219e-05,
+      "loss": 3.3796,
       "step": 2405
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.000607790601813685,
-      "loss": 4.9121,
+      "learning_rate": 6.0717230008244034e-05,
+      "loss": 3.0578,
       "step": 2406
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006075845012366034,
-      "loss": 4.8255,
+      "learning_rate": 6.069661995053586e-05,
+      "loss": 2.8674,
       "step": 2407
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006073784006595218,
-      "loss": 4.996,
+      "learning_rate": 6.06760098928277e-05,
+      "loss": 2.9745,
       "step": 2408
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006071723000824402,
-      "loss": 5.1797,
+      "learning_rate": 6.065539983511954e-05,
+      "loss": 2.6305,
       "step": 2409
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006069661995053585,
-      "loss": 4.9398,
+      "learning_rate": 6.063478977741138e-05,
+      "loss": 2.4415,
       "step": 2410
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006067600989282769,
-      "loss": 4.8418,
+      "learning_rate": 6.061417971970322e-05,
+      "loss": 2.8933,
       "step": 2411
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006065539983511954,
-      "loss": 4.8482,
+      "learning_rate": 6.059356966199505e-05,
+      "loss": 2.743,
       "step": 2412
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006063478977741138,
-      "loss": 5.5902,
+      "learning_rate": 6.0572959604286897e-05,
+      "loss": 3.325,
       "step": 2413
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006061417971970322,
-      "loss": 4.8184,
+      "learning_rate": 6.055234954657873e-05,
+      "loss": 2.3968,
       "step": 2414
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006059356966199505,
-      "loss": 4.8406,
+      "learning_rate": 6.0531739488870576e-05,
+      "loss": 2.4678,
       "step": 2415
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.000605729596042869,
-      "loss": 5.9371,
+      "learning_rate": 6.0511129431162406e-05,
+      "loss": 3.4014,
       "step": 2416
     },
     {
       "epoch": 1.35,
-      "learning_rate": 0.0006055234954657874,
-      "loss": 4.9705,
+      "learning_rate": 6.049051937345425e-05,
+      "loss": 2.3017,
       "step": 2417
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006053173948887058,
-      "loss": 5.1964,
+      "learning_rate": 6.0469909315746086e-05,
+      "loss": 3.091,
       "step": 2418
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006051112943116241,
-      "loss": 5.2744,
+      "learning_rate": 6.044929925803793e-05,
+      "loss": 2.7368,
       "step": 2419
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006049051937345425,
-      "loss": 4.9904,
+      "learning_rate": 6.0428689200329766e-05,
+      "loss": 2.7886,
       "step": 2420
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006046990931574609,
-      "loss": 5.239,
+      "learning_rate": 6.0408079142621596e-05,
+      "loss": 2.9839,
       "step": 2421
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006044929925803793,
-      "loss": 5.1387,
+      "learning_rate": 6.038746908491344e-05,
+      "loss": 2.3665,
       "step": 2422
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006042868920032977,
-      "loss": 4.7403,
+      "learning_rate": 6.036685902720528e-05,
+      "loss": 2.1925,
       "step": 2423
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.000604080791426216,
-      "loss": 5.0144,
+      "learning_rate": 6.034624896949712e-05,
+      "loss": 2.6814,
       "step": 2424
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006038746908491344,
-      "loss": 4.9431,
+      "learning_rate": 6.032563891178895e-05,
+      "loss": 1.9783,
       "step": 2425
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006036685902720528,
-      "loss": 5.0025,
+      "learning_rate": 6.030502885408079e-05,
+      "loss": 2.8178,
       "step": 2426
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006034624896949712,
-      "loss": 4.8716,
+      "learning_rate": 6.028441879637263e-05,
+      "loss": 2.0982,
       "step": 2427
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006032563891178895,
-      "loss": 4.8287,
+      "learning_rate": 6.026380873866447e-05,
+      "loss": 2.0928,
       "step": 2428
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006030502885408079,
-      "loss": 4.7636,
+      "learning_rate": 6.0243198680956316e-05,
+      "loss": 1.8741,
       "step": 2429
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006028441879637263,
-      "loss": 4.6974,
+      "learning_rate": 6.0222588623248146e-05,
+      "loss": 1.7497,
       "step": 2430
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006026380873866448,
-      "loss": 4.9268,
+      "learning_rate": 6.020197856553998e-05,
+      "loss": 1.9365,
       "step": 2431
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006024319868095632,
-      "loss": 4.4698,
+      "learning_rate": 6.0181368507831826e-05,
+      "loss": 1.4967,
       "step": 2432
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006022258862324815,
-      "loss": 7.1452,
+      "learning_rate": 6.016075845012367e-05,
+      "loss": 2.0995,
       "step": 2433
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006020197856553999,
-      "loss": 4.905,
+      "learning_rate": 6.01401483924155e-05,
+      "loss": 1.689,
       "step": 2434
     },
     {
       "epoch": 1.36,
-      "learning_rate": 0.0006018136850783183,
-      "loss": 5.0231,
+      "learning_rate": 6.0119538334707336e-05,
+      "loss": 2.8742,
       "step": 2435
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0006016075845012367,
-      "loss": 5.8013,
+      "learning_rate": 6.009892827699918e-05,
+      "loss": 3.8193,
       "step": 2436
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.000601401483924155,
-      "loss": 5.5135,
+      "learning_rate": 6.0078318219291016e-05,
+      "loss": 3.5339,
       "step": 2437
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0006011953833470734,
-      "loss": 5.3022,
+      "learning_rate": 6.005770816158286e-05,
+      "loss": 3.1693,
       "step": 2438
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0006009892827699918,
-      "loss": 5.0997,
+      "learning_rate": 6.003709810387469e-05,
+      "loss": 2.6553,
       "step": 2439
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0006007831821929102,
-      "loss": 4.9978,
+      "learning_rate": 6.001648804616653e-05,
+      "loss": 2.7825,
       "step": 2440
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0006005770816158286,
-      "loss": 5.2948,
+      "learning_rate": 5.999587798845837e-05,
+      "loss": 3.0228,
       "step": 2441
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0006003709810387469,
-      "loss": 5.1099,
+      "learning_rate": 5.997526793075021e-05,
+      "loss": 2.72,
       "step": 2442
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0006001648804616653,
-      "loss": 4.9653,
+      "learning_rate": 5.9954657873042056e-05,
+      "loss": 2.601,
       "step": 2443
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005999587798845837,
-      "loss": 4.9423,
+      "learning_rate": 5.9934047815333885e-05,
+      "loss": 2.7526,
       "step": 2444
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005997526793075021,
-      "loss": 5.2526,
+      "learning_rate": 5.991343775762572e-05,
+      "loss": 3.2897,
       "step": 2445
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005995465787304205,
-      "loss": 5.3132,
+      "learning_rate": 5.9892827699917565e-05,
+      "loss": 3.1954,
       "step": 2446
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005993404781533388,
-      "loss": 5.2025,
+      "learning_rate": 5.98722176422094e-05,
+      "loss": 2.8525,
       "step": 2447
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005991343775762572,
-      "loss": 5.227,
+      "learning_rate": 5.985160758450123e-05,
+      "loss": 2.8722,
       "step": 2448
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005989282769991756,
-      "loss": 5.0464,
+      "learning_rate": 5.9830997526793075e-05,
+      "loss": 2.5258,
       "step": 2449
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.000598722176422094,
-      "loss": 5.1404,
+      "learning_rate": 5.981038746908492e-05,
+      "loss": 2.9769,
       "step": 2450
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005985160758450123,
-      "loss": 4.9662,
+      "learning_rate": 5.9789777411376755e-05,
+      "loss": 2.9419,
       "step": 2451
     },
     {
       "epoch": 1.37,
-      "learning_rate": 0.0005983099752679308,
-      "loss": 5.353,
+      "learning_rate": 5.97691673536686e-05,
+      "loss": 3.1774,
       "step": 2452
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005981038746908492,
-      "loss": 5.2103,
+      "learning_rate": 5.974855729596043e-05,
+      "loss": 3.4088,
       "step": 2453
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005978977741137676,
-      "loss": 4.8355,
+      "learning_rate": 5.972794723825227e-05,
+      "loss": 2.9289,
       "step": 2454
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.000597691673536686,
-      "loss": 4.9085,
+      "learning_rate": 5.970733718054411e-05,
+      "loss": 2.3732,
       "step": 2455
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005974855729596043,
-      "loss": 4.9233,
+      "learning_rate": 5.968672712283595e-05,
+      "loss": 2.6231,
       "step": 2456
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005972794723825227,
-      "loss": 5.3614,
+      "learning_rate": 5.966611706512778e-05,
+      "loss": 2.6747,
       "step": 2457
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005970733718054411,
-      "loss": 4.9241,
+      "learning_rate": 5.964550700741962e-05,
+      "loss": 3.2799,
       "step": 2458
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005968672712283595,
-      "loss": 5.1216,
+      "learning_rate": 5.962489694971146e-05,
+      "loss": 2.81,
       "step": 2459
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005966611706512778,
-      "loss": 4.6808,
+      "learning_rate": 5.9604286892003305e-05,
+      "loss": 2.2646,
       "step": 2460
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005964550700741962,
-      "loss": 4.9575,
+      "learning_rate": 5.958367683429514e-05,
+      "loss": 2.4862,
       "step": 2461
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005962489694971146,
-      "loss": 4.9834,
+      "learning_rate": 5.956306677658697e-05,
+      "loss": 2.6333,
       "step": 2462
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.000596042868920033,
-      "loss": 5.1103,
+      "learning_rate": 5.9542456718878815e-05,
+      "loss": 2.9533,
       "step": 2463
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005958367683429514,
-      "loss": 4.9371,
+      "learning_rate": 5.952184666117066e-05,
+      "loss": 2.4432,
       "step": 2464
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005956306677658697,
-      "loss": 5.0491,
+      "learning_rate": 5.9501236603462495e-05,
+      "loss": 2.6672,
       "step": 2465
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005954245671887881,
-      "loss": 4.703,
+      "learning_rate": 5.9480626545754325e-05,
+      "loss": 2.1499,
       "step": 2466
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005952184666117065,
-      "loss": 5.0961,
+      "learning_rate": 5.946001648804617e-05,
+      "loss": 3.028,
       "step": 2467
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005950123660346249,
-      "loss": 5.564,
+      "learning_rate": 5.9439406430338005e-05,
+      "loss": 2.8239,
       "step": 2468
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005948062654575432,
-      "loss": 5.2638,
+      "learning_rate": 5.941879637262985e-05,
+      "loss": 2.8436,
       "step": 2469
     },
     {
       "epoch": 1.38,
-      "learning_rate": 0.0005946001648804616,
-      "loss": 5.0757,
+      "learning_rate": 5.939818631492169e-05,
+      "loss": 2.4546,
       "step": 2470
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.00059439406430338,
-      "loss": 4.9441,
+      "learning_rate": 5.937757625721352e-05,
+      "loss": 2.6532,
       "step": 2471
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005941879637262985,
-      "loss": 4.7709,
+      "learning_rate": 5.935696619950536e-05,
+      "loss": 2.7255,
       "step": 2472
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005939818631492169,
-      "loss": 4.779,
+      "learning_rate": 5.93363561417972e-05,
+      "loss": 2.952,
       "step": 2473
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005937757625721352,
-      "loss": 4.6202,
+      "learning_rate": 5.931574608408904e-05,
+      "loss": 2.4525,
       "step": 2474
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005935696619950536,
-      "loss": 5.1384,
+      "learning_rate": 5.929513602638088e-05,
+      "loss": 2.8355,
       "step": 2475
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.000593363561417972,
-      "loss": 5.6382,
+      "learning_rate": 5.927452596867271e-05,
+      "loss": 3.2174,
       "step": 2476
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005931574608408904,
-      "loss": 5.0036,
+      "learning_rate": 5.9253915910964554e-05,
+      "loss": 2.5174,
       "step": 2477
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005929513602638088,
-      "loss": 5.2854,
+      "learning_rate": 5.923330585325639e-05,
+      "loss": 2.1773,
       "step": 2478
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005927452596867271,
-      "loss": 4.5547,
+      "learning_rate": 5.9212695795548234e-05,
+      "loss": 2.2883,
       "step": 2479
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005925391591096455,
-      "loss": 5.3497,
+      "learning_rate": 5.9192085737840064e-05,
+      "loss": 2.3535,
       "step": 2480
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005923330585325639,
-      "loss": 4.6184,
+      "learning_rate": 5.917147568013191e-05,
+      "loss": 1.579,
       "step": 2481
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005921269579554823,
-      "loss": 4.6025,
+      "learning_rate": 5.9150865622423744e-05,
+      "loss": 1.6437,
       "step": 2482
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005919208573784006,
-      "loss": 4.6355,
+      "learning_rate": 5.913025556471559e-05,
+      "loss": 1.4447,
       "step": 2483
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.000591714756801319,
-      "loss": 5.2645,
+      "learning_rate": 5.9109645507007424e-05,
+      "loss": 1.8651,
       "step": 2484
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005915086562242374,
-      "loss": 6.3168,
+      "learning_rate": 5.9089035449299254e-05,
+      "loss": 3.7344,
       "step": 2485
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005913025556471558,
-      "loss": 4.9091,
+      "learning_rate": 5.90684253915911e-05,
+      "loss": 2.674,
       "step": 2486
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005910964550700742,
-      "loss": 4.9234,
+      "learning_rate": 5.904781533388294e-05,
+      "loss": 2.6046,
       "step": 2487
     },
     {
       "epoch": 1.39,
-      "learning_rate": 0.0005908903544929925,
-      "loss": 5.3417,
+      "learning_rate": 5.902720527617478e-05,
+      "loss": 3.397,
       "step": 2488
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005906842539159109,
-      "loss": 4.9838,
+      "learning_rate": 5.900659521846661e-05,
+      "loss": 2.874,
       "step": 2489
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005904781533388293,
-      "loss": 5.2376,
+      "learning_rate": 5.898598516075845e-05,
+      "loss": 2.789,
       "step": 2490
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005902720527617477,
-      "loss": 4.967,
+      "learning_rate": 5.8965375103050294e-05,
+      "loss": 2.9742,
       "step": 2491
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.000590065952184666,
-      "loss": 5.5947,
+      "learning_rate": 5.894476504534213e-05,
+      "loss": 3.877,
       "step": 2492
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005898598516075845,
-      "loss": 4.8256,
+      "learning_rate": 5.8924154987633974e-05,
+      "loss": 3.0205,
       "step": 2493
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005896537510305029,
-      "loss": 4.8316,
+      "learning_rate": 5.8903544929925804e-05,
+      "loss": 3.0722,
       "step": 2494
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005894476504534213,
-      "loss": 4.8631,
+      "learning_rate": 5.888293487221764e-05,
+      "loss": 3.0743,
       "step": 2495
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005892415498763397,
-      "loss": 4.861,
+      "learning_rate": 5.8862324814509484e-05,
+      "loss": 2.6067,
       "step": 2496
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.000589035449299258,
-      "loss": 5.2099,
+      "learning_rate": 5.884171475680133e-05,
+      "loss": 3.1679,
       "step": 2497
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005888293487221764,
-      "loss": 5.1522,
+      "learning_rate": 5.8821104699093164e-05,
+      "loss": 3.018,
       "step": 2498
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005886232481450948,
-      "loss": 4.8832,
+      "learning_rate": 5.8800494641384993e-05,
+      "loss": 2.8599,
       "step": 2499
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005884171475680132,
-      "loss": 4.9413,
+      "learning_rate": 5.877988458367684e-05,
+      "loss": 3.0003,
       "step": 2500
     },
     {
       "epoch": 1.4,
-      "eval_loss": 5.506562232971191,
-      "eval_runtime": 1247.7608,
-      "eval_samples_per_second": 2.117,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.9556326854422847,
+      "eval_loss": 6.095541477203369,
+      "eval_runtime": 1238.2516,
+      "eval_samples_per_second": 2.134,
+      "eval_steps_per_second": 0.267,
+      "eval_wer": 1.7624752082506943,
       "step": 2500
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005882110469909316,
-      "loss": 4.7913,
+      "learning_rate": 5.875927452596868e-05,
+      "loss": 2.7114,
       "step": 2501
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005880049464138499,
-      "loss": 4.9295,
+      "learning_rate": 5.873866446826052e-05,
+      "loss": 3.3152,
       "step": 2502
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005877988458367683,
-      "loss": 5.0334,
+      "learning_rate": 5.871805441055235e-05,
+      "loss": 2.8311,
       "step": 2503
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005875927452596867,
-      "loss": 4.5795,
+      "learning_rate": 5.869744435284419e-05,
+      "loss": 2.2888,
       "step": 2504
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005873866446826051,
-      "loss": 4.7373,
+      "learning_rate": 5.867683429513603e-05,
+      "loss": 2.5095,
       "step": 2505
     },
     {
       "epoch": 1.4,
-      "learning_rate": 0.0005871805441055234,
-      "loss": 5.14,
+      "learning_rate": 5.865622423742787e-05,
+      "loss": 2.9124,
       "step": 2506
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005869744435284418,
-      "loss": 4.7387,
+      "learning_rate": 5.8635614179719713e-05,
+      "loss": 2.3412,
       "step": 2507
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005867683429513602,
-      "loss": 4.9077,
+      "learning_rate": 5.861500412201154e-05,
+      "loss": 2.3546,
       "step": 2508
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005865622423742787,
-      "loss": 4.6913,
+      "learning_rate": 5.859439406430338e-05,
+      "loss": 2.3909,
       "step": 2509
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005863561417971971,
-      "loss": 5.4268,
+      "learning_rate": 5.857378400659522e-05,
+      "loss": 3.3647,
       "step": 2510
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005861500412201154,
-      "loss": 4.8315,
+      "learning_rate": 5.8553173948887067e-05,
+      "loss": 2.3893,
       "step": 2511
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005859439406430339,
-      "loss": 5.0394,
+      "learning_rate": 5.8532563891178896e-05,
+      "loss": 2.6503,
       "step": 2512
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005857378400659523,
-      "loss": 4.6983,
+      "learning_rate": 5.851195383347073e-05,
+      "loss": 2.4587,
       "step": 2513
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005855317394888707,
-      "loss": 4.8267,
+      "learning_rate": 5.8491343775762576e-05,
+      "loss": 2.3209,
       "step": 2514
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.000585325638911789,
-      "loss": 5.5017,
+      "learning_rate": 5.847073371805441e-05,
+      "loss": 3.3875,
       "step": 2515
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005851195383347074,
-      "loss": 4.9261,
+      "learning_rate": 5.8450123660346256e-05,
+      "loss": 2.7579,
       "step": 2516
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005849134377576258,
-      "loss": 4.8778,
+      "learning_rate": 5.8429513602638086e-05,
+      "loss": 2.908,
       "step": 2517
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005847073371805442,
-      "loss": 5.4911,
+      "learning_rate": 5.840890354492993e-05,
+      "loss": 2.9482,
       "step": 2518
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005845012366034626,
-      "loss": 5.2547,
+      "learning_rate": 5.8388293487221766e-05,
+      "loss": 3.5646,
       "step": 2519
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005842951360263809,
-      "loss": 4.9811,
+      "learning_rate": 5.836768342951361e-05,
+      "loss": 2.6137,
       "step": 2520
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005840890354492993,
-      "loss": 4.8932,
+      "learning_rate": 5.834707337180544e-05,
+      "loss": 2.2288,
       "step": 2521
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005838829348722177,
-      "loss": 4.6617,
+      "learning_rate": 5.8326463314097276e-05,
+      "loss": 2.1965,
       "step": 2522
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005836768342951361,
-      "loss": 4.4748,
+      "learning_rate": 5.830585325638912e-05,
+      "loss": 2.1787,
       "step": 2523
     },
     {
       "epoch": 1.41,
-      "learning_rate": 0.0005834707337180544,
-      "loss": 5.2021,
+      "learning_rate": 5.828524319868096e-05,
+      "loss": 2.8854,
       "step": 2524
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005832646331409728,
-      "loss": 4.5598,
+      "learning_rate": 5.82646331409728e-05,
+      "loss": 2.3586,
       "step": 2525
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005830585325638912,
-      "loss": 5.0887,
+      "learning_rate": 5.824402308326463e-05,
+      "loss": 2.1884,
       "step": 2526
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005828524319868096,
-      "loss": 5.33,
+      "learning_rate": 5.822341302555647e-05,
+      "loss": 2.5619,
       "step": 2527
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.000582646331409728,
-      "loss": 4.765,
+      "learning_rate": 5.8202802967848316e-05,
+      "loss": 2.2084,
       "step": 2528
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005824402308326463,
-      "loss": 4.6768,
+      "learning_rate": 5.818219291014015e-05,
+      "loss": 2.0146,
       "step": 2529
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005822341302555647,
-      "loss": 5.7151,
+      "learning_rate": 5.8161582852431996e-05,
+      "loss": 2.6248,
       "step": 2530
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005820280296784831,
-      "loss": 4.93,
+      "learning_rate": 5.8140972794723826e-05,
+      "loss": 1.8787,
       "step": 2531
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005818219291014016,
-      "loss": 5.4786,
+      "learning_rate": 5.812036273701566e-05,
+      "loss": 2.3823,
       "step": 2532
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.00058161582852432,
-      "loss": 5.3676,
+      "learning_rate": 5.8099752679307506e-05,
+      "loss": 1.8951,
       "step": 2533
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005814097279472383,
-      "loss": 5.1701,
+      "learning_rate": 5.807914262159935e-05,
+      "loss": 1.9408,
       "step": 2534
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005812036273701567,
-      "loss": 4.8913,
+      "learning_rate": 5.805853256389118e-05,
+      "loss": 2.8705,
       "step": 2535
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005809975267930751,
-      "loss": 5.4508,
+      "learning_rate": 5.8037922506183016e-05,
+      "loss": 3.2583,
       "step": 2536
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005807914262159935,
-      "loss": 5.6667,
+      "learning_rate": 5.801731244847486e-05,
+      "loss": 3.4612,
       "step": 2537
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005805853256389118,
-      "loss": 6.0381,
+      "learning_rate": 5.79967023907667e-05,
+      "loss": 3.4066,
       "step": 2538
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005803792250618302,
-      "loss": 5.2134,
+      "learning_rate": 5.797609233305854e-05,
+      "loss": 2.9945,
       "step": 2539
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005801731244847486,
-      "loss": 5.1166,
+      "learning_rate": 5.795548227535037e-05,
+      "loss": 2.6599,
       "step": 2540
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.000579967023907667,
-      "loss": 5.0468,
+      "learning_rate": 5.793487221764221e-05,
+      "loss": 3.0533,
       "step": 2541
     },
     {
       "epoch": 1.42,
-      "learning_rate": 0.0005797609233305854,
-      "loss": 4.9933,
+      "learning_rate": 5.791426215993405e-05,
+      "loss": 2.9956,
       "step": 2542
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005795548227535037,
-      "loss": 4.8563,
+      "learning_rate": 5.789365210222589e-05,
+      "loss": 2.8221,
       "step": 2543
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005793487221764221,
-      "loss": 4.8789,
+      "learning_rate": 5.787304204451772e-05,
+      "loss": 2.7069,
       "step": 2544
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005791426215993405,
-      "loss": 4.8522,
+      "learning_rate": 5.7852431986809565e-05,
+      "loss": 3.2652,
       "step": 2545
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005789365210222589,
-      "loss": 4.8898,
+      "learning_rate": 5.78318219291014e-05,
+      "loss": 2.6537,
       "step": 2546
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005787304204451772,
-      "loss": 5.0933,
+      "learning_rate": 5.7811211871393245e-05,
+      "loss": 2.9564,
       "step": 2547
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005785243198680956,
-      "loss": 4.9885,
+      "learning_rate": 5.779060181368509e-05,
+      "loss": 2.7268,
       "step": 2548
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.000578318219291014,
-      "loss": 4.612,
+      "learning_rate": 5.776999175597692e-05,
+      "loss": 2.4479,
       "step": 2549
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005781121187139324,
-      "loss": 4.8738,
+      "learning_rate": 5.7749381698268755e-05,
+      "loss": 2.587,
       "step": 2550
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005779060181368508,
-      "loss": 4.7944,
+      "learning_rate": 5.77287716405606e-05,
+      "loss": 2.4494,
       "step": 2551
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005776999175597691,
-      "loss": 5.0589,
+      "learning_rate": 5.7708161582852435e-05,
+      "loss": 3.1919,
       "step": 2552
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005774938169826876,
-      "loss": 5.0891,
+      "learning_rate": 5.7687551525144265e-05,
+      "loss": 2.7558,
       "step": 2553
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.000577287716405606,
-      "loss": 5.512,
+      "learning_rate": 5.766694146743611e-05,
+      "loss": 3.4355,
       "step": 2554
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005770816158285244,
-      "loss": 5.0868,
+      "learning_rate": 5.764633140972795e-05,
+      "loss": 2.673,
       "step": 2555
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005768755152514427,
-      "loss": 4.6583,
+      "learning_rate": 5.762572135201979e-05,
+      "loss": 2.9026,
       "step": 2556
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005766694146743611,
-      "loss": 5.1158,
+      "learning_rate": 5.760511129431163e-05,
+      "loss": 2.8247,
       "step": 2557
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005764633140972795,
-      "loss": 4.7543,
+      "learning_rate": 5.758450123660346e-05,
+      "loss": 2.7505,
       "step": 2558
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005762572135201979,
-      "loss": 5.1819,
+      "learning_rate": 5.7563891178895305e-05,
+      "loss": 2.9012,
       "step": 2559
     },
     {
       "epoch": 1.43,
-      "learning_rate": 0.0005760511129431163,
-      "loss": 5.0667,
+      "learning_rate": 5.754328112118714e-05,
+      "loss": 3.1595,
       "step": 2560
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005758450123660346,
-      "loss": 4.8594,
+      "learning_rate": 5.7522671063478985e-05,
+      "loss": 2.6385,
       "step": 2561
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.000575638911788953,
-      "loss": 4.8421,
+      "learning_rate": 5.750206100577082e-05,
+      "loss": 2.7437,
       "step": 2562
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005754328112118714,
-      "loss": 4.8633,
+      "learning_rate": 5.748145094806265e-05,
+      "loss": 2.4477,
       "step": 2563
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005752267106347898,
-      "loss": 4.9137,
+      "learning_rate": 5.7460840890354495e-05,
+      "loss": 2.549,
       "step": 2564
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005750206100577082,
-      "loss": 4.946,
+      "learning_rate": 5.744023083264634e-05,
+      "loss": 2.9459,
       "step": 2565
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005748145094806265,
-      "loss": 4.928,
+      "learning_rate": 5.7419620774938175e-05,
+      "loss": 2.4985,
       "step": 2566
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005746084089035449,
-      "loss": 5.1521,
+      "learning_rate": 5.7399010717230004e-05,
+      "loss": 2.843,
       "step": 2567
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005744023083264633,
-      "loss": 4.8473,
+      "learning_rate": 5.737840065952185e-05,
+      "loss": 2.5881,
       "step": 2568
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005741962077493817,
-      "loss": 5.0472,
+      "learning_rate": 5.735779060181369e-05,
+      "loss": 2.7705,
       "step": 2569
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005739901071723,
-      "loss": 5.2403,
+      "learning_rate": 5.733718054410553e-05,
+      "loss": 2.671,
       "step": 2570
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005737840065952184,
-      "loss": 5.1523,
+      "learning_rate": 5.731657048639737e-05,
+      "loss": 2.2987,
       "step": 2571
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005735779060181368,
-      "loss": 5.2161,
+      "learning_rate": 5.72959604286892e-05,
+      "loss": 2.857,
       "step": 2572
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005733718054410553,
-      "loss": 5.8123,
+      "learning_rate": 5.727535037098104e-05,
+      "loss": 3.3149,
       "step": 2573
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005731657048639737,
-      "loss": 5.4737,
+      "learning_rate": 5.725474031327288e-05,
+      "loss": 3.1982,
       "step": 2574
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.000572959604286892,
-      "loss": 4.6615,
+      "learning_rate": 5.7234130255564724e-05,
+      "loss": 2.2183,
       "step": 2575
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005727535037098104,
-      "loss": 4.8602,
+      "learning_rate": 5.7213520197856554e-05,
+      "loss": 2.4399,
       "step": 2576
     },
     {
       "epoch": 1.44,
-      "learning_rate": 0.0005725474031327288,
-      "loss": 5.575,
+      "learning_rate": 5.719291014014839e-05,
+      "loss": 3.1773,
       "step": 2577
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005723413025556472,
-      "loss": 4.7768,
+      "learning_rate": 5.7172300082440234e-05,
+      "loss": 2.1794,
       "step": 2578
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005721352019785655,
-      "loss": 4.6421,
+      "learning_rate": 5.715169002473207e-05,
+      "loss": 1.8338,
       "step": 2579
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005719291014014839,
-      "loss": 4.9369,
+      "learning_rate": 5.7131079967023914e-05,
+      "loss": 2.1556,
       "step": 2580
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005717230008244023,
-      "loss": 5.3888,
+      "learning_rate": 5.7110469909315744e-05,
+      "loss": 2.3237,
       "step": 2581
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005715169002473207,
-      "loss": 6.6137,
+      "learning_rate": 5.708985985160759e-05,
+      "loss": 2.1791,
       "step": 2582
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005713107996702391,
-      "loss": 4.9025,
+      "learning_rate": 5.7069249793899424e-05,
+      "loss": 2.3684,
       "step": 2583
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005711046990931574,
-      "loss": 4.7283,
+      "learning_rate": 5.704863973619127e-05,
+      "loss": 1.7419,
       "step": 2584
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005708985985160758,
-      "loss": 5.0137,
+      "learning_rate": 5.70280296784831e-05,
+      "loss": 3.2892,
       "step": 2585
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005706924979389942,
-      "loss": 5.0905,
+      "learning_rate": 5.700741962077494e-05,
+      "loss": 3.4524,
       "step": 2586
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005704863973619126,
-      "loss": 5.0072,
+      "learning_rate": 5.698680956306678e-05,
+      "loss": 3.2494,
       "step": 2587
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005702802967848309,
-      "loss": 4.9008,
+      "learning_rate": 5.696619950535862e-05,
+      "loss": 2.6842,
       "step": 2588
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005700741962077493,
-      "loss": 4.7538,
+      "learning_rate": 5.694558944765046e-05,
+      "loss": 2.5109,
       "step": 2589
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005698680956306677,
-      "loss": 4.9587,
+      "learning_rate": 5.692497938994229e-05,
+      "loss": 3.0176,
       "step": 2590
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005696619950535861,
-      "loss": 4.807,
+      "learning_rate": 5.690436933223413e-05,
+      "loss": 2.8526,
       "step": 2591
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005694558944765045,
-      "loss": 4.9758,
+      "learning_rate": 5.6883759274525974e-05,
+      "loss": 2.838,
       "step": 2592
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005692497938994228,
-      "loss": 4.9983,
+      "learning_rate": 5.686314921681781e-05,
+      "loss": 2.9057,
       "step": 2593
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005690436933223412,
-      "loss": 4.9617,
+      "learning_rate": 5.6842539159109654e-05,
+      "loss": 2.7319,
       "step": 2594
     },
     {
       "epoch": 1.45,
-      "learning_rate": 0.0005688375927452597,
-      "loss": 4.9015,
+      "learning_rate": 5.6821929101401484e-05,
+      "loss": 2.874,
       "step": 2595
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005686314921681781,
-      "loss": 4.7615,
+      "learning_rate": 5.680131904369333e-05,
+      "loss": 2.5215,
       "step": 2596
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005684253915910965,
-      "loss": 4.9225,
+      "learning_rate": 5.6780708985985164e-05,
+      "loss": 2.5739,
       "step": 2597
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005682192910140148,
-      "loss": 5.0747,
+      "learning_rate": 5.676009892827701e-05,
+      "loss": 2.9298,
       "step": 2598
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005680131904369332,
-      "loss": 4.7409,
+      "learning_rate": 5.673948887056884e-05,
+      "loss": 2.6998,
       "step": 2599
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005678070898598516,
-      "loss": 5.337,
+      "learning_rate": 5.671887881286067e-05,
+      "loss": 3.3217,
       "step": 2600
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.00056760098928277,
-      "loss": 5.0503,
+      "learning_rate": 5.669826875515252e-05,
+      "loss": 2.775,
       "step": 2601
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005673948887056883,
-      "loss": 4.8812,
+      "learning_rate": 5.667765869744436e-05,
+      "loss": 2.5044,
       "step": 2602
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005671887881286068,
-      "loss": 4.982,
+      "learning_rate": 5.66570486397362e-05,
+      "loss": 2.9763,
       "step": 2603
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005669826875515252,
-      "loss": 4.967,
+      "learning_rate": 5.6636438582028027e-05,
+      "loss": 2.3294,
       "step": 2604
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005667765869744436,
-      "loss": 5.1395,
+      "learning_rate": 5.661582852431987e-05,
+      "loss": 3.1067,
       "step": 2605
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.000566570486397362,
-      "loss": 4.6981,
+      "learning_rate": 5.659521846661171e-05,
+      "loss": 2.2273,
       "step": 2606
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005663643858202803,
-      "loss": 4.7929,
+      "learning_rate": 5.657460840890355e-05,
+      "loss": 2.4426,
       "step": 2607
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005661582852431987,
-      "loss": 4.8682,
+      "learning_rate": 5.655399835119538e-05,
+      "loss": 2.3481,
       "step": 2608
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005659521846661171,
-      "loss": 4.5313,
+      "learning_rate": 5.653338829348722e-05,
+      "loss": 2.3774,
       "step": 2609
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005657460840890355,
-      "loss": 5.3326,
+      "learning_rate": 5.651277823577906e-05,
+      "loss": 3.1233,
       "step": 2610
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005655399835119538,
-      "loss": 5.0761,
+      "learning_rate": 5.64921681780709e-05,
+      "loss": 2.9499,
       "step": 2611
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005653338829348722,
-      "loss": 4.8847,
+      "learning_rate": 5.6471558120362747e-05,
+      "loss": 2.7667,
       "step": 2612
     },
     {
       "epoch": 1.46,
-      "learning_rate": 0.0005651277823577907,
-      "loss": 5.0449,
+      "learning_rate": 5.6450948062654576e-05,
+      "loss": 2.5643,
       "step": 2613
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005649216817807091,
-      "loss": 5.0013,
+      "learning_rate": 5.643033800494641e-05,
+      "loss": 2.6271,
       "step": 2614
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005647155812036275,
-      "loss": 5.3472,
+      "learning_rate": 5.6409727947238256e-05,
+      "loss": 2.7431,
       "step": 2615
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005645094806265458,
-      "loss": 4.8797,
+      "learning_rate": 5.63891178895301e-05,
+      "loss": 2.3825,
       "step": 2616
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005643033800494642,
-      "loss": 4.7996,
+      "learning_rate": 5.6368507831821936e-05,
+      "loss": 2.1966,
       "step": 2617
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005640972794723826,
-      "loss": 4.5358,
+      "learning_rate": 5.6347897774113766e-05,
+      "loss": 2.2054,
       "step": 2618
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.000563891178895301,
-      "loss": 5.0358,
+      "learning_rate": 5.632728771640561e-05,
+      "loss": 2.7318,
       "step": 2619
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005636850783182194,
-      "loss": 4.6239,
+      "learning_rate": 5.6306677658697446e-05,
+      "loss": 2.2658,
       "step": 2620
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005634789777411377,
-      "loss": 4.9326,
+      "learning_rate": 5.628606760098929e-05,
+      "loss": 2.6563,
       "step": 2621
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005632728771640561,
-      "loss": 4.9566,
+      "learning_rate": 5.626545754328112e-05,
+      "loss": 2.2576,
       "step": 2622
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005630667765869745,
-      "loss": 4.8354,
+      "learning_rate": 5.624484748557296e-05,
+      "loss": 1.976,
       "step": 2623
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005628606760098929,
-      "loss": 4.81,
+      "learning_rate": 5.62242374278648e-05,
+      "loss": 2.0908,
       "step": 2624
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005626545754328112,
-      "loss": 4.9386,
+      "learning_rate": 5.620362737015664e-05,
+      "loss": 2.6435,
       "step": 2625
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005624484748557296,
-      "loss": 4.6555,
+      "learning_rate": 5.618301731244848e-05,
+      "loss": 1.8451,
       "step": 2626
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.000562242374278648,
-      "loss": 4.7439,
+      "learning_rate": 5.616240725474031e-05,
+      "loss": 2.3766,
       "step": 2627
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005620362737015664,
-      "loss": 5.6666,
+      "learning_rate": 5.614179719703215e-05,
+      "loss": 2.8715,
       "step": 2628
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005618301731244848,
-      "loss": 4.7649,
+      "learning_rate": 5.6121187139323996e-05,
+      "loss": 2.1292,
       "step": 2629
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005616240725474031,
-      "loss": 4.9463,
+      "learning_rate": 5.610057708161583e-05,
+      "loss": 2.1746,
       "step": 2630
     },
     {
       "epoch": 1.47,
-      "learning_rate": 0.0005614179719703215,
-      "loss": 4.8409,
+      "learning_rate": 5.607996702390766e-05,
+      "loss": 1.5679,
       "step": 2631
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005612118713932399,
-      "loss": 5.016,
+      "learning_rate": 5.6059356966199506e-05,
+      "loss": 1.6243,
       "step": 2632
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005610057708161584,
-      "loss": 4.6575,
+      "learning_rate": 5.603874690849135e-05,
+      "loss": 1.5346,
       "step": 2633
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005607996702390767,
-      "loss": 4.4669,
+      "learning_rate": 5.6018136850783186e-05,
+      "loss": 1.6169,
       "step": 2634
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005605935696619951,
-      "loss": 4.8681,
+      "learning_rate": 5.599752679307503e-05,
+      "loss": 2.6903,
       "step": 2635
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005603874690849135,
-      "loss": 4.9413,
+      "learning_rate": 5.597691673536686e-05,
+      "loss": 3.0026,
       "step": 2636
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005601813685078319,
-      "loss": 5.0674,
+      "learning_rate": 5.5956306677658695e-05,
+      "loss": 2.8018,
       "step": 2637
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005599752679307503,
-      "loss": 5.0034,
+      "learning_rate": 5.593569661995054e-05,
+      "loss": 3.4575,
       "step": 2638
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005597691673536686,
-      "loss": 4.8807,
+      "learning_rate": 5.591508656224238e-05,
+      "loss": 2.6909,
       "step": 2639
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.000559563066776587,
-      "loss": 5.1134,
+      "learning_rate": 5.589447650453421e-05,
+      "loss": 2.9334,
       "step": 2640
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005593569661995054,
-      "loss": 5.0428,
+      "learning_rate": 5.587386644682605e-05,
+      "loss": 3.2478,
       "step": 2641
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005591508656224238,
-      "loss": 4.8957,
+      "learning_rate": 5.585325638911789e-05,
+      "loss": 2.7456,
       "step": 2642
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005589447650453421,
-      "loss": 5.0808,
+      "learning_rate": 5.5832646331409735e-05,
+      "loss": 2.8118,
       "step": 2643
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005587386644682605,
-      "loss": 4.8374,
+      "learning_rate": 5.581203627370157e-05,
+      "loss": 2.5401,
       "step": 2644
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005585325638911789,
-      "loss": 5.1177,
+      "learning_rate": 5.57914262159934e-05,
+      "loss": 2.9089,
       "step": 2645
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005583264633140973,
-      "loss": 4.8372,
+      "learning_rate": 5.5770816158285245e-05,
+      "loss": 2.6198,
       "step": 2646
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005581203627370157,
-      "loss": 5.0784,
+      "learning_rate": 5.575020610057708e-05,
+      "loss": 3.2296,
       "step": 2647
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.000557914262159934,
-      "loss": 4.7165,
+      "learning_rate": 5.5729596042868925e-05,
+      "loss": 2.4391,
       "step": 2648
     },
     {
       "epoch": 1.48,
-      "learning_rate": 0.0005577081615828524,
-      "loss": 4.9549,
+      "learning_rate": 5.570898598516077e-05,
+      "loss": 2.9456,
       "step": 2649
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005575020610057708,
-      "loss": 4.7211,
+      "learning_rate": 5.56883759274526e-05,
+      "loss": 2.7575,
       "step": 2650
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005572959604286892,
-      "loss": 4.9956,
+      "learning_rate": 5.5667765869744435e-05,
+      "loss": 3.3713,
       "step": 2651
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005570898598516076,
-      "loss": 4.9969,
+      "learning_rate": 5.564715581203628e-05,
+      "loss": 3.0924,
       "step": 2652
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005568837592745259,
-      "loss": 5.108,
+      "learning_rate": 5.562654575432812e-05,
+      "loss": 2.8934,
       "step": 2653
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005566776586974444,
-      "loss": 5.428,
+      "learning_rate": 5.560593569661995e-05,
+      "loss": 3.3517,
       "step": 2654
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005564715581203628,
-      "loss": 5.005,
+      "learning_rate": 5.558532563891179e-05,
+      "loss": 3.1342,
       "step": 2655
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005562654575432812,
-      "loss": 5.0474,
+      "learning_rate": 5.556471558120363e-05,
+      "loss": 2.934,
       "step": 2656
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005560593569661995,
-      "loss": 4.9852,
+      "learning_rate": 5.554410552349547e-05,
+      "loss": 2.8191,
       "step": 2657
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005558532563891179,
-      "loss": 4.8614,
+      "learning_rate": 5.552349546578731e-05,
+      "loss": 2.4588,
       "step": 2658
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005556471558120363,
-      "loss": 4.9481,
+      "learning_rate": 5.550288540807914e-05,
+      "loss": 2.9618,
       "step": 2659
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005554410552349547,
-      "loss": 5.0162,
+      "learning_rate": 5.5482275350370985e-05,
+      "loss": 2.7618,
       "step": 2660
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005552349546578731,
-      "loss": 4.9602,
+      "learning_rate": 5.546166529266282e-05,
+      "loss": 2.7694,
       "step": 2661
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005550288540807914,
-      "loss": 4.9237,
+      "learning_rate": 5.5441055234954665e-05,
+      "loss": 3.0036,
       "step": 2662
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005548227535037098,
-      "loss": 5.4024,
+      "learning_rate": 5.5420445177246495e-05,
+      "loss": 3.0967,
       "step": 2663
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005546166529266282,
-      "loss": 4.7845,
+      "learning_rate": 5.539983511953834e-05,
+      "loss": 2.4334,
       "step": 2664
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005544105523495466,
-      "loss": 5.2859,
+      "learning_rate": 5.5379225061830175e-05,
+      "loss": 2.7005,
       "step": 2665
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005542044517724649,
-      "loss": 5.0567,
+      "learning_rate": 5.535861500412202e-05,
+      "loss": 3.0466,
       "step": 2666
     },
     {
       "epoch": 1.49,
-      "learning_rate": 0.0005539983511953833,
-      "loss": 5.3535,
+      "learning_rate": 5.5338004946413855e-05,
+      "loss": 3.1519,
       "step": 2667
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005537922506183017,
-      "loss": 4.9933,
+      "learning_rate": 5.5317394888705684e-05,
+      "loss": 2.5935,
       "step": 2668
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005535861500412201,
-      "loss": 5.0895,
+      "learning_rate": 5.529678483099753e-05,
+      "loss": 3.199,
       "step": 2669
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005533800494641385,
-      "loss": 5.3002,
+      "learning_rate": 5.527617477328937e-05,
+      "loss": 3.2114,
       "step": 2670
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005531739488870568,
-      "loss": 5.2384,
+      "learning_rate": 5.525556471558121e-05,
+      "loss": 3.1908,
       "step": 2671
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005529678483099752,
-      "loss": 5.1788,
+      "learning_rate": 5.523495465787304e-05,
+      "loss": 2.9136,
       "step": 2672
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005527617477328936,
-      "loss": 4.7467,
+      "learning_rate": 5.521434460016488e-05,
+      "loss": 2.0977,
       "step": 2673
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.000552555647155812,
-      "loss": 4.81,
+      "learning_rate": 5.519373454245672e-05,
+      "loss": 2.0569,
       "step": 2674
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005523495465787303,
-      "loss": 5.4366,
+      "learning_rate": 5.517312448474856e-05,
+      "loss": 2.8568,
       "step": 2675
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005521434460016488,
-      "loss": 5.5348,
+      "learning_rate": 5.5152514427040404e-05,
+      "loss": 2.7402,
       "step": 2676
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005519373454245672,
-      "loss": 4.8261,
+      "learning_rate": 5.5131904369332234e-05,
+      "loss": 1.9119,
       "step": 2677
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005517312448474856,
-      "loss": 5.1766,
+      "learning_rate": 5.511129431162407e-05,
+      "loss": 2.1823,
       "step": 2678
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.000551525144270404,
-      "loss": 4.9959,
+      "learning_rate": 5.5090684253915914e-05,
+      "loss": 2.3839,
       "step": 2679
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005513190436933223,
-      "loss": 4.6046,
+      "learning_rate": 5.507007419620776e-05,
+      "loss": 1.6375,
       "step": 2680
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005511129431162407,
-      "loss": 4.7935,
+      "learning_rate": 5.5049464138499594e-05,
+      "loss": 1.7163,
       "step": 2681
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005509068425391591,
-      "loss": 4.7336,
+      "learning_rate": 5.5028854080791424e-05,
+      "loss": 1.5136,
       "step": 2682
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005507007419620775,
-      "loss": 4.9861,
+      "learning_rate": 5.500824402308327e-05,
+      "loss": 1.8643,
       "step": 2683
     },
     {
       "epoch": 1.5,
-      "learning_rate": 0.0005504946413849959,
-      "loss": 4.015,
+      "learning_rate": 5.4987633965375104e-05,
+      "loss": 1.5154,
       "step": 2684
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005502885408079142,
-      "loss": 5.3946,
+      "learning_rate": 5.496702390766695e-05,
+      "loss": 3.9332,
       "step": 2685
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005500824402308326,
-      "loss": 4.828,
+      "learning_rate": 5.494641384995878e-05,
+      "loss": 2.6596,
       "step": 2686
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.000549876339653751,
-      "loss": 4.921,
+      "learning_rate": 5.492580379225062e-05,
+      "loss": 2.7027,
       "step": 2687
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005496702390766694,
-      "loss": 4.9531,
+      "learning_rate": 5.490519373454246e-05,
+      "loss": 3.0398,
       "step": 2688
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005494641384995877,
-      "loss": 4.8759,
+      "learning_rate": 5.48845836768343e-05,
+      "loss": 2.5309,
       "step": 2689
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005492580379225061,
-      "loss": 4.8267,
+      "learning_rate": 5.4863973619126144e-05,
+      "loss": 2.7535,
       "step": 2690
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005490519373454245,
-      "loss": 4.871,
+      "learning_rate": 5.4843363561417974e-05,
+      "loss": 2.6019,
       "step": 2691
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005488458367683429,
-      "loss": 5.0513,
+      "learning_rate": 5.482275350370981e-05,
+      "loss": 2.8653,
       "step": 2692
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005486397361912613,
-      "loss": 4.9383,
+      "learning_rate": 5.4802143446001654e-05,
+      "loss": 3.1344,
       "step": 2693
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005484336356141796,
-      "loss": 4.8699,
+      "learning_rate": 5.478153338829349e-05,
+      "loss": 2.562,
       "step": 2694
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.000548227535037098,
-      "loss": 5.255,
+      "learning_rate": 5.476092333058532e-05,
+      "loss": 3.0333,
       "step": 2695
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005480214344600166,
-      "loss": 4.8584,
+      "learning_rate": 5.4740313272877163e-05,
+      "loss": 2.5692,
       "step": 2696
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.000547815333882935,
-      "loss": 4.7653,
+      "learning_rate": 5.471970321516901e-05,
+      "loss": 2.3675,
       "step": 2697
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005476092333058533,
-      "loss": 4.6637,
+      "learning_rate": 5.4699093157460843e-05,
+      "loss": 2.4984,
       "step": 2698
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005474031327287717,
-      "loss": 5.0332,
+      "learning_rate": 5.467848309975269e-05,
+      "loss": 2.7051,
       "step": 2699
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005471970321516901,
-      "loss": 5.6386,
+      "learning_rate": 5.465787304204452e-05,
+      "loss": 3.2881,
       "step": 2700
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005469909315746085,
-      "loss": 4.836,
+      "learning_rate": 5.463726298433636e-05,
+      "loss": 2.6037,
       "step": 2701
     },
     {
       "epoch": 1.51,
-      "learning_rate": 0.0005467848309975269,
-      "loss": 4.5148,
+      "learning_rate": 5.46166529266282e-05,
+      "loss": 2.4974,
       "step": 2702
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005465787304204452,
-      "loss": 4.712,
+      "learning_rate": 5.459604286892004e-05,
+      "loss": 2.6682,
       "step": 2703
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005463726298433636,
-      "loss": 5.0119,
+      "learning_rate": 5.457543281121187e-05,
+      "loss": 2.3202,
       "step": 2704
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.000546166529266282,
-      "loss": 4.7069,
+      "learning_rate": 5.4554822753503706e-05,
+      "loss": 2.7947,
       "step": 2705
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005459604286892004,
-      "loss": 5.0023,
+      "learning_rate": 5.453421269579555e-05,
+      "loss": 2.6245,
       "step": 2706
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005457543281121187,
-      "loss": 4.8547,
+      "learning_rate": 5.451360263808739e-05,
+      "loss": 2.2548,
       "step": 2707
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005455482275350371,
-      "loss": 4.9683,
+      "learning_rate": 5.449299258037923e-05,
+      "loss": 2.6476,
       "step": 2708
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005453421269579555,
-      "loss": 4.9358,
+      "learning_rate": 5.447238252267106e-05,
+      "loss": 2.4118,
       "step": 2709
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005451360263808739,
-      "loss": 4.7607,
+      "learning_rate": 5.44517724649629e-05,
+      "loss": 2.6812,
       "step": 2710
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005449299258037923,
-      "loss": 4.6785,
+      "learning_rate": 5.4431162407254746e-05,
+      "loss": 2.1894,
       "step": 2711
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005447238252267106,
-      "loss": 4.9307,
+      "learning_rate": 5.441055234954658e-05,
+      "loss": 2.7612,
       "step": 2712
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.000544517724649629,
-      "loss": 5.212,
+      "learning_rate": 5.4389942291838426e-05,
+      "loss": 2.8598,
       "step": 2713
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005443116240725475,
-      "loss": 5.3907,
+      "learning_rate": 5.4369332234130256e-05,
+      "loss": 2.6132,
       "step": 2714
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005441055234954659,
-      "loss": 4.5373,
+      "learning_rate": 5.434872217642209e-05,
+      "loss": 2.1706,
       "step": 2715
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005438994229183843,
-      "loss": 5.2777,
+      "learning_rate": 5.4328112118713936e-05,
+      "loss": 2.8051,
       "step": 2716
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005436933223413026,
-      "loss": 5.0851,
+      "learning_rate": 5.430750206100578e-05,
+      "loss": 2.5089,
       "step": 2717
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.000543487221764221,
-      "loss": 4.7496,
+      "learning_rate": 5.428689200329761e-05,
+      "loss": 2.1592,
       "step": 2718
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005432811211871394,
-      "loss": 4.6809,
+      "learning_rate": 5.4266281945589446e-05,
+      "loss": 2.2947,
       "step": 2719
     },
     {
       "epoch": 1.52,
-      "learning_rate": 0.0005430750206100578,
-      "loss": 5.1033,
+      "learning_rate": 5.424567188788129e-05,
+      "loss": 3.144,
       "step": 2720
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005428689200329761,
-      "loss": 4.9552,
+      "learning_rate": 5.422506183017313e-05,
+      "loss": 2.852,
       "step": 2721
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005426628194558945,
-      "loss": 4.9282,
+      "learning_rate": 5.420445177246497e-05,
+      "loss": 2.0652,
       "step": 2722
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005424567188788129,
-      "loss": 4.9737,
+      "learning_rate": 5.41838417147568e-05,
+      "loss": 2.4554,
       "step": 2723
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005422506183017313,
-      "loss": 4.8318,
+      "learning_rate": 5.416323165704864e-05,
+      "loss": 2.0676,
       "step": 2724
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005420445177246497,
-      "loss": 5.1933,
+      "learning_rate": 5.414262159934048e-05,
+      "loss": 2.8723,
       "step": 2725
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.000541838417147568,
-      "loss": 4.9295,
+      "learning_rate": 5.412201154163232e-05,
+      "loss": 2.1425,
       "step": 2726
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005416323165704864,
-      "loss": 5.0298,
+      "learning_rate": 5.410140148392415e-05,
+      "loss": 1.9983,
       "step": 2727
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005414262159934048,
-      "loss": 4.85,
+      "learning_rate": 5.4080791426215996e-05,
+      "loss": 1.8426,
       "step": 2728
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005412201154163232,
-      "loss": 4.9146,
+      "learning_rate": 5.406018136850783e-05,
+      "loss": 2.1683,
       "step": 2729
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005410140148392415,
-      "loss": 5.6855,
+      "learning_rate": 5.4039571310799676e-05,
+      "loss": 3.2827,
       "step": 2730
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005408079142621599,
-      "loss": 4.5711,
+      "learning_rate": 5.401896125309151e-05,
+      "loss": 1.5183,
       "step": 2731
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005406018136850783,
-      "loss": 4.3342,
+      "learning_rate": 5.399835119538334e-05,
+      "loss": 1.5356,
       "step": 2732
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005403957131079967,
-      "loss": 5.4346,
+      "learning_rate": 5.3977741137675186e-05,
+      "loss": 1.8003,
       "step": 2733
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005401896125309152,
-      "loss": 4.5293,
+      "learning_rate": 5.395713107996703e-05,
+      "loss": 1.8805,
       "step": 2734
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005399835119538334,
-      "loss": 5.8772,
+      "learning_rate": 5.3936521022258866e-05,
+      "loss": 3.7202,
       "step": 2735
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005397774113767519,
-      "loss": 5.1576,
+      "learning_rate": 5.391591096455071e-05,
+      "loss": 3.1568,
       "step": 2736
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005395713107996703,
-      "loss": 5.1358,
+      "learning_rate": 5.389530090684254e-05,
+      "loss": 3.1879,
       "step": 2737
     },
     {
       "epoch": 1.53,
-      "learning_rate": 0.0005393652102225887,
-      "loss": 4.8923,
+      "learning_rate": 5.387469084913438e-05,
+      "loss": 2.6399,
       "step": 2738
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005391591096455071,
-      "loss": 4.9491,
+      "learning_rate": 5.385408079142622e-05,
+      "loss": 2.5355,
       "step": 2739
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005389530090684254,
-      "loss": 4.7672,
+      "learning_rate": 5.383347073371806e-05,
+      "loss": 2.5363,
       "step": 2740
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005387469084913438,
-      "loss": 5.1035,
+      "learning_rate": 5.381286067600989e-05,
+      "loss": 2.8563,
       "step": 2741
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005385408079142622,
-      "loss": 4.8735,
+      "learning_rate": 5.379225061830173e-05,
+      "loss": 2.5238,
       "step": 2742
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005383347073371806,
-      "loss": 4.9367,
+      "learning_rate": 5.377164056059357e-05,
+      "loss": 3.4647,
       "step": 2743
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005381286067600989,
-      "loss": 5.1649,
+      "learning_rate": 5.3751030502885415e-05,
+      "loss": 3.2502,
       "step": 2744
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005379225061830173,
-      "loss": 5.1688,
+      "learning_rate": 5.373042044517725e-05,
+      "loss": 2.793,
       "step": 2745
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005377164056059357,
-      "loss": 5.3375,
+      "learning_rate": 5.370981038746908e-05,
+      "loss": 3.1141,
       "step": 2746
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005375103050288541,
-      "loss": 5.0156,
+      "learning_rate": 5.3689200329760925e-05,
+      "loss": 3.024,
       "step": 2747
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005373042044517725,
-      "loss": 5.1242,
+      "learning_rate": 5.366859027205277e-05,
+      "loss": 3.3851,
       "step": 2748
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005370981038746908,
-      "loss": 5.7294,
+      "learning_rate": 5.3647980214344605e-05,
+      "loss": 3.4171,
       "step": 2749
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005368920032976092,
-      "loss": 5.8141,
+      "learning_rate": 5.3627370156636435e-05,
+      "loss": 3.444,
       "step": 2750
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005366859027205276,
-      "loss": 4.907,
+      "learning_rate": 5.360676009892828e-05,
+      "loss": 3.0291,
       "step": 2751
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.000536479802143446,
-      "loss": 4.7108,
+      "learning_rate": 5.3586150041220115e-05,
+      "loss": 2.9882,
       "step": 2752
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005362737015663643,
-      "loss": 4.6842,
+      "learning_rate": 5.356553998351196e-05,
+      "loss": 2.3329,
       "step": 2753
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005360676009892827,
-      "loss": 5.1712,
+      "learning_rate": 5.35449299258038e-05,
+      "loss": 2.9618,
       "step": 2754
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005358615004122011,
-      "loss": 5.1099,
+      "learning_rate": 5.352431986809563e-05,
+      "loss": 2.8274,
       "step": 2755
     },
     {
       "epoch": 1.54,
-      "learning_rate": 0.0005356553998351196,
-      "loss": 5.0424,
+      "learning_rate": 5.350370981038747e-05,
+      "loss": 2.9826,
       "step": 2756
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.000535449299258038,
-      "loss": 5.1221,
+      "learning_rate": 5.348309975267931e-05,
+      "loss": 2.797,
       "step": 2757
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005352431986809563,
-      "loss": 5.7373,
+      "learning_rate": 5.3462489694971155e-05,
+      "loss": 3.1836,
       "step": 2758
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005350370981038747,
-      "loss": 5.2545,
+      "learning_rate": 5.3441879637262985e-05,
+      "loss": 3.148,
       "step": 2759
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005348309975267931,
-      "loss": 5.0077,
+      "learning_rate": 5.342126957955482e-05,
+      "loss": 2.6634,
       "step": 2760
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005346248969497115,
-      "loss": 4.8473,
+      "learning_rate": 5.3400659521846665e-05,
+      "loss": 2.4629,
       "step": 2761
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005344187963726298,
-      "loss": 5.0185,
+      "learning_rate": 5.33800494641385e-05,
+      "loss": 2.5789,
       "step": 2762
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005342126957955482,
-      "loss": 4.9529,
+      "learning_rate": 5.3359439406430345e-05,
+      "loss": 3.3708,
       "step": 2763
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005340065952184666,
-      "loss": 5.1013,
+      "learning_rate": 5.3338829348722174e-05,
+      "loss": 2.9955,
       "step": 2764
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.000533800494641385,
-      "loss": 4.6892,
+      "learning_rate": 5.331821929101402e-05,
+      "loss": 2.7936,
       "step": 2765
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005335943940643034,
-      "loss": 4.5607,
+      "learning_rate": 5.3297609233305854e-05,
+      "loss": 2.6752,
       "step": 2766
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005333882934872217,
-      "loss": 4.9135,
+      "learning_rate": 5.32769991755977e-05,
+      "loss": 2.2901,
       "step": 2767
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005331821929101401,
-      "loss": 5.1268,
+      "learning_rate": 5.325638911788954e-05,
+      "loss": 2.8424,
       "step": 2768
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005329760923330585,
-      "loss": 4.4053,
+      "learning_rate": 5.323577906018137e-05,
+      "loss": 1.9238,
       "step": 2769
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005327699917559769,
-      "loss": 4.7188,
+      "learning_rate": 5.321516900247321e-05,
+      "loss": 2.3362,
       "step": 2770
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005325638911788953,
-      "loss": 4.8435,
+      "learning_rate": 5.319455894476505e-05,
+      "loss": 2.3358,
       "step": 2771
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005323577906018136,
-      "loss": 4.752,
+      "learning_rate": 5.317394888705689e-05,
+      "loss": 2.2449,
       "step": 2772
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.000532151690024732,
-      "loss": 4.8333,
+      "learning_rate": 5.315333882934872e-05,
+      "loss": 2.6245,
       "step": 2773
     },
     {
       "epoch": 1.55,
-      "learning_rate": 0.0005319455894476504,
-      "loss": 5.0294,
+      "learning_rate": 5.313272877164056e-05,
+      "loss": 2.8557,
       "step": 2774
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005317394888705688,
-      "loss": 4.5488,
+      "learning_rate": 5.3112118713932404e-05,
+      "loss": 1.8417,
       "step": 2775
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005315333882934871,
-      "loss": 4.9429,
+      "learning_rate": 5.309150865622424e-05,
+      "loss": 2.0875,
       "step": 2776
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005313272877164056,
-      "loss": 5.6044,
+      "learning_rate": 5.3070898598516084e-05,
+      "loss": 2.5101,
       "step": 2777
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.000531121187139324,
-      "loss": 4.9306,
+      "learning_rate": 5.3050288540807914e-05,
+      "loss": 1.9583,
       "step": 2778
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005309150865622424,
-      "loss": 4.769,
+      "learning_rate": 5.302967848309975e-05,
+      "loss": 1.6007,
       "step": 2779
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005307089859851608,
-      "loss": 4.7281,
+      "learning_rate": 5.3009068425391594e-05,
+      "loss": 1.5597,
       "step": 2780
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005305028854080791,
-      "loss": 4.1484,
+      "learning_rate": 5.298845836768344e-05,
+      "loss": 1.2247,
       "step": 2781
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005302967848309975,
-      "loss": 4.5629,
+      "learning_rate": 5.296784830997527e-05,
+      "loss": 1.5327,
       "step": 2782
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005300906842539159,
-      "loss": 4.6512,
+      "learning_rate": 5.2947238252267104e-05,
+      "loss": 1.4495,
       "step": 2783
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005298845836768343,
-      "loss": 4.5502,
+      "learning_rate": 5.292662819455895e-05,
+      "loss": 1.8075,
       "step": 2784
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005296784830997526,
-      "loss": 5.3159,
+      "learning_rate": 5.290601813685079e-05,
+      "loss": 3.0722,
       "step": 2785
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.000529472382522671,
-      "loss": 5.0415,
+      "learning_rate": 5.288540807914263e-05,
+      "loss": 3.3857,
       "step": 2786
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005292662819455894,
-      "loss": 5.0897,
+      "learning_rate": 5.286479802143446e-05,
+      "loss": 2.8129,
       "step": 2787
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005290601813685078,
-      "loss": 4.9424,
+      "learning_rate": 5.28441879637263e-05,
+      "loss": 3.3286,
       "step": 2788
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005288540807914263,
-      "loss": 4.6518,
+      "learning_rate": 5.282357790601814e-05,
+      "loss": 2.3456,
       "step": 2789
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.0005286479802143446,
-      "loss": 4.9314,
+      "learning_rate": 5.280296784830998e-05,
+      "loss": 3.0951,
       "step": 2790
     },
     {
       "epoch": 1.56,
-      "learning_rate": 0.000528441879637263,
-      "loss": 4.7148,
+      "learning_rate": 5.278235779060181e-05,
+      "loss": 3.203,
       "step": 2791
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005282357790601814,
-      "loss": 5.095,
+      "learning_rate": 5.2761747732893654e-05,
+      "loss": 3.1787,
       "step": 2792
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005280296784830998,
-      "loss": 4.9162,
+      "learning_rate": 5.274113767518549e-05,
+      "loss": 2.4024,
       "step": 2793
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005278235779060181,
-      "loss": 4.9853,
+      "learning_rate": 5.2720527617477334e-05,
+      "loss": 2.9316,
       "step": 2794
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005276174773289365,
-      "loss": 4.8752,
+      "learning_rate": 5.269991755976918e-05,
+      "loss": 2.4973,
       "step": 2795
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.000527411376751855,
-      "loss": 4.8529,
+      "learning_rate": 5.267930750206101e-05,
+      "loss": 3.0246,
       "step": 2796
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005272052761747734,
-      "loss": 4.8728,
+      "learning_rate": 5.2658697444352843e-05,
+      "loss": 2.6762,
       "step": 2797
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005269991755976918,
-      "loss": 5.0403,
+      "learning_rate": 5.263808738664469e-05,
+      "loss": 3.198,
       "step": 2798
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005267930750206101,
-      "loss": 4.9942,
+      "learning_rate": 5.261747732893652e-05,
+      "loss": 2.8032,
       "step": 2799
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005265869744435285,
-      "loss": 5.5591,
+      "learning_rate": 5.259686727122837e-05,
+      "loss": 3.2281,
       "step": 2800
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005263808738664469,
-      "loss": 4.7695,
+      "learning_rate": 5.2576257213520197e-05,
+      "loss": 2.354,
       "step": 2801
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005261747732893653,
-      "loss": 4.6509,
+      "learning_rate": 5.255564715581204e-05,
+      "loss": 2.8333,
       "step": 2802
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005259686727122837,
-      "loss": 4.8735,
+      "learning_rate": 5.2535037098103877e-05,
+      "loss": 2.4432,
       "step": 2803
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.000525762572135202,
-      "loss": 5.4841,
+      "learning_rate": 5.251442704039572e-05,
+      "loss": 3.0679,
       "step": 2804
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005255564715581204,
-      "loss": 4.9709,
+      "learning_rate": 5.249381698268755e-05,
+      "loss": 2.5776,
       "step": 2805
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005253503709810388,
-      "loss": 5.0337,
+      "learning_rate": 5.247320692497939e-05,
+      "loss": 2.7713,
       "step": 2806
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005251442704039572,
-      "loss": 5.4682,
+      "learning_rate": 5.245259686727123e-05,
+      "loss": 3.2463,
       "step": 2807
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005249381698268755,
-      "loss": 4.7683,
+      "learning_rate": 5.243198680956307e-05,
+      "loss": 2.367,
       "step": 2808
     },
     {
       "epoch": 1.57,
-      "learning_rate": 0.0005247320692497939,
-      "loss": 4.7322,
+      "learning_rate": 5.241137675185491e-05,
+      "loss": 2.1724,
       "step": 2809
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005245259686727123,
-      "loss": 4.4981,
+      "learning_rate": 5.239076669414674e-05,
+      "loss": 2.3364,
       "step": 2810
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005243198680956307,
-      "loss": 4.6683,
+      "learning_rate": 5.237015663643858e-05,
+      "loss": 2.865,
       "step": 2811
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005241137675185491,
-      "loss": 5.1613,
+      "learning_rate": 5.2349546578730426e-05,
+      "loss": 3.1586,
       "step": 2812
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005239076669414674,
-      "loss": 5.0958,
+      "learning_rate": 5.232893652102226e-05,
+      "loss": 2.881,
       "step": 2813
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005237015663643858,
-      "loss": 5.0264,
+      "learning_rate": 5.230832646331409e-05,
+      "loss": 2.6814,
       "step": 2814
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005234954657873042,
-      "loss": 4.8238,
+      "learning_rate": 5.2287716405605936e-05,
+      "loss": 2.3298,
       "step": 2815
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005232893652102227,
-      "loss": 4.9071,
+      "learning_rate": 5.226710634789778e-05,
+      "loss": 2.7382,
       "step": 2816
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.000523083264633141,
-      "loss": 4.7696,
+      "learning_rate": 5.2246496290189616e-05,
+      "loss": 2.7891,
       "step": 2817
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005228771640560594,
-      "loss": 4.6439,
+      "learning_rate": 5.222588623248146e-05,
+      "loss": 2.2381,
       "step": 2818
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005226710634789778,
-      "loss": 4.8542,
+      "learning_rate": 5.220527617477329e-05,
+      "loss": 2.3219,
       "step": 2819
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005224649629018962,
-      "loss": 4.6634,
+      "learning_rate": 5.2184666117065126e-05,
+      "loss": 2.5157,
       "step": 2820
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005222588623248146,
-      "loss": 4.6023,
+      "learning_rate": 5.216405605935697e-05,
+      "loss": 2.0395,
       "step": 2821
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005220527617477329,
-      "loss": 5.0339,
+      "learning_rate": 5.214344600164881e-05,
+      "loss": 2.7835,
       "step": 2822
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005218466611706513,
-      "loss": 4.8188,
+      "learning_rate": 5.212283594394065e-05,
+      "loss": 1.9736,
       "step": 2823
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005216405605935697,
-      "loss": 5.0043,
+      "learning_rate": 5.210222588623248e-05,
+      "loss": 2.4044,
       "step": 2824
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005214344600164881,
-      "loss": 5.1227,
+      "learning_rate": 5.208161582852432e-05,
+      "loss": 2.2286,
       "step": 2825
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005212283594394065,
-      "loss": 4.7445,
+      "learning_rate": 5.206100577081616e-05,
+      "loss": 1.9433,
       "step": 2826
     },
     {
       "epoch": 1.58,
-      "learning_rate": 0.0005210222588623248,
-      "loss": 4.9777,
+      "learning_rate": 5.2040395713108e-05,
+      "loss": 2.6586,
       "step": 2827
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005208161582852432,
-      "loss": 5.2002,
+      "learning_rate": 5.201978565539983e-05,
+      "loss": 1.9845,
       "step": 2828
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005206100577081616,
-      "loss": 4.9054,
+      "learning_rate": 5.1999175597691676e-05,
+      "loss": 2.4989,
       "step": 2829
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.00052040395713108,
-      "loss": 5.0842,
+      "learning_rate": 5.197856553998351e-05,
+      "loss": 1.8858,
       "step": 2830
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005201978565539983,
-      "loss": 4.2801,
+      "learning_rate": 5.1957955482275356e-05,
+      "loss": 1.4654,
       "step": 2831
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005199917559769167,
-      "loss": 4.5727,
+      "learning_rate": 5.19373454245672e-05,
+      "loss": 1.4997,
       "step": 2832
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005197856553998351,
-      "loss": 4.4683,
+      "learning_rate": 5.191673536685903e-05,
+      "loss": 1.6369,
       "step": 2833
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005195795548227535,
-      "loss": 4.9113,
+      "learning_rate": 5.1896125309150865e-05,
+      "loss": 1.6764,
       "step": 2834
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.000519373454245672,
-      "loss": 4.9849,
+      "learning_rate": 5.187551525144271e-05,
+      "loss": 2.7731,
       "step": 2835
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005191673536685902,
-      "loss": 4.8494,
+      "learning_rate": 5.1854905193734545e-05,
+      "loss": 3.0824,
       "step": 2836
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005189612530915087,
-      "loss": 4.6293,
+      "learning_rate": 5.1834295136026375e-05,
+      "loss": 3.1002,
       "step": 2837
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005187551525144271,
-      "loss": 4.832,
+      "learning_rate": 5.181368507831822e-05,
+      "loss": 2.5875,
       "step": 2838
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005185490519373455,
-      "loss": 4.8129,
+      "learning_rate": 5.179307502061006e-05,
+      "loss": 2.5151,
       "step": 2839
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005183429513602638,
-      "loss": 4.6738,
+      "learning_rate": 5.17724649629019e-05,
+      "loss": 2.4772,
       "step": 2840
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005181368507831822,
-      "loss": 5.3822,
+      "learning_rate": 5.175185490519374e-05,
+      "loss": 3.1165,
       "step": 2841
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005179307502061006,
-      "loss": 5.0753,
+      "learning_rate": 5.173124484748557e-05,
+      "loss": 2.9401,
       "step": 2842
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.000517724649629019,
-      "loss": 5.0748,
+      "learning_rate": 5.1710634789777415e-05,
+      "loss": 2.634,
       "step": 2843
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005175185490519374,
-      "loss": 5.2758,
+      "learning_rate": 5.169002473206925e-05,
+      "loss": 3.2103,
       "step": 2844
     },
     {
       "epoch": 1.59,
-      "learning_rate": 0.0005173124484748557,
-      "loss": 4.8947,
+      "learning_rate": 5.1669414674361095e-05,
+      "loss": 2.8278,
       "step": 2845
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005171063478977741,
-      "loss": 4.8727,
+      "learning_rate": 5.1648804616652925e-05,
+      "loss": 2.7229,
       "step": 2846
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005169002473206925,
-      "loss": 5.3107,
+      "learning_rate": 5.162819455894476e-05,
+      "loss": 3.0072,
       "step": 2847
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005166941467436109,
-      "loss": 5.0071,
+      "learning_rate": 5.1607584501236605e-05,
+      "loss": 2.6653,
       "step": 2848
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005164880461665292,
-      "loss": 4.8001,
+      "learning_rate": 5.158697444352845e-05,
+      "loss": 2.3119,
       "step": 2849
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005162819455894476,
-      "loss": 4.8467,
+      "learning_rate": 5.1566364385820285e-05,
+      "loss": 2.9657,
       "step": 2850
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.000516075845012366,
-      "loss": 4.9072,
+      "learning_rate": 5.1545754328112115e-05,
+      "loss": 2.5275,
       "step": 2851
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005158697444352844,
-      "loss": 4.8425,
+      "learning_rate": 5.152514427040396e-05,
+      "loss": 2.4599,
       "step": 2852
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005156636438582028,
-      "loss": 5.0271,
+      "learning_rate": 5.15045342126958e-05,
+      "loss": 3.0848,
       "step": 2853
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005154575432811211,
-      "loss": 4.6145,
+      "learning_rate": 5.148392415498764e-05,
+      "loss": 2.2661,
       "step": 2854
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005152514427040395,
-      "loss": 4.4781,
+      "learning_rate": 5.146331409727948e-05,
+      "loss": 2.4292,
       "step": 2855
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.000515045342126958,
-      "loss": 4.7813,
+      "learning_rate": 5.144270403957131e-05,
+      "loss": 2.7316,
       "step": 2856
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005148392415498764,
-      "loss": 4.8357,
+      "learning_rate": 5.142209398186315e-05,
+      "loss": 2.5683,
       "step": 2857
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005146331409727948,
-      "loss": 4.8528,
+      "learning_rate": 5.140148392415499e-05,
+      "loss": 2.3117,
       "step": 2858
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005144270403957131,
-      "loss": 5.5872,
+      "learning_rate": 5.1380873866446835e-05,
+      "loss": 3.2677,
       "step": 2859
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005142209398186315,
-      "loss": 4.8197,
+      "learning_rate": 5.1360263808738665e-05,
+      "loss": 2.8463,
       "step": 2860
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005140148392415499,
-      "loss": 5.1407,
+      "learning_rate": 5.13396537510305e-05,
+      "loss": 3.1606,
       "step": 2861
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005138087386644683,
-      "loss": 4.8898,
+      "learning_rate": 5.1319043693322345e-05,
+      "loss": 3.2506,
       "step": 2862
     },
     {
       "epoch": 1.6,
-      "learning_rate": 0.0005136026380873866,
-      "loss": 5.065,
+      "learning_rate": 5.129843363561419e-05,
+      "loss": 2.8785,
       "step": 2863
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.000513396537510305,
-      "loss": 4.9242,
+      "learning_rate": 5.1277823577906025e-05,
+      "loss": 2.6227,
       "step": 2864
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005131904369332234,
-      "loss": 4.5865,
+      "learning_rate": 5.1257213520197854e-05,
+      "loss": 2.1188,
       "step": 2865
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005129843363561418,
-      "loss": 4.9237,
+      "learning_rate": 5.12366034624897e-05,
+      "loss": 2.4772,
       "step": 2866
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005127782357790602,
-      "loss": 4.8197,
+      "learning_rate": 5.1215993404781534e-05,
+      "loss": 2.4023,
       "step": 2867
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005125721352019785,
-      "loss": 4.6222,
+      "learning_rate": 5.119538334707338e-05,
+      "loss": 2.1397,
       "step": 2868
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005123660346248969,
-      "loss": 5.5099,
+      "learning_rate": 5.117477328936521e-05,
+      "loss": 3.0184,
       "step": 2869
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005121599340478153,
-      "loss": 4.6911,
+      "learning_rate": 5.115416323165705e-05,
+      "loss": 2.1807,
       "step": 2870
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005119538334707337,
-      "loss": 4.7161,
+      "learning_rate": 5.113355317394889e-05,
+      "loss": 2.1853,
       "step": 2871
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.000511747732893652,
-      "loss": 4.8067,
+      "learning_rate": 5.111294311624073e-05,
+      "loss": 1.9552,
       "step": 2872
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005115416323165704,
-      "loss": 5.6408,
+      "learning_rate": 5.1092333058532574e-05,
+      "loss": 3.0378,
       "step": 2873
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005113355317394888,
-      "loss": 4.9481,
+      "learning_rate": 5.10717230008244e-05,
+      "loss": 2.5366,
       "step": 2874
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005111294311624072,
-      "loss": 5.2645,
+      "learning_rate": 5.105111294311624e-05,
+      "loss": 3.0736,
       "step": 2875
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005109233305853256,
-      "loss": 4.8979,
+      "learning_rate": 5.1030502885408084e-05,
+      "loss": 2.9876,
       "step": 2876
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.000510717230008244,
-      "loss": 4.8281,
+      "learning_rate": 5.100989282769992e-05,
+      "loss": 2.2132,
       "step": 2877
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005105111294311624,
-      "loss": 5.0836,
+      "learning_rate": 5.098928276999175e-05,
+      "loss": 2.8034,
       "step": 2878
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005103050288540808,
-      "loss": 4.7873,
+      "learning_rate": 5.0968672712283594e-05,
+      "loss": 1.8175,
       "step": 2879
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005100989282769992,
-      "loss": 4.5298,
+      "learning_rate": 5.094806265457544e-05,
+      "loss": 2.0383,
       "step": 2880
     },
     {
       "epoch": 1.61,
-      "learning_rate": 0.0005098928276999175,
-      "loss": 4.7793,
+      "learning_rate": 5.0927452596867274e-05,
+      "loss": 1.5059,
       "step": 2881
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005096867271228359,
-      "loss": 5.8203,
+      "learning_rate": 5.090684253915912e-05,
+      "loss": 2.1021,
       "step": 2882
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005094806265457544,
-      "loss": 4.9854,
+      "learning_rate": 5.088623248145095e-05,
+      "loss": 1.5639,
       "step": 2883
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005092745259686728,
-      "loss": 4.6577,
+      "learning_rate": 5.0865622423742784e-05,
+      "loss": 1.5593,
       "step": 2884
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005090684253915912,
-      "loss": 5.2622,
+      "learning_rate": 5.084501236603463e-05,
+      "loss": 3.1255,
       "step": 2885
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005088623248145095,
-      "loss": 4.8151,
+      "learning_rate": 5.082440230832647e-05,
+      "loss": 2.572,
       "step": 2886
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005086562242374279,
-      "loss": 4.9556,
+      "learning_rate": 5.080379225061831e-05,
+      "loss": 2.9392,
       "step": 2887
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005084501236603463,
-      "loss": 4.926,
+      "learning_rate": 5.078318219291014e-05,
+      "loss": 2.8495,
       "step": 2888
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005082440230832647,
-      "loss": 4.7872,
+      "learning_rate": 5.076257213520198e-05,
+      "loss": 2.3776,
       "step": 2889
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005080379225061831,
-      "loss": 4.8647,
+      "learning_rate": 5.0741962077493824e-05,
+      "loss": 2.6531,
       "step": 2890
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005078318219291014,
-      "loss": 4.793,
+      "learning_rate": 5.072135201978566e-05,
+      "loss": 2.6822,
       "step": 2891
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005076257213520198,
-      "loss": 4.9721,
+      "learning_rate": 5.070074196207749e-05,
+      "loss": 2.9599,
       "step": 2892
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005074196207749382,
-      "loss": 5.0686,
+      "learning_rate": 5.0680131904369333e-05,
+      "loss": 2.8679,
       "step": 2893
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005072135201978566,
-      "loss": 4.7767,
+      "learning_rate": 5.065952184666117e-05,
+      "loss": 2.406,
       "step": 2894
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005070074196207749,
-      "loss": 4.9068,
+      "learning_rate": 5.0638911788953013e-05,
+      "loss": 2.9524,
       "step": 2895
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005068013190436933,
-      "loss": 4.7194,
+      "learning_rate": 5.061830173124486e-05,
+      "loss": 2.6264,
       "step": 2896
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005065952184666118,
-      "loss": 4.9561,
+      "learning_rate": 5.059769167353669e-05,
+      "loss": 2.5241,
       "step": 2897
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005063891178895302,
-      "loss": 4.8854,
+      "learning_rate": 5.057708161582852e-05,
+      "loss": 2.6021,
       "step": 2898
     },
     {
       "epoch": 1.62,
-      "learning_rate": 0.0005061830173124486,
-      "loss": 4.669,
+      "learning_rate": 5.055647155812037e-05,
+      "loss": 2.2926,
       "step": 2899
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005059769167353669,
-      "loss": 4.8871,
+      "learning_rate": 5.053586150041221e-05,
+      "loss": 2.9221,
       "step": 2900
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005057708161582853,
-      "loss": 4.8804,
+      "learning_rate": 5.051525144270404e-05,
+      "loss": 2.9595,
       "step": 2901
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005055647155812037,
-      "loss": 5.0735,
+      "learning_rate": 5.0494641384995876e-05,
+      "loss": 2.5774,
       "step": 2902
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005053586150041221,
-      "loss": 4.9173,
+      "learning_rate": 5.047403132728772e-05,
+      "loss": 2.3763,
       "step": 2903
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005051525144270404,
-      "loss": 4.9308,
+      "learning_rate": 5.0453421269579556e-05,
+      "loss": 2.8637,
       "step": 2904
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005049464138499588,
-      "loss": 4.956,
+      "learning_rate": 5.04328112118714e-05,
+      "loss": 2.527,
       "step": 2905
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005047403132728772,
-      "loss": 4.9552,
+      "learning_rate": 5.041220115416323e-05,
+      "loss": 2.8287,
       "step": 2906
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005045342126957956,
-      "loss": 5.2224,
+      "learning_rate": 5.039159109645507e-05,
+      "loss": 2.6758,
       "step": 2907
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.000504328112118714,
-      "loss": 5.0884,
+      "learning_rate": 5.037098103874691e-05,
+      "loss": 2.3843,
       "step": 2908
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005041220115416323,
-      "loss": 4.9645,
+      "learning_rate": 5.035037098103875e-05,
+      "loss": 2.7233,
       "step": 2909
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005039159109645507,
-      "loss": 5.2955,
+      "learning_rate": 5.032976092333058e-05,
+      "loss": 2.9163,
       "step": 2910
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005037098103874691,
-      "loss": 4.808,
+      "learning_rate": 5.0309150865622426e-05,
+      "loss": 2.1432,
       "step": 2911
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005035037098103875,
-      "loss": 4.5859,
+      "learning_rate": 5.028854080791426e-05,
+      "loss": 2.302,
       "step": 2912
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005032976092333058,
-      "loss": 5.5587,
+      "learning_rate": 5.0267930750206106e-05,
+      "loss": 3.0538,
       "step": 2913
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005030915086562242,
-      "loss": 4.9497,
+      "learning_rate": 5.024732069249794e-05,
+      "loss": 2.947,
       "step": 2914
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.0005028854080791426,
-      "loss": 4.7373,
+      "learning_rate": 5.022671063478977e-05,
+      "loss": 2.4969,
       "step": 2915
     },
     {
       "epoch": 1.63,
-      "learning_rate": 0.000502679307502061,
-      "loss": 4.4535,
+      "learning_rate": 5.0206100577081616e-05,
+      "loss": 2.1434,
       "step": 2916
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005024732069249795,
-      "loss": 4.6037,
+      "learning_rate": 5.018549051937346e-05,
+      "loss": 2.4951,
       "step": 2917
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005022671063478978,
-      "loss": 5.2585,
+      "learning_rate": 5.0164880461665296e-05,
+      "loss": 2.5305,
       "step": 2918
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005020610057708162,
-      "loss": 4.6104,
+      "learning_rate": 5.014427040395714e-05,
+      "loss": 2.2184,
       "step": 2919
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005018549051937346,
-      "loss": 4.8446,
+      "learning_rate": 5.012366034624897e-05,
+      "loss": 2.2619,
       "step": 2920
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.000501648804616653,
-      "loss": 4.651,
+      "learning_rate": 5.010305028854081e-05,
+      "loss": 2.2248,
       "step": 2921
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005014427040395714,
-      "loss": 4.4523,
+      "learning_rate": 5.008244023083265e-05,
+      "loss": 1.9354,
       "step": 2922
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005012366034624897,
-      "loss": 5.0576,
+      "learning_rate": 5.006183017312449e-05,
+      "loss": 3.1649,
       "step": 2923
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005010305028854081,
-      "loss": 4.9897,
+      "learning_rate": 5.004122011541632e-05,
+      "loss": 2.1646,
       "step": 2924
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005008244023083265,
-      "loss": 4.9498,
+      "learning_rate": 5.002061005770816e-05,
+      "loss": 2.1781,
       "step": 2925
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005006183017312449,
-      "loss": 5.13,
+      "learning_rate": 5e-05,
+      "loss": 2.8066,
       "step": 2926
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005004122011541632,
-      "loss": 4.7719,
+      "learning_rate": 4.9979389942291846e-05,
+      "loss": 1.719,
       "step": 2927
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005002061005770816,
-      "loss": 4.9221,
+      "learning_rate": 4.9958779884583676e-05,
+      "loss": 1.9223,
       "step": 2928
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0005,
-      "loss": 5.1253,
+      "learning_rate": 4.993816982687552e-05,
+      "loss": 2.2745,
       "step": 2929
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0004997938994229184,
-      "loss": 5.1202,
+      "learning_rate": 4.9917559769167356e-05,
+      "loss": 1.7799,
       "step": 2930
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0004995877988458368,
-      "loss": 4.5824,
+      "learning_rate": 4.989694971145919e-05,
+      "loss": 1.9457,
       "step": 2931
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0004993816982687552,
-      "loss": 4.8614,
+      "learning_rate": 4.987633965375103e-05,
+      "loss": 1.6336,
       "step": 2932
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0004991755976916735,
-      "loss": 5.9299,
+      "learning_rate": 4.985572959604287e-05,
+      "loss": 2.0525,
       "step": 2933
     },
     {
       "epoch": 1.64,
-      "learning_rate": 0.0004989694971145919,
-      "loss": 3.9866,
+      "learning_rate": 4.9835119538334716e-05,
+      "loss": 1.4917,
       "step": 2934
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004987633965375103,
-      "loss": 4.7628,
+      "learning_rate": 4.9814509480626545e-05,
+      "loss": 3.0299,
       "step": 2935
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004985572959604287,
-      "loss": 5.5465,
+      "learning_rate": 4.979389942291839e-05,
+      "loss": 2.8622,
       "step": 2936
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004983511953833472,
-      "loss": 4.982,
+      "learning_rate": 4.9773289365210225e-05,
+      "loss": 2.7851,
       "step": 2937
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004981450948062655,
-      "loss": 4.689,
+      "learning_rate": 4.975267930750206e-05,
+      "loss": 2.9581,
       "step": 2938
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004979389942291839,
-      "loss": 4.8461,
+      "learning_rate": 4.97320692497939e-05,
+      "loss": 3.1783,
       "step": 2939
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004977328936521023,
-      "loss": 4.9766,
+      "learning_rate": 4.971145919208574e-05,
+      "loss": 3.4938,
       "step": 2940
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004975267930750207,
-      "loss": 4.7843,
+      "learning_rate": 4.969084913437758e-05,
+      "loss": 2.7015,
       "step": 2941
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.000497320692497939,
-      "loss": 4.7863,
+      "learning_rate": 4.9670239076669415e-05,
+      "loss": 2.5868,
       "step": 2942
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004971145919208574,
-      "loss": 4.9879,
+      "learning_rate": 4.964962901896126e-05,
+      "loss": 3.0534,
       "step": 2943
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004969084913437758,
-      "loss": 5.1038,
+      "learning_rate": 4.9629018961253095e-05,
+      "loss": 2.9347,
       "step": 2944
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004967023907666942,
-      "loss": 5.4049,
+      "learning_rate": 4.960840890354493e-05,
+      "loss": 3.2423,
       "step": 2945
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004964962901896126,
-      "loss": 4.7133,
+      "learning_rate": 4.958779884583677e-05,
+      "loss": 2.4394,
       "step": 2946
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004962901896125309,
-      "loss": 4.8305,
+      "learning_rate": 4.956718878812861e-05,
+      "loss": 2.5427,
       "step": 2947
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004960840890354493,
-      "loss": 4.7094,
+      "learning_rate": 4.954657873042045e-05,
+      "loss": 2.4464,
       "step": 2948
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004958779884583677,
-      "loss": 4.923,
+      "learning_rate": 4.9525968672712285e-05,
+      "loss": 2.9952,
       "step": 2949
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004956718878812861,
-      "loss": 4.5092,
+      "learning_rate": 4.950535861500413e-05,
+      "loss": 2.4973,
       "step": 2950
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004954657873042044,
-      "loss": 4.8073,
+      "learning_rate": 4.9484748557295965e-05,
+      "loss": 2.4622,
       "step": 2951
     },
     {
       "epoch": 1.65,
-      "learning_rate": 0.0004952596867271228,
-      "loss": 4.6826,
+      "learning_rate": 4.94641384995878e-05,
+      "loss": 2.3206,
       "step": 2952
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004950535861500412,
-      "loss": 4.526,
+      "learning_rate": 4.944352844187964e-05,
+      "loss": 2.1742,
       "step": 2953
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004948474855729596,
-      "loss": 5.1821,
+      "learning_rate": 4.942291838417148e-05,
+      "loss": 2.7398,
       "step": 2954
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.000494641384995878,
-      "loss": 4.5528,
+      "learning_rate": 4.940230832646331e-05,
+      "loss": 2.3414,
       "step": 2955
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004944352844187963,
-      "loss": 5.471,
+      "learning_rate": 4.9381698268755155e-05,
+      "loss": 3.1018,
       "step": 2956
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004942291838417147,
-      "loss": 4.4559,
+      "learning_rate": 4.936108821104699e-05,
+      "loss": 2.205,
       "step": 2957
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004940230832646332,
-      "loss": 4.9998,
+      "learning_rate": 4.9340478153338835e-05,
+      "loss": 2.9122,
       "step": 2958
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004938169826875516,
-      "loss": 5.0525,
+      "learning_rate": 4.931986809563067e-05,
+      "loss": 2.7623,
       "step": 2959
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004936108821104699,
-      "loss": 4.7632,
+      "learning_rate": 4.929925803792251e-05,
+      "loss": 2.076,
       "step": 2960
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004934047815333883,
-      "loss": 4.5558,
+      "learning_rate": 4.927864798021435e-05,
+      "loss": 3.1052,
       "step": 2961
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004931986809563067,
-      "loss": 4.5775,
+      "learning_rate": 4.925803792250618e-05,
+      "loss": 2.5827,
       "step": 2962
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004929925803792251,
-      "loss": 5.1002,
+      "learning_rate": 4.9237427864798024e-05,
+      "loss": 2.7947,
       "step": 2963
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004927864798021435,
-      "loss": 5.4826,
+      "learning_rate": 4.921681780708986e-05,
+      "loss": 3.2296,
       "step": 2964
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004925803792250618,
-      "loss": 4.6726,
+      "learning_rate": 4.91962077493817e-05,
+      "loss": 2.459,
       "step": 2965
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004923742786479802,
-      "loss": 4.5173,
+      "learning_rate": 4.917559769167354e-05,
+      "loss": 2.3332,
       "step": 2966
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004921681780708986,
-      "loss": 4.703,
+      "learning_rate": 4.915498763396538e-05,
+      "loss": 1.97,
       "step": 2967
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.000491962077493817,
-      "loss": 4.6806,
+      "learning_rate": 4.913437757625722e-05,
+      "loss": 2.2898,
       "step": 2968
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004917559769167354,
-      "loss": 4.8013,
+      "learning_rate": 4.911376751854905e-05,
+      "loss": 2.5292,
       "step": 2969
     },
     {
       "epoch": 1.66,
-      "learning_rate": 0.0004915498763396537,
-      "loss": 4.8185,
+      "learning_rate": 4.9093157460840894e-05,
+      "loss": 2.6848,
       "step": 2970
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004913437757625721,
-      "loss": 4.9222,
+      "learning_rate": 4.907254740313273e-05,
+      "loss": 2.3614,
       "step": 2971
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004911376751854905,
-      "loss": 5.0195,
+      "learning_rate": 4.905193734542457e-05,
+      "loss": 2.3842,
       "step": 2972
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004909315746084089,
-      "loss": 5.2666,
+      "learning_rate": 4.9031327287716404e-05,
+      "loss": 3.2936,
       "step": 2973
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004907254740313272,
-      "loss": 5.074,
+      "learning_rate": 4.901071723000825e-05,
+      "loss": 2.6843,
       "step": 2974
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004905193734542456,
-      "loss": 4.854,
+      "learning_rate": 4.8990107172300084e-05,
+      "loss": 2.9627,
       "step": 2975
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.000490313272877164,
-      "loss": 5.1185,
+      "learning_rate": 4.896949711459192e-05,
+      "loss": 2.9565,
       "step": 2976
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004901071723000824,
-      "loss": 4.8451,
+      "learning_rate": 4.8948887056883764e-05,
+      "loss": 2.0827,
       "step": 2977
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004899010717230009,
-      "loss": 5.6358,
+      "learning_rate": 4.89282769991756e-05,
+      "loss": 2.6365,
       "step": 2978
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004896949711459193,
-      "loss": 4.8059,
+      "learning_rate": 4.890766694146744e-05,
+      "loss": 1.9217,
       "step": 2979
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004894888705688377,
-      "loss": 4.728,
+      "learning_rate": 4.8887056883759274e-05,
+      "loss": 1.7172,
       "step": 2980
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.000489282769991756,
-      "loss": 4.6566,
+      "learning_rate": 4.886644682605112e-05,
+      "loss": 1.6104,
       "step": 2981
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004890766694146744,
-      "loss": 4.991,
+      "learning_rate": 4.8845836768342954e-05,
+      "loss": 1.5194,
       "step": 2982
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004888705688375928,
-      "loss": 5.1569,
+      "learning_rate": 4.882522671063479e-05,
+      "loss": 1.5773,
       "step": 2983
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004886644682605112,
-      "loss": 5.0346,
+      "learning_rate": 4.8804616652926634e-05,
+      "loss": 1.8241,
       "step": 2984
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004884583676834296,
-      "loss": 5.0383,
+      "learning_rate": 4.878400659521847e-05,
+      "loss": 3.0166,
       "step": 2985
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004882522671063479,
-      "loss": 5.0121,
+      "learning_rate": 4.876339653751031e-05,
+      "loss": 2.9331,
       "step": 2986
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.0004880461665292663,
-      "loss": 4.7907,
+      "learning_rate": 4.8742786479802144e-05,
+      "loss": 2.4597,
       "step": 2987
     },
     {
       "epoch": 1.67,
-      "learning_rate": 0.00048784006595218465,
-      "loss": 4.6709,
+      "learning_rate": 4.872217642209399e-05,
+      "loss": 2.3679,
       "step": 2988
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048763396537510306,
-      "loss": 5.4123,
+      "learning_rate": 4.870156636438582e-05,
+      "loss": 2.9411,
       "step": 2989
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004874278647980214,
-      "loss": 5.0189,
+      "learning_rate": 4.868095630667766e-05,
+      "loss": 2.6619,
       "step": 2990
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004872217642209398,
-      "loss": 5.2049,
+      "learning_rate": 4.8660346248969504e-05,
+      "loss": 2.9871,
       "step": 2991
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048701566364385817,
-      "loss": 4.7734,
+      "learning_rate": 4.863973619126134e-05,
+      "loss": 2.6421,
       "step": 2992
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004868095630667766,
-      "loss": 5.6477,
+      "learning_rate": 4.861912613355318e-05,
+      "loss": 3.4024,
       "step": 2993
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.000486603462489695,
-      "loss": 4.7488,
+      "learning_rate": 4.8598516075845013e-05,
+      "loss": 2.3474,
       "step": 2994
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048639736191261333,
-      "loss": 4.8331,
+      "learning_rate": 4.857790601813686e-05,
+      "loss": 2.3881,
       "step": 2995
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048619126133553174,
-      "loss": 4.7743,
+      "learning_rate": 4.8557295960428687e-05,
+      "loss": 2.4117,
       "step": 2996
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004859851607584501,
-      "loss": 5.0226,
+      "learning_rate": 4.853668590272053e-05,
+      "loss": 2.48,
       "step": 2997
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004857790601813685,
-      "loss": 4.6922,
+      "learning_rate": 4.851607584501237e-05,
+      "loss": 2.7303,
       "step": 2998
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048557295960428685,
-      "loss": 4.7133,
+      "learning_rate": 4.84954657873042e-05,
+      "loss": 2.4258,
       "step": 2999
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004853668590272053,
-      "loss": 4.7653,
+      "learning_rate": 4.8474855729596047e-05,
+      "loss": 2.9677,
       "step": 3000
     },
     {
       "epoch": 1.68,
-      "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
+      "eval_loss": 6.28408145904541,
+      "eval_runtime": 1248.6504,
+      "eval_samples_per_second": 2.116,
+      "eval_steps_per_second": 0.265,
+      "eval_wer": 1.673145577151924,
+      "step": 3000
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004851607584501237,
-      "loss": 4.7718,
+      "learning_rate": 4.845424567188788e-05,
+      "loss": 2.4348,
       "step": 3001
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048495465787304207,
-      "loss": 4.922,
+      "learning_rate": 4.8433635614179727e-05,
+      "loss": 3.1212,
       "step": 3002
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004847485572959605,
-      "loss": 4.7143,
+      "learning_rate": 4.8413025556471556e-05,
+      "loss": 2.3002,
       "step": 3003
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048454245671887883,
-      "loss": 4.6927,
+      "learning_rate": 4.83924154987634e-05,
+      "loss": 2.2918,
       "step": 3004
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.00048433635614179724,
-      "loss": 4.6121,
+      "learning_rate": 4.8371805441055236e-05,
+      "loss": 2.8036,
       "step": 3005
     },
     {
       "epoch": 1.68,
-      "learning_rate": 0.0004841302555647156,
-      "loss": 5.0751,
+      "learning_rate": 4.835119538334707e-05,
+      "loss": 3.4228,
       "step": 3006
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.000483924154987634,
-      "loss": 5.1179,
+      "learning_rate": 4.8330585325638916e-05,
+      "loss": 2.9835,
       "step": 3007
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048371805441055235,
-      "loss": 4.7184,
+      "learning_rate": 4.830997526793075e-05,
+      "loss": 2.5344,
       "step": 3008
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048351195383347076,
-      "loss": 4.6881,
+      "learning_rate": 4.828936521022259e-05,
+      "loss": 2.6888,
       "step": 3009
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048330585325638916,
-      "loss": 5.1034,
+      "learning_rate": 4.8268755152514426e-05,
+      "loss": 3.0755,
       "step": 3010
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004830997526793075,
-      "loss": 4.9019,
+      "learning_rate": 4.824814509480627e-05,
+      "loss": 2.4876,
       "step": 3011
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004828936521022259,
-      "loss": 4.4907,
+      "learning_rate": 4.8227535037098106e-05,
+      "loss": 2.4951,
       "step": 3012
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004826875515251443,
-      "loss": 5.4752,
+      "learning_rate": 4.820692497938994e-05,
+      "loss": 3.4925,
       "step": 3013
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004824814509480627,
-      "loss": 4.9775,
+      "learning_rate": 4.8186314921681786e-05,
+      "loss": 2.2179,
       "step": 3014
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048227535037098103,
-      "loss": 4.58,
+      "learning_rate": 4.816570486397362e-05,
+      "loss": 1.9884,
       "step": 3015
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048206924979389944,
-      "loss": 4.8092,
+      "learning_rate": 4.814509480626546e-05,
+      "loss": 2.7093,
       "step": 3016
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048186314921681785,
-      "loss": 6.0983,
+      "learning_rate": 4.8124484748557296e-05,
+      "loss": 3.4762,
       "step": 3017
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004816570486397362,
-      "loss": 4.6628,
+      "learning_rate": 4.810387469084914e-05,
+      "loss": 2.4305,
       "step": 3018
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004814509480626546,
-      "loss": 4.9202,
+      "learning_rate": 4.8083264633140976e-05,
+      "loss": 2.5847,
       "step": 3019
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048124484748557296,
-      "loss": 4.6934,
+      "learning_rate": 4.806265457543281e-05,
+      "loss": 2.1768,
       "step": 3020
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.00048103874690849137,
-      "loss": 4.7527,
+      "learning_rate": 4.804204451772465e-05,
+      "loss": 2.2175,
       "step": 3021
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004808326463314097,
-      "loss": 4.7332,
+      "learning_rate": 4.802143446001649e-05,
+      "loss": 2.7998,
       "step": 3022
     },
     {
       "epoch": 1.69,
-      "learning_rate": 0.0004806265457543281,
-      "loss": 5.1831,
+      "learning_rate": 4.800082440230833e-05,
+      "loss": 2.5893,
       "step": 3023
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0004804204451772465,
-      "loss": 4.6306,
+      "learning_rate": 4.7980214344600166e-05,
+      "loss": 2.7552,
       "step": 3024
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0004802143446001649,
-      "loss": 4.9321,
+      "learning_rate": 4.795960428689201e-05,
+      "loss": 2.2703,
       "step": 3025
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0004800082440230833,
-      "loss": 4.7831,
+      "learning_rate": 4.7938994229183846e-05,
+      "loss": 1.9365,
       "step": 3026
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047980214344600164,
-      "loss": 4.6309,
+      "learning_rate": 4.791838417147568e-05,
+      "loss": 1.8788,
       "step": 3027
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047959604286892005,
-      "loss": 5.9696,
+      "learning_rate": 4.789777411376752e-05,
+      "loss": 3.0648,
       "step": 3028
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0004793899422918384,
-      "loss": 4.7136,
+      "learning_rate": 4.787716405605936e-05,
+      "loss": 1.6102,
       "step": 3029
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0004791838417147568,
-      "loss": 5.6978,
+      "learning_rate": 4.78565539983512e-05,
+      "loss": 2.8617,
       "step": 3030
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047897774113767516,
-      "loss": 4.7382,
+      "learning_rate": 4.7835943940643035e-05,
+      "loss": 1.6981,
       "step": 3031
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047877164056059357,
-      "loss": 4.8475,
+      "learning_rate": 4.781533388293488e-05,
+      "loss": 1.7365,
       "step": 3032
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.000478565539983512,
-      "loss": 4.7631,
+      "learning_rate": 4.779472382522671e-05,
+      "loss": 1.3406,
       "step": 3033
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047835943940643033,
-      "loss": 4.6395,
+      "learning_rate": 4.777411376751855e-05,
+      "loss": 1.5162,
       "step": 3034
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047815333882934873,
-      "loss": 4.832,
+      "learning_rate": 4.775350370981039e-05,
+      "loss": 3.1128,
       "step": 3035
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0004779472382522671,
-      "loss": 4.709,
+      "learning_rate": 4.7732893652102225e-05,
+      "loss": 3.0556,
       "step": 3036
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.0004777411376751855,
-      "loss": 4.8878,
+      "learning_rate": 4.771228359439407e-05,
+      "loss": 3.0932,
       "step": 3037
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047753503709810385,
-      "loss": 4.8311,
+      "learning_rate": 4.7691673536685905e-05,
+      "loss": 3.1211,
       "step": 3038
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047732893652102225,
-      "loss": 4.8308,
+      "learning_rate": 4.767106347897775e-05,
+      "loss": 2.5564,
       "step": 3039
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.00047712283594394066,
-      "loss": 5.2709,
+      "learning_rate": 4.765045342126958e-05,
+      "loss": 3.1324,
       "step": 3040
     },
     {
       "epoch": 1.7,
-      "learning_rate": 0.000476916735366859,
-      "loss": 4.8662,
+      "learning_rate": 4.762984336356142e-05,
+      "loss": 2.5781,
       "step": 3041
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0004767106347897774,
-      "loss": 4.7272,
+      "learning_rate": 4.760923330585326e-05,
+      "loss": 2.4702,
       "step": 3042
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047650453421269577,
-      "loss": 4.7876,
+      "learning_rate": 4.7588623248145095e-05,
+      "loss": 2.5851,
       "step": 3043
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0004762984336356142,
-      "loss": 4.6993,
+      "learning_rate": 4.756801319043693e-05,
+      "loss": 2.3691,
       "step": 3044
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047609233305853253,
-      "loss": 4.7552,
+      "learning_rate": 4.7547403132728775e-05,
+      "loss": 2.4905,
       "step": 3045
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.000475886232481451,
-      "loss": 4.7379,
+      "learning_rate": 4.752679307502061e-05,
+      "loss": 2.3401,
       "step": 3046
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047568013190436934,
-      "loss": 4.6416,
+      "learning_rate": 4.750618301731245e-05,
+      "loss": 2.2958,
       "step": 3047
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047547403132728775,
-      "loss": 5.1788,
+      "learning_rate": 4.748557295960429e-05,
+      "loss": 3.3119,
       "step": 3048
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047526793075020616,
-      "loss": 4.6729,
+      "learning_rate": 4.746496290189613e-05,
+      "loss": 2.5694,
       "step": 3049
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0004750618301731245,
-      "loss": 4.5329,
+      "learning_rate": 4.7444352844187965e-05,
+      "loss": 2.3703,
       "step": 3050
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0004748557295960429,
-      "loss": 4.5627,
+      "learning_rate": 4.74237427864798e-05,
+      "loss": 2.4066,
       "step": 3051
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047464962901896127,
-      "loss": 4.9644,
+      "learning_rate": 4.7403132728771645e-05,
+      "loss": 2.8847,
       "step": 3052
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0004744435284418797,
-      "loss": 4.8746,
+      "learning_rate": 4.738252267106348e-05,
+      "loss": 2.8694,
       "step": 3053
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047423742786479803,
-      "loss": 4.7598,
+      "learning_rate": 4.736191261335532e-05,
+      "loss": 3.066,
       "step": 3054
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047403132728771643,
-      "loss": 4.9776,
+      "learning_rate": 4.734130255564716e-05,
+      "loss": 2.5408,
       "step": 3055
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047382522671063484,
-      "loss": 4.8499,
+      "learning_rate": 4.7320692497939e-05,
+      "loss": 2.7794,
       "step": 3056
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0004736191261335532,
-      "loss": 4.9817,
+      "learning_rate": 4.7300082440230835e-05,
+      "loss": 2.9915,
       "step": 3057
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.0004734130255564716,
-      "loss": 4.9736,
+      "learning_rate": 4.727947238252267e-05,
+      "loss": 2.6612,
       "step": 3058
     },
     {
       "epoch": 1.71,
-      "learning_rate": 0.00047320692497938995,
-      "loss": 4.6397,
+      "learning_rate": 4.7258862324814515e-05,
+      "loss": 2.3497,
       "step": 3059
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047300082440230836,
-      "loss": 4.674,
+      "learning_rate": 4.7238252267106344e-05,
+      "loss": 2.1956,
       "step": 3060
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004727947238252267,
-      "loss": 4.5434,
+      "learning_rate": 4.721764220939819e-05,
+      "loss": 2.1942,
       "step": 3061
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004725886232481451,
-      "loss": 4.8244,
+      "learning_rate": 4.719703215169003e-05,
+      "loss": 2.566,
       "step": 3062
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047238252267106347,
-      "loss": 5.0157,
+      "learning_rate": 4.717642209398187e-05,
+      "loss": 2.6864,
       "step": 3063
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004721764220939819,
-      "loss": 4.9611,
+      "learning_rate": 4.7155812036273704e-05,
+      "loss": 2.7088,
       "step": 3064
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004719703215169003,
-      "loss": 4.4803,
+      "learning_rate": 4.713520197856554e-05,
+      "loss": 1.9172,
       "step": 3065
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047176422093981864,
-      "loss": 5.0906,
+      "learning_rate": 4.7114591920857384e-05,
+      "loss": 2.5563,
       "step": 3066
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047155812036273704,
-      "loss": 4.6382,
+      "learning_rate": 4.7093981863149214e-05,
+      "loss": 2.1895,
       "step": 3067
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004713520197856554,
-      "loss": 5.3363,
+      "learning_rate": 4.707337180544106e-05,
+      "loss": 2.7341,
       "step": 3068
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004711459192085738,
-      "loss": 5.2619,
+      "learning_rate": 4.70527617477329e-05,
+      "loss": 2.806,
       "step": 3069
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047093981863149216,
-      "loss": 4.8895,
+      "learning_rate": 4.703215169002473e-05,
+      "loss": 2.6331,
       "step": 3070
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047073371805441056,
-      "loss": 4.8402,
+      "learning_rate": 4.7011541632316574e-05,
+      "loss": 2.3165,
       "step": 3071
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047052761747732897,
-      "loss": 4.7542,
+      "learning_rate": 4.699093157460841e-05,
+      "loss": 2.2691,
       "step": 3072
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004703215169002473,
-      "loss": 4.8944,
+      "learning_rate": 4.6970321516900254e-05,
+      "loss": 2.1118,
       "step": 3073
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00047011541632316573,
-      "loss": 4.8044,
+      "learning_rate": 4.6949711459192084e-05,
+      "loss": 2.0361,
       "step": 3074
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004699093157460841,
-      "loss": 5.2358,
+      "learning_rate": 4.692910140148393e-05,
+      "loss": 2.9748,
       "step": 3075
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.0004697032151690025,
-      "loss": 6.0982,
+      "learning_rate": 4.6908491343775764e-05,
+      "loss": 3.1366,
       "step": 3076
     },
     {
       "epoch": 1.72,
-      "learning_rate": 0.00046949711459192084,
-      "loss": 4.9189,
+      "learning_rate": 4.68878812860676e-05,
+      "loss": 2.5697,
       "step": 3077
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00046929101401483925,
-      "loss": 5.1576,
+      "learning_rate": 4.6867271228359444e-05,
+      "loss": 2.6917,
       "step": 3078
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004690849134377576,
-      "loss": 5.5019,
+      "learning_rate": 4.684666117065128e-05,
+      "loss": 2.8376,
       "step": 3079
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.000468878812860676,
-      "loss": 5.232,
+      "learning_rate": 4.682605111294312e-05,
+      "loss": 1.9518,
       "step": 3080
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004686727122835944,
-      "loss": 4.5148,
+      "learning_rate": 4.6805441055234954e-05,
+      "loss": 1.5764,
       "step": 3081
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00046846661170651276,
-      "loss": 5.0405,
+      "learning_rate": 4.67848309975268e-05,
+      "loss": 2.0084,
       "step": 3082
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00046826051112943117,
-      "loss": 4.7981,
+      "learning_rate": 4.6764220939818634e-05,
+      "loss": 2.0444,
       "step": 3083
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004680544105523495,
-      "loss": 8.6501,
+      "learning_rate": 4.674361088211047e-05,
+      "loss": 2.8447,
       "step": 3084
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00046784830997526793,
-      "loss": 5.0449,
+      "learning_rate": 4.6723000824402314e-05,
+      "loss": 2.8176,
       "step": 3085
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004676422093981863,
-      "loss": 4.9747,
+      "learning_rate": 4.670239076669415e-05,
+      "loss": 3.166,
       "step": 3086
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004674361088211047,
-      "loss": 4.8174,
+      "learning_rate": 4.668178070898599e-05,
+      "loss": 2.839,
       "step": 3087
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004672300082440231,
-      "loss": 4.945,
+      "learning_rate": 4.6661170651277824e-05,
+      "loss": 2.8115,
       "step": 3088
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00046702390766694145,
-      "loss": 4.9772,
+      "learning_rate": 4.664056059356967e-05,
+      "loss": 2.7769,
       "step": 3089
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00046681780708985986,
-      "loss": 4.5791,
+      "learning_rate": 4.6619950535861504e-05,
+      "loss": 2.2579,
       "step": 3090
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004666117065127782,
-      "loss": 4.9822,
+      "learning_rate": 4.659934047815334e-05,
+      "loss": 2.5943,
       "step": 3091
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004664056059356966,
-      "loss": 5.1276,
+      "learning_rate": 4.657873042044518e-05,
+      "loss": 3.0815,
       "step": 3092
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.000466199505358615,
-      "loss": 4.6729,
+      "learning_rate": 4.655812036273702e-05,
+      "loss": 2.7034,
       "step": 3093
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.00046599340478153343,
-      "loss": 5.0321,
+      "learning_rate": 4.653751030502886e-05,
+      "loss": 2.8431,
       "step": 3094
     },
     {
       "epoch": 1.73,
-      "learning_rate": 0.0004657873042044518,
-      "loss": 5.0349,
+      "learning_rate": 4.651690024732069e-05,
+      "loss": 2.8544,
       "step": 3095
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004655812036273702,
-      "loss": 4.955,
+      "learning_rate": 4.649629018961254e-05,
+      "loss": 2.8777,
       "step": 3096
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004653751030502886,
-      "loss": 4.9472,
+      "learning_rate": 4.647568013190437e-05,
+      "loss": 2.9223,
       "step": 3097
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046516900247320695,
-      "loss": 4.6128,
+      "learning_rate": 4.645507007419621e-05,
+      "loss": 2.3661,
       "step": 3098
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046496290189612535,
-      "loss": 4.6475,
+      "learning_rate": 4.6434460016488046e-05,
+      "loss": 2.662,
       "step": 3099
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004647568013190437,
-      "loss": 4.9007,
+      "learning_rate": 4.641384995877989e-05,
+      "loss": 2.8351,
       "step": 3100
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004645507007419621,
-      "loss": 5.0667,
+      "learning_rate": 4.6393239901071726e-05,
+      "loss": 2.7067,
       "step": 3101
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046434460016488046,
-      "loss": 4.9717,
+      "learning_rate": 4.637262984336356e-05,
+      "loss": 2.8372,
       "step": 3102
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046413849958779887,
-      "loss": 5.3151,
+      "learning_rate": 4.6352019785655406e-05,
+      "loss": 2.9956,
       "step": 3103
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004639323990107173,
-      "loss": 4.615,
+      "learning_rate": 4.6331409727947236e-05,
+      "loss": 2.3721,
       "step": 3104
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046372629843363563,
-      "loss": 5.5692,
+      "learning_rate": 4.631079967023908e-05,
+      "loss": 3.3388,
       "step": 3105
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046352019785655404,
-      "loss": 5.0304,
+      "learning_rate": 4.6290189612530916e-05,
+      "loss": 2.6021,
       "step": 3106
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004633140972794724,
-      "loss": 4.8689,
+      "learning_rate": 4.626957955482275e-05,
+      "loss": 2.651,
       "step": 3107
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004631079967023908,
-      "loss": 4.6382,
+      "learning_rate": 4.624896949711459e-05,
+      "loss": 2.1955,
       "step": 3108
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046290189612530915,
-      "loss": 4.8791,
+      "learning_rate": 4.622835943940643e-05,
+      "loss": 2.2058,
       "step": 3109
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.00046269579554822756,
-      "loss": 5.1656,
+      "learning_rate": 4.6207749381698276e-05,
+      "loss": 2.6595,
       "step": 3110
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004624896949711459,
-      "loss": 4.749,
+      "learning_rate": 4.6187139323990106e-05,
+      "loss": 3.0881,
       "step": 3111
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004622835943940643,
-      "loss": 5.302,
+      "learning_rate": 4.616652926628195e-05,
+      "loss": 3.0814,
       "step": 3112
     },
     {
       "epoch": 1.74,
-      "learning_rate": 0.0004620774938169827,
-      "loss": 4.8968,
+      "learning_rate": 4.6145919208573786e-05,
+      "loss": 2.7971,
       "step": 3113
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004618713932399011,
-      "loss": 4.7596,
+      "learning_rate": 4.612530915086562e-05,
+      "loss": 2.1389,
       "step": 3114
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004616652926628195,
-      "loss": 4.8576,
+      "learning_rate": 4.610469909315746e-05,
+      "loss": 2.8014,
       "step": 3115
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00046145919208573783,
-      "loss": 4.6059,
+      "learning_rate": 4.60840890354493e-05,
+      "loss": 2.2351,
       "step": 3116
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00046125309150865624,
-      "loss": 4.6033,
+      "learning_rate": 4.606347897774114e-05,
+      "loss": 2.358,
       "step": 3117
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004610469909315746,
-      "loss": 4.6653,
+      "learning_rate": 4.6042868920032976e-05,
+      "loss": 2.0683,
       "step": 3118
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.000460840890354493,
-      "loss": 4.7146,
+      "learning_rate": 4.602225886232482e-05,
+      "loss": 2.5401,
       "step": 3119
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004606347897774114,
-      "loss": 4.7012,
+      "learning_rate": 4.6001648804616656e-05,
+      "loss": 2.7071,
       "step": 3120
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00046042868920032976,
-      "loss": 5.0652,
+      "learning_rate": 4.598103874690849e-05,
+      "loss": 2.8192,
       "step": 3121
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00046022258862324817,
-      "loss": 4.7243,
+      "learning_rate": 4.596042868920033e-05,
+      "loss": 2.1048,
       "step": 3122
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004600164880461665,
-      "loss": 4.7944,
+      "learning_rate": 4.593981863149217e-05,
+      "loss": 2.0311,
       "step": 3123
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004598103874690849,
-      "loss": 4.9057,
+      "learning_rate": 4.591920857378401e-05,
+      "loss": 2.188,
       "step": 3124
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004596042868920033,
-      "loss": 4.7848,
+      "learning_rate": 4.5898598516075846e-05,
+      "loss": 2.7421,
       "step": 3125
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004593981863149217,
-      "loss": 4.7799,
+      "learning_rate": 4.587798845836769e-05,
+      "loss": 2.1573,
       "step": 3126
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00045919208573784004,
-      "loss": 4.6397,
+      "learning_rate": 4.5857378400659526e-05,
+      "loss": 1.94,
       "step": 3127
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00045898598516075844,
-      "loss": 4.5853,
+      "learning_rate": 4.583676834295136e-05,
+      "loss": 1.575,
       "step": 3128
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.00045877988458367685,
-      "loss": 4.6781,
+      "learning_rate": 4.58161582852432e-05,
+      "loss": 1.59,
       "step": 3129
     },
     {
       "epoch": 1.75,
-      "learning_rate": 0.0004585737840065952,
-      "loss": 4.5539,
+      "learning_rate": 4.579554822753504e-05,
+      "loss": 1.3304,
       "step": 3130
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004583676834295136,
-      "loss": 4.9331,
+      "learning_rate": 4.577493816982687e-05,
+      "loss": 1.602,
       "step": 3131
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045816158285243196,
-      "loss": 5.8975,
+      "learning_rate": 4.5754328112118715e-05,
+      "loss": 2.4049,
       "step": 3132
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045795548227535037,
-      "loss": 4.6423,
+      "learning_rate": 4.573371805441056e-05,
+      "loss": 1.3065,
       "step": 3133
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004577493816982687,
-      "loss": 4.3845,
+      "learning_rate": 4.5713107996702395e-05,
+      "loss": 1.771,
       "step": 3134
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004575432811211871,
-      "loss": 5.4291,
+      "learning_rate": 4.569249793899423e-05,
+      "loss": 3.1797,
       "step": 3135
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045733718054410553,
-      "loss": 5.2008,
+      "learning_rate": 4.567188788128607e-05,
+      "loss": 2.9335,
       "step": 3136
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004571310799670239,
-      "loss": 5.0435,
+      "learning_rate": 4.565127782357791e-05,
+      "loss": 2.9021,
       "step": 3137
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004569249793899423,
-      "loss": 4.9647,
+      "learning_rate": 4.563066776586974e-05,
+      "loss": 2.9664,
       "step": 3138
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045671887881286065,
-      "loss": 4.8151,
+      "learning_rate": 4.5610057708161585e-05,
+      "loss": 3.0992,
       "step": 3139
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004565127782357791,
-      "loss": 4.8969,
+      "learning_rate": 4.558944765045342e-05,
+      "loss": 2.781,
       "step": 3140
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045630667765869746,
-      "loss": 5.0369,
+      "learning_rate": 4.556883759274526e-05,
+      "loss": 3.0464,
       "step": 3141
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045610057708161587,
-      "loss": 5.3135,
+      "learning_rate": 4.55482275350371e-05,
+      "loss": 3.9457,
       "step": 3142
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004558944765045342,
-      "loss": 4.9621,
+      "learning_rate": 4.552761747732894e-05,
+      "loss": 2.7499,
       "step": 3143
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004556883759274526,
-      "loss": 4.8998,
+      "learning_rate": 4.550700741962078e-05,
+      "loss": 2.7107,
       "step": 3144
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045548227535037103,
-      "loss": 4.5532,
+      "learning_rate": 4.548639736191261e-05,
+      "loss": 3.2184,
       "step": 3145
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004552761747732894,
-      "loss": 4.5489,
+      "learning_rate": 4.5465787304204455e-05,
+      "loss": 2.8308,
       "step": 3146
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.0004550700741962078,
-      "loss": 4.7167,
+      "learning_rate": 4.544517724649629e-05,
+      "loss": 2.2172,
       "step": 3147
     },
     {
       "epoch": 1.76,
-      "learning_rate": 0.00045486397361912614,
-      "loss": 4.7675,
+      "learning_rate": 4.542456718878813e-05,
+      "loss": 2.9728,
       "step": 3148
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045465787304204455,
-      "loss": 4.7578,
+      "learning_rate": 4.540395713107997e-05,
+      "loss": 2.5852,
       "step": 3149
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004544517724649629,
-      "loss": 4.8245,
+      "learning_rate": 4.538334707337181e-05,
+      "loss": 2.4878,
       "step": 3150
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004542456718878813,
-      "loss": 4.7631,
+      "learning_rate": 4.5362737015663645e-05,
+      "loss": 2.5133,
       "step": 3151
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004540395713107997,
-      "loss": 5.2156,
+      "learning_rate": 4.534212695795548e-05,
+      "loss": 2.9504,
       "step": 3152
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045383347073371807,
-      "loss": 4.8006,
+      "learning_rate": 4.5321516900247325e-05,
+      "loss": 2.4554,
       "step": 3153
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004536273701566365,
-      "loss": 5.3038,
+      "learning_rate": 4.530090684253916e-05,
+      "loss": 3.0374,
       "step": 3154
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004534212695795548,
-      "loss": 4.785,
+      "learning_rate": 4.5280296784831e-05,
+      "loss": 2.3788,
       "step": 3155
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045321516900247323,
-      "loss": 4.4697,
+      "learning_rate": 4.525968672712284e-05,
+      "loss": 2.3286,
       "step": 3156
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004530090684253916,
-      "loss": 4.9542,
+      "learning_rate": 4.523907666941468e-05,
+      "loss": 3.1929,
       "step": 3157
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045280296784831,
-      "loss": 4.7452,
+      "learning_rate": 4.5218466611706515e-05,
+      "loss": 2.3037,
       "step": 3158
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004525968672712284,
-      "loss": 4.7631,
+      "learning_rate": 4.519785655399835e-05,
+      "loss": 2.7927,
       "step": 3159
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045239076669414675,
-      "loss": 4.9631,
+      "learning_rate": 4.5177246496290194e-05,
+      "loss": 2.6141,
       "step": 3160
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045218466611706516,
-      "loss": 4.7429,
+      "learning_rate": 4.515663643858203e-05,
+      "loss": 2.4185,
       "step": 3161
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004519785655399835,
-      "loss": 4.557,
+      "learning_rate": 4.513602638087387e-05,
+      "loss": 2.1819,
       "step": 3162
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004517724649629019,
-      "loss": 4.8137,
+      "learning_rate": 4.5115416323165704e-05,
+      "loss": 2.372,
       "step": 3163
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045156636438582027,
-      "loss": 5.2951,
+      "learning_rate": 4.509480626545755e-05,
+      "loss": 2.9027,
       "step": 3164
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.0004513602638087387,
-      "loss": 5.1023,
+      "learning_rate": 4.5074196207749384e-05,
+      "loss": 3.0154,
       "step": 3165
     },
     {
       "epoch": 1.77,
-      "learning_rate": 0.00045115416323165703,
-      "loss": 4.5265,
+      "learning_rate": 4.505358615004122e-05,
+      "loss": 2.1591,
       "step": 3166
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00045094806265457544,
-      "loss": 4.8865,
+      "learning_rate": 4.5032976092333064e-05,
+      "loss": 2.3673,
       "step": 3167
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00045074196207749384,
-      "loss": 4.7221,
+      "learning_rate": 4.50123660346249e-05,
+      "loss": 2.6403,
       "step": 3168
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004505358615004122,
-      "loss": 4.6497,
+      "learning_rate": 4.499175597691674e-05,
+      "loss": 2.1314,
       "step": 3169
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004503297609233306,
-      "loss": 4.6578,
+      "learning_rate": 4.4971145919208574e-05,
+      "loss": 2.5364,
       "step": 3170
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00045012366034624895,
-      "loss": 4.9729,
+      "learning_rate": 4.495053586150042e-05,
+      "loss": 2.3505,
       "step": 3171
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00044991755976916736,
-      "loss": 4.5774,
+      "learning_rate": 4.4929925803792254e-05,
+      "loss": 2.1201,
       "step": 3172
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004497114591920857,
-      "loss": 4.8821,
+      "learning_rate": 4.490931574608409e-05,
+      "loss": 2.0565,
       "step": 3173
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004495053586150041,
-      "loss": 4.6988,
+      "learning_rate": 4.4888705688375934e-05,
+      "loss": 1.8583,
       "step": 3174
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004492992580379225,
-      "loss": 4.8031,
+      "learning_rate": 4.4868095630667764e-05,
+      "loss": 2.0037,
       "step": 3175
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004490931574608409,
-      "loss": 4.9744,
+      "learning_rate": 4.484748557295961e-05,
+      "loss": 2.7244,
       "step": 3176
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004488870568837593,
-      "loss": 5.0486,
+      "learning_rate": 4.4826875515251444e-05,
+      "loss": 1.8505,
       "step": 3177
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00044868095630667764,
-      "loss": 4.7017,
+      "learning_rate": 4.480626545754329e-05,
+      "loss": 1.702,
       "step": 3178
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00044847485572959605,
-      "loss": 4.7223,
+      "learning_rate": 4.478565539983512e-05,
+      "loss": 1.6028,
       "step": 3179
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004482687551525144,
-      "loss": 5.1026,
+      "learning_rate": 4.476504534212696e-05,
+      "loss": 2.1868,
       "step": 3180
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.0004480626545754328,
-      "loss": 5.2464,
+      "learning_rate": 4.4744435284418804e-05,
+      "loss": 1.765,
       "step": 3181
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00044785655399835116,
-      "loss": 4.194,
+      "learning_rate": 4.4723825226710634e-05,
+      "loss": 1.2682,
       "step": 3182
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00044765045342126956,
-      "loss": 4.8353,
+      "learning_rate": 4.470321516900248e-05,
+      "loss": 1.5935,
       "step": 3183
     },
     {
       "epoch": 1.78,
-      "learning_rate": 0.00044744435284418797,
-      "loss": 4.6733,
+      "learning_rate": 4.4682605111294314e-05,
+      "loss": 1.5676,
       "step": 3184
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004472382522671063,
-      "loss": 4.8886,
+      "learning_rate": 4.466199505358615e-05,
+      "loss": 2.6557,
       "step": 3185
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004470321516900248,
-      "loss": 4.8368,
+      "learning_rate": 4.464138499587799e-05,
+      "loss": 2.9908,
       "step": 3186
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044682605111294314,
-      "loss": 4.8884,
+      "learning_rate": 4.462077493816983e-05,
+      "loss": 2.7847,
       "step": 3187
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044661995053586154,
-      "loss": 5.0182,
+      "learning_rate": 4.460016488046167e-05,
+      "loss": 2.5687,
       "step": 3188
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004464138499587799,
-      "loss": 5.0725,
+      "learning_rate": 4.4579554822753503e-05,
+      "loss": 3.2724,
       "step": 3189
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004462077493816983,
-      "loss": 5.0742,
+      "learning_rate": 4.455894476504535e-05,
+      "loss": 2.3899,
       "step": 3190
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004460016488046167,
-      "loss": 4.7844,
+      "learning_rate": 4.4538334707337183e-05,
+      "loss": 2.8789,
       "step": 3191
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044579554822753506,
-      "loss": 5.1153,
+      "learning_rate": 4.451772464962902e-05,
+      "loss": 3.5786,
       "step": 3192
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044558944765045347,
-      "loss": 4.9863,
+      "learning_rate": 4.449711459192086e-05,
+      "loss": 3.5573,
       "step": 3193
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004453833470733718,
-      "loss": 4.9356,
+      "learning_rate": 4.44765045342127e-05,
+      "loss": 3.5013,
       "step": 3194
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044517724649629023,
-      "loss": 4.9145,
+      "learning_rate": 4.4455894476504537e-05,
+      "loss": 2.4237,
       "step": 3195
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004449711459192086,
-      "loss": 5.1526,
+      "learning_rate": 4.443528441879637e-05,
+      "loss": 2.6512,
       "step": 3196
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.000444765045342127,
-      "loss": 4.6843,
+      "learning_rate": 4.4414674361088217e-05,
+      "loss": 2.3869,
       "step": 3197
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044455894476504534,
-      "loss": 5.0706,
+      "learning_rate": 4.439406430338005e-05,
+      "loss": 3.6391,
       "step": 3198
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044435284418796375,
-      "loss": 5.4601,
+      "learning_rate": 4.437345424567189e-05,
+      "loss": 3.3065,
       "step": 3199
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.00044414674361088215,
-      "loss": 5.3174,
+      "learning_rate": 4.4352844187963726e-05,
+      "loss": 3.2221,
       "step": 3200
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004439406430338005,
-      "loss": 4.7116,
+      "learning_rate": 4.433223413025557e-05,
+      "loss": 2.3152,
       "step": 3201
     },
     {
       "epoch": 1.79,
-      "learning_rate": 0.0004437345424567189,
-      "loss": 4.8322,
+      "learning_rate": 4.4311624072547406e-05,
+      "loss": 3.2677,
       "step": 3202
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044352844187963726,
-      "loss": 4.9297,
+      "learning_rate": 4.429101401483924e-05,
+      "loss": 2.7326,
       "step": 3203
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044332234130255567,
-      "loss": 4.9091,
+      "learning_rate": 4.4270403957131086e-05,
+      "loss": 3.0858,
       "step": 3204
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.000443116240725474,
-      "loss": 5.2406,
+      "learning_rate": 4.424979389942292e-05,
+      "loss": 2.947,
       "step": 3205
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044291014014839243,
-      "loss": 4.9367,
+      "learning_rate": 4.422918384171476e-05,
+      "loss": 2.648,
       "step": 3206
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044270403957131084,
-      "loss": 4.8079,
+      "learning_rate": 4.4208573784006596e-05,
+      "loss": 2.8332,
       "step": 3207
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004424979389942292,
-      "loss": 4.8947,
+      "learning_rate": 4.418796372629844e-05,
+      "loss": 2.4821,
       "step": 3208
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004422918384171476,
-      "loss": 4.8248,
+      "learning_rate": 4.416735366859027e-05,
+      "loss": 2.7649,
       "step": 3209
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044208573784006595,
-      "loss": 5.2378,
+      "learning_rate": 4.414674361088211e-05,
+      "loss": 2.8527,
       "step": 3210
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044187963726298435,
-      "loss": 4.9258,
+      "learning_rate": 4.412613355317395e-05,
+      "loss": 2.6508,
       "step": 3211
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004416735366859027,
-      "loss": 4.6727,
+      "learning_rate": 4.4105523495465786e-05,
+      "loss": 2.2536,
       "step": 3212
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004414674361088211,
-      "loss": 4.7168,
+      "learning_rate": 4.408491343775763e-05,
+      "loss": 2.8145,
       "step": 3213
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044126133553173947,
-      "loss": 5.2165,
+      "learning_rate": 4.4064303380049466e-05,
+      "loss": 3.0039,
       "step": 3214
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004410552349546579,
-      "loss": 4.8579,
+      "learning_rate": 4.404369332234131e-05,
+      "loss": 2.2185,
       "step": 3215
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004408491343775763,
-      "loss": 4.8613,
+      "learning_rate": 4.402308326463314e-05,
+      "loss": 2.7949,
       "step": 3216
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044064303380049463,
-      "loss": 4.6757,
+      "learning_rate": 4.400247320692498e-05,
+      "loss": 2.5589,
       "step": 3217
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.00044043693322341304,
-      "loss": 4.6173,
+      "learning_rate": 4.398186314921682e-05,
+      "loss": 2.0439,
       "step": 3218
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004402308326463314,
-      "loss": 4.7392,
+      "learning_rate": 4.3961253091508656e-05,
+      "loss": 2.117,
       "step": 3219
     },
     {
       "epoch": 1.8,
-      "learning_rate": 0.0004400247320692498,
-      "loss": 4.6896,
+      "learning_rate": 4.39406430338005e-05,
+      "loss": 2.1112,
       "step": 3220
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00043981863149216815,
-      "loss": 5.211,
+      "learning_rate": 4.3920032976092336e-05,
+      "loss": 2.477,
       "step": 3221
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00043961253091508656,
-      "loss": 4.8211,
+      "learning_rate": 4.389942291838417e-05,
+      "loss": 2.5964,
       "step": 3222
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00043940643033800496,
-      "loss": 4.5102,
+      "learning_rate": 4.387881286067601e-05,
+      "loss": 1.7906,
       "step": 3223
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004392003297609233,
-      "loss": 4.8341,
+      "learning_rate": 4.385820280296785e-05,
+      "loss": 2.3689,
       "step": 3224
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004389942291838417,
-      "loss": 4.6401,
+      "learning_rate": 4.383759274525969e-05,
+      "loss": 1.9736,
       "step": 3225
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004387881286067601,
-      "loss": 5.2141,
+      "learning_rate": 4.3816982687551526e-05,
+      "loss": 2.3419,
       "step": 3226
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004385820280296785,
-      "loss": 5.369,
+      "learning_rate": 4.379637262984336e-05,
+      "loss": 2.8586,
       "step": 3227
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00043837592745259683,
-      "loss": 4.6513,
+      "learning_rate": 4.3775762572135205e-05,
+      "loss": 1.8265,
       "step": 3228
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00043816982687551524,
-      "loss": 4.6688,
+      "learning_rate": 4.375515251442704e-05,
+      "loss": 1.6107,
       "step": 3229
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004379637262984336,
-      "loss": 4.4774,
+      "learning_rate": 4.373454245671888e-05,
+      "loss": 1.6418,
       "step": 3230
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.000437757625721352,
-      "loss": 4.4189,
+      "learning_rate": 4.371393239901072e-05,
+      "loss": 1.467,
       "step": 3231
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00043755152514427046,
-      "loss": 4.6987,
+      "learning_rate": 4.369332234130256e-05,
+      "loss": 1.5988,
       "step": 3232
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004373454245671888,
-      "loss": 5.2783,
+      "learning_rate": 4.3672712283594395e-05,
+      "loss": 1.4546,
       "step": 3233
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004371393239901072,
-      "loss": 4.6701,
+      "learning_rate": 4.365210222588623e-05,
+      "loss": 1.6001,
       "step": 3234
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.0004369332234130256,
-      "loss": 5.7257,
+      "learning_rate": 4.3631492168178075e-05,
+      "loss": 3.6545,
       "step": 3235
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.000436727122835944,
-      "loss": 5.0093,
+      "learning_rate": 4.361088211046991e-05,
+      "loss": 3.3651,
       "step": 3236
     },
     {
       "epoch": 1.81,
-      "learning_rate": 0.00043652102225886233,
-      "loss": 4.7871,
+      "learning_rate": 4.359027205276175e-05,
+      "loss": 2.5394,
       "step": 3237
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043631492168178074,
-      "loss": 4.7474,
+      "learning_rate": 4.356966199505359e-05,
+      "loss": 2.7002,
       "step": 3238
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043610882110469915,
-      "loss": 5.1011,
+      "learning_rate": 4.354905193734543e-05,
+      "loss": 2.7719,
       "step": 3239
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004359027205276175,
-      "loss": 4.7328,
+      "learning_rate": 4.3528441879637265e-05,
+      "loss": 2.2408,
       "step": 3240
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004356966199505359,
-      "loss": 4.6768,
+      "learning_rate": 4.35078318219291e-05,
+      "loss": 2.5287,
       "step": 3241
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043549051937345426,
-      "loss": 5.5402,
+      "learning_rate": 4.3487221764220945e-05,
+      "loss": 3.3514,
       "step": 3242
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043528441879637266,
-      "loss": 5.0311,
+      "learning_rate": 4.3466611706512775e-05,
+      "loss": 2.5434,
       "step": 3243
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.000435078318219291,
-      "loss": 4.999,
+      "learning_rate": 4.344600164880462e-05,
+      "loss": 2.639,
       "step": 3244
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004348722176422094,
-      "loss": 4.944,
+      "learning_rate": 4.342539159109646e-05,
+      "loss": 3.3005,
       "step": 3245
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004346661170651278,
-      "loss": 4.8487,
+      "learning_rate": 4.340478153338829e-05,
+      "loss": 2.7742,
       "step": 3246
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004344600164880462,
-      "loss": 4.9389,
+      "learning_rate": 4.3384171475680135e-05,
+      "loss": 2.8584,
       "step": 3247
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004342539159109646,
-      "loss": 4.7366,
+      "learning_rate": 4.336356141797197e-05,
+      "loss": 2.331,
       "step": 3248
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043404781533388294,
-      "loss": 4.5171,
+      "learning_rate": 4.3342951360263815e-05,
+      "loss": 2.5383,
       "step": 3249
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043384171475680135,
-      "loss": 4.7362,
+      "learning_rate": 4.3322341302555645e-05,
+      "loss": 2.5879,
       "step": 3250
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004336356141797197,
-      "loss": 4.7415,
+      "learning_rate": 4.330173124484749e-05,
+      "loss": 2.6587,
       "step": 3251
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004334295136026381,
-      "loss": 4.8461,
+      "learning_rate": 4.328112118713933e-05,
+      "loss": 2.416,
       "step": 3252
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043322341302555646,
-      "loss": 5.076,
+      "learning_rate": 4.326051112943116e-05,
+      "loss": 2.8562,
       "step": 3253
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.00043301731244847487,
-      "loss": 4.828,
+      "learning_rate": 4.3239901071723005e-05,
+      "loss": 2.4214,
       "step": 3254
     },
     {
       "epoch": 1.82,
-      "learning_rate": 0.0004328112118713933,
-      "loss": 4.9678,
+      "learning_rate": 4.321929101401484e-05,
+      "loss": 2.3829,
       "step": 3255
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004326051112943116,
-      "loss": 4.6477,
+      "learning_rate": 4.319868095630668e-05,
+      "loss": 2.3527,
       "step": 3256
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00043239901071723003,
-      "loss": 4.6926,
+      "learning_rate": 4.3178070898598514e-05,
+      "loss": 2.2371,
       "step": 3257
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004321929101401484,
-      "loss": 4.9184,
+      "learning_rate": 4.315746084089036e-05,
+      "loss": 2.6561,
       "step": 3258
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004319868095630668,
-      "loss": 4.8205,
+      "learning_rate": 4.3136850783182194e-05,
+      "loss": 2.2816,
       "step": 3259
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00043178070898598514,
-      "loss": 5.1338,
+      "learning_rate": 4.311624072547403e-05,
+      "loss": 2.805,
       "step": 3260
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00043157460840890355,
-      "loss": 5.0879,
+      "learning_rate": 4.3095630667765874e-05,
+      "loss": 2.8155,
       "step": 3261
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00043136850783182196,
-      "loss": 5.5863,
+      "learning_rate": 4.307502061005771e-05,
+      "loss": 3.3295,
       "step": 3262
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004311624072547403,
-      "loss": 4.9713,
+      "learning_rate": 4.305441055234955e-05,
+      "loss": 2.5535,
       "step": 3263
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004309563066776587,
-      "loss": 4.6285,
+      "learning_rate": 4.3033800494641384e-05,
+      "loss": 2.0859,
       "step": 3264
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00043075020610057707,
-      "loss": 5.0367,
+      "learning_rate": 4.301319043693323e-05,
+      "loss": 2.7079,
       "step": 3265
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004305441055234955,
-      "loss": 4.6544,
+      "learning_rate": 4.2992580379225064e-05,
+      "loss": 2.6893,
       "step": 3266
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00043033800494641383,
-      "loss": 4.5245,
+      "learning_rate": 4.29719703215169e-05,
+      "loss": 2.1208,
       "step": 3267
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00043013190436933224,
-      "loss": 4.6522,
+      "learning_rate": 4.2951360263808744e-05,
+      "loss": 2.5739,
       "step": 3268
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004299258037922506,
-      "loss": 4.8567,
+      "learning_rate": 4.293075020610058e-05,
+      "loss": 3.131,
       "step": 3269
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.000429719703215169,
-      "loss": 5.1076,
+      "learning_rate": 4.291014014839242e-05,
+      "loss": 2.5259,
       "step": 3270
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.0004295136026380874,
-      "loss": 4.6998,
+      "learning_rate": 4.2889530090684254e-05,
+      "loss": 2.3574,
       "step": 3271
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00042930750206100575,
-      "loss": 4.6232,
+      "learning_rate": 4.28689200329761e-05,
+      "loss": 1.9012,
       "step": 3272
     },
     {
       "epoch": 1.83,
-      "learning_rate": 0.00042910140148392416,
-      "loss": 4.7186,
+      "learning_rate": 4.2848309975267934e-05,
+      "loss": 2.1862,
       "step": 3273
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004288953009068425,
-      "loss": 4.803,
+      "learning_rate": 4.282769991755977e-05,
+      "loss": 2.126,
       "step": 3274
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004286892003297609,
-      "loss": 5.0615,
+      "learning_rate": 4.2807089859851614e-05,
+      "loss": 2.8298,
       "step": 3275
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00042848309975267927,
-      "loss": 4.8424,
+      "learning_rate": 4.278647980214345e-05,
+      "loss": 1.9434,
       "step": 3276
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004282769991755977,
-      "loss": 4.3099,
+      "learning_rate": 4.276586974443529e-05,
+      "loss": 1.8654,
       "step": 3277
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004280708985985161,
-      "loss": 5.034,
+      "learning_rate": 4.2745259686727124e-05,
+      "loss": 2.3031,
       "step": 3278
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004278647980214345,
-      "loss": 4.8562,
+      "learning_rate": 4.272464962901897e-05,
+      "loss": 2.1405,
       "step": 3279
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004276586974443529,
-      "loss": 4.4245,
+      "learning_rate": 4.27040395713108e-05,
+      "loss": 1.4788,
       "step": 3280
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00042745259686727125,
-      "loss": 4.8343,
+      "learning_rate": 4.268342951360264e-05,
+      "loss": 1.5599,
       "step": 3281
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00042724649629018966,
-      "loss": 4.3519,
+      "learning_rate": 4.266281945589448e-05,
+      "loss": 1.3699,
       "step": 3282
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.000427040395713108,
-      "loss": 5.2552,
+      "learning_rate": 4.2642209398186314e-05,
+      "loss": 1.9312,
       "step": 3283
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004268342951360264,
-      "loss": 4.2039,
+      "learning_rate": 4.262159934047816e-05,
+      "loss": 1.5273,
       "step": 3284
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00042662819455894477,
-      "loss": 4.5619,
+      "learning_rate": 4.2600989282769994e-05,
+      "loss": 2.9023,
       "step": 3285
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004264220939818632,
-      "loss": 5.1431,
+      "learning_rate": 4.258037922506184e-05,
+      "loss": 2.7552,
       "step": 3286
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004262159934047816,
-      "loss": 4.6661,
+      "learning_rate": 4.255976916735367e-05,
+      "loss": 2.4385,
       "step": 3287
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00042600989282769994,
-      "loss": 4.4522,
+      "learning_rate": 4.253915910964551e-05,
+      "loss": 2.2706,
       "step": 3288
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.00042580379225061834,
-      "loss": 5.4291,
+      "learning_rate": 4.251854905193735e-05,
+      "loss": 3.4234,
       "step": 3289
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004255976916735367,
-      "loss": 4.8745,
+      "learning_rate": 4.249793899422918e-05,
+      "loss": 2.5137,
       "step": 3290
     },
     {
       "epoch": 1.84,
-      "learning_rate": 0.0004253915910964551,
-      "loss": 5.3573,
+      "learning_rate": 4.247732893652103e-05,
+      "loss": 3.2378,
       "step": 3291
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042518549051937345,
-      "loss": 4.8001,
+      "learning_rate": 4.245671887881286e-05,
+      "loss": 2.5558,
       "step": 3292
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042497938994229186,
-      "loss": 4.7845,
+      "learning_rate": 4.24361088211047e-05,
+      "loss": 2.3999,
       "step": 3293
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042477328936521027,
-      "loss": 4.6345,
+      "learning_rate": 4.2415498763396537e-05,
+      "loss": 2.4852,
       "step": 3294
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004245671887881286,
-      "loss": 4.9105,
+      "learning_rate": 4.239488870568838e-05,
+      "loss": 2.7254,
       "step": 3295
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.000424361088211047,
-      "loss": 4.7087,
+      "learning_rate": 4.2374278647980216e-05,
+      "loss": 2.5217,
       "step": 3296
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004241549876339654,
-      "loss": 4.8645,
+      "learning_rate": 4.235366859027205e-05,
+      "loss": 2.4559,
       "step": 3297
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004239488870568838,
-      "loss": 4.7795,
+      "learning_rate": 4.233305853256389e-05,
+      "loss": 2.4568,
       "step": 3298
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042374278647980214,
-      "loss": 5.0883,
+      "learning_rate": 4.231244847485573e-05,
+      "loss": 2.8353,
       "step": 3299
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042353668590272054,
-      "loss": 4.7145,
+      "learning_rate": 4.229183841714757e-05,
+      "loss": 2.4434,
       "step": 3300
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004233305853256389,
-      "loss": 4.8431,
+      "learning_rate": 4.2271228359439406e-05,
+      "loss": 2.3523,
       "step": 3301
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004231244847485573,
-      "loss": 4.5773,
+      "learning_rate": 4.225061830173125e-05,
+      "loss": 2.204,
       "step": 3302
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004229183841714757,
-      "loss": 4.8097,
+      "learning_rate": 4.2230008244023086e-05,
+      "loss": 2.7096,
       "step": 3303
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042271228359439406,
-      "loss": 4.6245,
+      "learning_rate": 4.220939818631492e-05,
+      "loss": 2.7437,
       "step": 3304
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042250618301731247,
-      "loss": 4.8282,
+      "learning_rate": 4.218878812860676e-05,
+      "loss": 2.2291,
       "step": 3305
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004223000824402308,
-      "loss": 4.5493,
+      "learning_rate": 4.21681780708986e-05,
+      "loss": 2.3811,
       "step": 3306
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.00042209398186314923,
-      "loss": 4.8475,
+      "learning_rate": 4.214756801319044e-05,
+      "loss": 2.489,
       "step": 3307
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.0004218878812860676,
-      "loss": 4.9035,
+      "learning_rate": 4.2126957955482276e-05,
+      "loss": 2.4664,
       "step": 3308
     },
     {
       "epoch": 1.85,
-      "learning_rate": 0.000421681780708986,
-      "loss": 4.8557,
+      "learning_rate": 4.210634789777412e-05,
+      "loss": 3.0609,
       "step": 3309
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004214756801319044,
-      "loss": 4.9576,
+      "learning_rate": 4.2085737840065956e-05,
+      "loss": 3.5254,
       "step": 3310
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00042126957955482275,
-      "loss": 4.7932,
+      "learning_rate": 4.206512778235779e-05,
+      "loss": 2.69,
       "step": 3311
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00042106347897774115,
-      "loss": 4.5029,
+      "learning_rate": 4.204451772464963e-05,
+      "loss": 2.2555,
       "step": 3312
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004208573784006595,
-      "loss": 4.6352,
+      "learning_rate": 4.202390766694147e-05,
+      "loss": 2.0366,
       "step": 3313
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004206512778235779,
-      "loss": 4.9154,
+      "learning_rate": 4.20032976092333e-05,
+      "loss": 2.2549,
       "step": 3314
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00042044517724649627,
-      "loss": 4.7988,
+      "learning_rate": 4.1982687551525146e-05,
+      "loss": 2.6268,
       "step": 3315
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00042023907666941467,
-      "loss": 4.9443,
+      "learning_rate": 4.196207749381699e-05,
+      "loss": 2.2503,
       "step": 3316
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.000420032976092333,
-      "loss": 4.7826,
+      "learning_rate": 4.194146743610882e-05,
+      "loss": 2.5103,
       "step": 3317
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00041982687551525143,
-      "loss": 4.7367,
+      "learning_rate": 4.192085737840066e-05,
+      "loss": 2.4924,
       "step": 3318
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00041962077493816984,
-      "loss": 4.5704,
+      "learning_rate": 4.19002473206925e-05,
+      "loss": 2.1417,
       "step": 3319
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004194146743610882,
-      "loss": 4.7499,
+      "learning_rate": 4.187963726298434e-05,
+      "loss": 2.1792,
       "step": 3320
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004192085737840066,
-      "loss": 4.6772,
+      "learning_rate": 4.185902720527617e-05,
+      "loss": 2.1387,
       "step": 3321
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00041900247320692495,
-      "loss": 4.7127,
+      "learning_rate": 4.1838417147568016e-05,
+      "loss": 1.9194,
       "step": 3322
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00041879637262984336,
-      "loss": 5.0778,
+      "learning_rate": 4.181780708985986e-05,
+      "loss": 2.7438,
       "step": 3323
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004185902720527617,
-      "loss": 4.7434,
+      "learning_rate": 4.179719703215169e-05,
+      "loss": 2.3231,
       "step": 3324
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004183841714756801,
-      "loss": 4.7699,
+      "learning_rate": 4.177658697444353e-05,
+      "loss": 2.4042,
       "step": 3325
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.0004181780708985986,
-      "loss": 5.3764,
+      "learning_rate": 4.175597691673537e-05,
+      "loss": 3.0162,
       "step": 3326
     },
     {
       "epoch": 1.86,
-      "learning_rate": 0.00041797197032151693,
-      "loss": 4.9099,
+      "learning_rate": 4.1735366859027205e-05,
+      "loss": 2.1016,
       "step": 3327
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041776586974443534,
-      "loss": 4.5252,
+      "learning_rate": 4.171475680131904e-05,
+      "loss": 1.6573,
       "step": 3328
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004175597691673537,
-      "loss": 4.7284,
+      "learning_rate": 4.1694146743610885e-05,
+      "loss": 1.6513,
       "step": 3329
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004173536685902721,
-      "loss": 4.7711,
+      "learning_rate": 4.167353668590272e-05,
+      "loss": 1.7917,
       "step": 3330
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041714756801319045,
-      "loss": 4.6967,
+      "learning_rate": 4.165292662819456e-05,
+      "loss": 1.9126,
       "step": 3331
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041694146743610885,
-      "loss": 5.0961,
+      "learning_rate": 4.16323165704864e-05,
+      "loss": 1.5421,
       "step": 3332
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004167353668590272,
-      "loss": 5.1053,
+      "learning_rate": 4.161170651277824e-05,
+      "loss": 2.2262,
       "step": 3333
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004165292662819456,
-      "loss": 4.3585,
+      "learning_rate": 4.1591096455070075e-05,
+      "loss": 1.4906,
       "step": 3334
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.000416323165704864,
-      "loss": 4.8844,
+      "learning_rate": 4.157048639736191e-05,
+      "loss": 3.3322,
       "step": 3335
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041611706512778237,
-      "loss": 5.0525,
+      "learning_rate": 4.1549876339653755e-05,
+      "loss": 2.7272,
       "step": 3336
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004159109645507008,
-      "loss": 5.2952,
+      "learning_rate": 4.152926628194559e-05,
+      "loss": 3.4752,
       "step": 3337
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041570486397361913,
-      "loss": 4.8484,
+      "learning_rate": 4.150865622423743e-05,
+      "loss": 2.9008,
       "step": 3338
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041549876339653754,
-      "loss": 4.5423,
+      "learning_rate": 4.148804616652927e-05,
+      "loss": 2.5406,
       "step": 3339
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004152926628194559,
-      "loss": 4.7553,
+      "learning_rate": 4.146743610882111e-05,
+      "loss": 2.4909,
       "step": 3340
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004150865622423743,
-      "loss": 4.5723,
+      "learning_rate": 4.1446826051112945e-05,
+      "loss": 2.3677,
       "step": 3341
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.0004148804616652927,
-      "loss": 4.8167,
+      "learning_rate": 4.142621599340478e-05,
+      "loss": 2.7164,
       "step": 3342
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041467436108821106,
-      "loss": 4.9743,
+      "learning_rate": 4.1405605935696625e-05,
+      "loss": 2.9359,
       "step": 3343
     },
     {
       "epoch": 1.87,
-      "learning_rate": 0.00041446826051112946,
-      "loss": 4.5933,
+      "learning_rate": 4.138499587798846e-05,
+      "loss": 2.2712,
       "step": 3344
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004142621599340478,
-      "loss": 5.5062,
+      "learning_rate": 4.13643858202803e-05,
+      "loss": 3.4275,
       "step": 3345
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004140560593569662,
-      "loss": 4.4583,
+      "learning_rate": 4.1343775762572135e-05,
+      "loss": 2.3774,
       "step": 3346
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004138499587798846,
-      "loss": 4.3661,
+      "learning_rate": 4.132316570486398e-05,
+      "loss": 2.2158,
       "step": 3347
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.000413643858202803,
-      "loss": 4.6886,
+      "learning_rate": 4.1302555647155815e-05,
+      "loss": 2.1159,
       "step": 3348
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041343775762572133,
-      "loss": 4.5952,
+      "learning_rate": 4.128194558944765e-05,
+      "loss": 2.5436,
       "step": 3349
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041323165704863974,
-      "loss": 4.7484,
+      "learning_rate": 4.1261335531739495e-05,
+      "loss": 2.1375,
       "step": 3350
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041302555647155815,
-      "loss": 4.8139,
+      "learning_rate": 4.1240725474031325e-05,
+      "loss": 2.9311,
       "step": 3351
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004128194558944765,
-      "loss": 4.7335,
+      "learning_rate": 4.122011541632317e-05,
+      "loss": 2.4752,
       "step": 3352
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004126133553173949,
-      "loss": 4.6977,
+      "learning_rate": 4.1199505358615005e-05,
+      "loss": 2.4729,
       "step": 3353
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041240725474031326,
-      "loss": 4.9789,
+      "learning_rate": 4.117889530090685e-05,
+      "loss": 2.3137,
       "step": 3354
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041220115416323167,
-      "loss": 4.8396,
+      "learning_rate": 4.1158285243198685e-05,
+      "loss": 2.4931,
       "step": 3355
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041199505358615,
-      "loss": 4.6689,
+      "learning_rate": 4.113767518549052e-05,
+      "loss": 3.1288,
       "step": 3356
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004117889530090684,
-      "loss": 5.142,
+      "learning_rate": 4.1117065127782365e-05,
+      "loss": 3.2589,
       "step": 3357
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041158285243198683,
-      "loss": 4.5445,
+      "learning_rate": 4.1096455070074194e-05,
+      "loss": 2.3455,
       "step": 3358
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004113767518549052,
-      "loss": 5.269,
+      "learning_rate": 4.107584501236604e-05,
+      "loss": 3.5514,
       "step": 3359
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.0004111706512778236,
-      "loss": 5.2971,
+      "learning_rate": 4.1055234954657874e-05,
+      "loss": 3.4953,
       "step": 3360
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041096455070074194,
-      "loss": 5.1702,
+      "learning_rate": 4.103462489694971e-05,
+      "loss": 2.8691,
       "step": 3361
     },
     {
       "epoch": 1.88,
-      "learning_rate": 0.00041075845012366035,
-      "loss": 4.9609,
+      "learning_rate": 4.101401483924155e-05,
+      "loss": 2.3117,
       "step": 3362
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004105523495465787,
-      "loss": 4.5996,
+      "learning_rate": 4.099340478153339e-05,
+      "loss": 2.3283,
       "step": 3363
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004103462489694971,
-      "loss": 4.7091,
+      "learning_rate": 4.097279472382523e-05,
+      "loss": 2.7679,
       "step": 3364
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00041014014839241546,
-      "loss": 5.1736,
+      "learning_rate": 4.0952184666117064e-05,
+      "loss": 2.729,
       "step": 3365
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00040993404781533387,
-      "loss": 4.8153,
+      "learning_rate": 4.093157460840891e-05,
+      "loss": 2.4644,
       "step": 3366
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004097279472382523,
-      "loss": 4.9472,
+      "learning_rate": 4.0910964550700744e-05,
+      "loss": 3.0849,
       "step": 3367
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00040952184666117063,
-      "loss": 5.0991,
+      "learning_rate": 4.089035449299258e-05,
+      "loss": 2.2768,
       "step": 3368
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00040931574608408903,
-      "loss": 4.9006,
+      "learning_rate": 4.086974443528442e-05,
+      "loss": 3.0417,
       "step": 3369
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004091096455070074,
-      "loss": 4.7975,
+      "learning_rate": 4.084913437757626e-05,
+      "loss": 2.9209,
       "step": 3370
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004089035449299258,
-      "loss": 5.0763,
+      "learning_rate": 4.08285243198681e-05,
+      "loss": 2.9726,
       "step": 3371
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00040869744435284415,
-      "loss": 4.9482,
+      "learning_rate": 4.0807914262159934e-05,
+      "loss": 2.4809,
       "step": 3372
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004084913437757626,
-      "loss": 4.6109,
+      "learning_rate": 4.078730420445178e-05,
+      "loss": 2.3492,
       "step": 3373
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.000408285243198681,
-      "loss": 4.9554,
+      "learning_rate": 4.0766694146743614e-05,
+      "loss": 2.7744,
       "step": 3374
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00040807914262159937,
-      "loss": 5.4849,
+      "learning_rate": 4.074608408903545e-05,
+      "loss": 2.392,
       "step": 3375
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00040787304204451777,
-      "loss": 4.6041,
+      "learning_rate": 4.072547403132729e-05,
+      "loss": 1.8953,
       "step": 3376
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004076669414674361,
-      "loss": 5.0535,
+      "learning_rate": 4.070486397361913e-05,
+      "loss": 2.3126,
       "step": 3377
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.00040746084089035453,
-      "loss": 4.5565,
+      "learning_rate": 4.068425391591097e-05,
+      "loss": 2.045,
       "step": 3378
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004072547403132729,
-      "loss": 4.4918,
+      "learning_rate": 4.0663643858202804e-05,
+      "loss": 1.5107,
       "step": 3379
     },
     {
       "epoch": 1.89,
-      "learning_rate": 0.0004070486397361913,
-      "loss": 4.3907,
+      "learning_rate": 4.064303380049465e-05,
+      "loss": 1.5778,
       "step": 3380
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004068425391591097,
-      "loss": 6.3704,
+      "learning_rate": 4.0622423742786484e-05,
+      "loss": 2.439,
       "step": 3381
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040663643858202805,
-      "loss": 4.7241,
+      "learning_rate": 4.060181368507832e-05,
+      "loss": 1.5275,
       "step": 3382
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040643033800494646,
-      "loss": 4.9549,
+      "learning_rate": 4.058120362737016e-05,
+      "loss": 1.5211,
       "step": 3383
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004062242374278648,
-      "loss": 4.4062,
+      "learning_rate": 4.0560593569662e-05,
+      "loss": 1.5546,
       "step": 3384
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004060181368507832,
-      "loss": 4.7854,
+      "learning_rate": 4.053998351195383e-05,
+      "loss": 2.6316,
       "step": 3385
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040581203627370157,
-      "loss": 4.8542,
+      "learning_rate": 4.0519373454245673e-05,
+      "loss": 2.6369,
       "step": 3386
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040560593569662,
-      "loss": 4.9975,
+      "learning_rate": 4.049876339653752e-05,
+      "loss": 3.0145,
       "step": 3387
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040539983511953833,
-      "loss": 5.1747,
+      "learning_rate": 4.047815333882935e-05,
+      "loss": 2.9603,
       "step": 3388
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040519373454245673,
-      "loss": 4.7826,
+      "learning_rate": 4.045754328112119e-05,
+      "loss": 2.4983,
       "step": 3389
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040498763396537514,
-      "loss": 5.008,
+      "learning_rate": 4.043693322341303e-05,
+      "loss": 3.3296,
       "step": 3390
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004047815333882935,
-      "loss": 4.9458,
+      "learning_rate": 4.041632316570487e-05,
+      "loss": 2.7705,
       "step": 3391
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004045754328112119,
-      "loss": 4.6367,
+      "learning_rate": 4.03957131079967e-05,
+      "loss": 2.8044,
       "step": 3392
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040436933223413025,
-      "loss": 4.6591,
+      "learning_rate": 4.037510305028854e-05,
+      "loss": 2.7043,
       "step": 3393
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.00040416323165704866,
-      "loss": 4.9568,
+      "learning_rate": 4.0354492992580387e-05,
+      "loss": 3.1213,
       "step": 3394
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.000403957131079967,
-      "loss": 4.8437,
+      "learning_rate": 4.0333882934872216e-05,
+      "loss": 2.96,
       "step": 3395
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004037510305028854,
-      "loss": 5.0203,
+      "learning_rate": 4.031327287716406e-05,
+      "loss": 3.2054,
       "step": 3396
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004035449299258038,
-      "loss": 4.5651,
+      "learning_rate": 4.0292662819455896e-05,
+      "loss": 2.6332,
       "step": 3397
     },
     {
       "epoch": 1.9,
-      "learning_rate": 0.0004033388293487222,
-      "loss": 5.0757,
+      "learning_rate": 4.027205276174773e-05,
+      "loss": 2.8258,
       "step": 3398
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0004031327287716406,
-      "loss": 4.6867,
+      "learning_rate": 4.025144270403957e-05,
+      "loss": 2.5712,
       "step": 3399
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040292662819455894,
-      "loss": 4.5968,
+      "learning_rate": 4.023083264633141e-05,
+      "loss": 2.4452,
       "step": 3400
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040272052761747734,
-      "loss": 4.6518,
+      "learning_rate": 4.021022258862325e-05,
+      "loss": 2.4873,
       "step": 3401
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0004025144270403957,
-      "loss": 4.5181,
+      "learning_rate": 4.0189612530915086e-05,
+      "loss": 2.3594,
       "step": 3402
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0004023083264633141,
-      "loss": 4.9078,
+      "learning_rate": 4.016900247320693e-05,
+      "loss": 2.9511,
       "step": 3403
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040210222588623246,
-      "loss": 4.7953,
+      "learning_rate": 4.0148392415498766e-05,
+      "loss": 3.2847,
       "step": 3404
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040189612530915086,
-      "loss": 4.6249,
+      "learning_rate": 4.01277823577906e-05,
+      "loss": 2.3453,
       "step": 3405
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040169002473206927,
-      "loss": 4.7323,
+      "learning_rate": 4.010717230008244e-05,
+      "loss": 2.1486,
       "step": 3406
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0004014839241549876,
-      "loss": 4.7391,
+      "learning_rate": 4.008656224237428e-05,
+      "loss": 2.6029,
       "step": 3407
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040127782357790603,
-      "loss": 5.1484,
+      "learning_rate": 4.006595218466612e-05,
+      "loss": 3.1623,
       "step": 3408
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0004010717230008244,
-      "loss": 5.1443,
+      "learning_rate": 4.0045342126957956e-05,
+      "loss": 2.8397,
       "step": 3409
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0004008656224237428,
-      "loss": 4.8747,
+      "learning_rate": 4.00247320692498e-05,
+      "loss": 2.3979,
       "step": 3410
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040065952184666114,
-      "loss": 4.756,
+      "learning_rate": 4.0004122011541636e-05,
+      "loss": 2.3542,
       "step": 3411
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040045342126957955,
-      "loss": 4.8372,
+      "learning_rate": 3.998351195383347e-05,
+      "loss": 2.1855,
       "step": 3412
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00040024732069249795,
-      "loss": 4.4609,
+      "learning_rate": 3.996290189612531e-05,
+      "loss": 2.247,
       "step": 3413
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0004000412201154163,
-      "loss": 4.6454,
+      "learning_rate": 3.994229183841715e-05,
+      "loss": 2.4286,
       "step": 3414
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.0003998351195383347,
-      "loss": 4.7215,
+      "learning_rate": 3.992168178070899e-05,
+      "loss": 2.3065,
       "step": 3415
     },
     {
       "epoch": 1.91,
-      "learning_rate": 0.00039962901896125306,
-      "loss": 4.6686,
+      "learning_rate": 3.9901071723000826e-05,
+      "loss": 2.2158,
       "step": 3416
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039942291838417147,
-      "loss": 4.7053,
+      "learning_rate": 3.988046166529266e-05,
+      "loss": 2.3462,
       "step": 3417
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0003992168178070898,
-      "loss": 4.5054,
+      "learning_rate": 3.9859851607584506e-05,
+      "loss": 2.181,
       "step": 3418
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0003990107172300083,
-      "loss": 4.6862,
+      "learning_rate": 3.983924154987634e-05,
+      "loss": 2.3973,
       "step": 3419
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039880461665292664,
-      "loss": 4.5206,
+      "learning_rate": 3.981863149216818e-05,
+      "loss": 1.9094,
       "step": 3420
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039859851607584504,
-      "loss": 4.9256,
+      "learning_rate": 3.979802143446002e-05,
+      "loss": 2.5718,
       "step": 3421
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039839241549876345,
-      "loss": 4.8188,
+      "learning_rate": 3.977741137675185e-05,
+      "loss": 2.9635,
       "step": 3422
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0003981863149216818,
-      "loss": 4.7555,
+      "learning_rate": 3.9756801319043696e-05,
+      "loss": 2.6985,
       "step": 3423
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0003979802143446002,
-      "loss": 4.5959,
+      "learning_rate": 3.973619126133553e-05,
+      "loss": 2.1686,
       "step": 3424
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039777411376751856,
-      "loss": 4.6231,
+      "learning_rate": 3.9715581203627376e-05,
+      "loss": 1.9487,
       "step": 3425
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039756801319043697,
-      "loss": 4.5784,
+      "learning_rate": 3.969497114591921e-05,
+      "loss": 2.0194,
       "step": 3426
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0003973619126133553,
-      "loss": 4.4765,
+      "learning_rate": 3.967436108821105e-05,
+      "loss": 1.7929,
       "step": 3427
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039715581203627373,
-      "loss": 4.8259,
+      "learning_rate": 3.965375103050289e-05,
+      "loss": 2.6202,
       "step": 3428
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039694971145919213,
-      "loss": 4.3661,
+      "learning_rate": 3.963314097279472e-05,
+      "loss": 1.6535,
       "step": 3429
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0003967436108821105,
-      "loss": 5.837,
+      "learning_rate": 3.9612530915086565e-05,
+      "loss": 2.3864,
       "step": 3430
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.0003965375103050289,
-      "loss": 4.5403,
+      "learning_rate": 3.95919208573784e-05,
+      "loss": 1.5554,
       "step": 3431
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039633140972794725,
-      "loss": 5.1954,
+      "learning_rate": 3.957131079967024e-05,
+      "loss": 2.2424,
       "step": 3432
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.00039612530915086565,
-      "loss": 4.274,
+      "learning_rate": 3.9550700741962075e-05,
+      "loss": 1.2913,
       "step": 3433
     },
     {
       "epoch": 1.92,
-      "learning_rate": 0.000395919208573784,
-      "loss": 4.4951,
+      "learning_rate": 3.953009068425392e-05,
+      "loss": 1.6073,
       "step": 3434
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003957131079967024,
-      "loss": 4.7472,
+      "learning_rate": 3.9509480626545755e-05,
+      "loss": 2.7603,
       "step": 3435
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039550700741962076,
-      "loss": 4.7832,
+      "learning_rate": 3.948887056883759e-05,
+      "loss": 2.7889,
       "step": 3436
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039530090684253917,
-      "loss": 4.7386,
+      "learning_rate": 3.9468260511129435e-05,
+      "loss": 2.9682,
       "step": 3437
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003950948062654576,
-      "loss": 4.7036,
+      "learning_rate": 3.944765045342127e-05,
+      "loss": 2.8027,
       "step": 3438
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039488870568837593,
-      "loss": 4.6697,
+      "learning_rate": 3.942704039571311e-05,
+      "loss": 2.5216,
       "step": 3439
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039468260511129434,
-      "loss": 4.7119,
+      "learning_rate": 3.9406430338004945e-05,
+      "loss": 3.1102,
       "step": 3440
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003944765045342127,
-      "loss": 5.2962,
+      "learning_rate": 3.938582028029679e-05,
+      "loss": 3.3938,
       "step": 3441
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003942704039571311,
-      "loss": 4.8954,
+      "learning_rate": 3.9365210222588625e-05,
+      "loss": 2.7588,
       "step": 3442
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039406430338004945,
-      "loss": 5.0259,
+      "learning_rate": 3.934460016488046e-05,
+      "loss": 2.9684,
       "step": 3443
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039385820280296786,
-      "loss": 4.7879,
+      "learning_rate": 3.9323990107172305e-05,
+      "loss": 2.8398,
       "step": 3444
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039365210222588626,
-      "loss": 4.8812,
+      "learning_rate": 3.930338004946414e-05,
+      "loss": 2.9794,
       "step": 3445
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003934460016488046,
-      "loss": 4.8308,
+      "learning_rate": 3.928276999175598e-05,
+      "loss": 2.5178,
       "step": 3446
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.000393239901071723,
-      "loss": 5.0648,
+      "learning_rate": 3.9262159934047815e-05,
+      "loss": 2.6355,
       "step": 3447
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003930338004946414,
-      "loss": 4.6062,
+      "learning_rate": 3.924154987633966e-05,
+      "loss": 2.3203,
       "step": 3448
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003928276999175598,
-      "loss": 4.8874,
+      "learning_rate": 3.9220939818631495e-05,
+      "loss": 3.1074,
       "step": 3449
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039262159934047813,
-      "loss": 5.0936,
+      "learning_rate": 3.920032976092333e-05,
+      "loss": 3.0187,
       "step": 3450
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.00039241549876339654,
-      "loss": 4.5317,
+      "learning_rate": 3.9179719703215175e-05,
+      "loss": 2.5897,
       "step": 3451
     },
     {
       "epoch": 1.93,
-      "learning_rate": 0.0003922093981863149,
-      "loss": 4.8123,
+      "learning_rate": 3.915910964550701e-05,
+      "loss": 2.4237,
       "step": 3452
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003920032976092333,
-      "loss": 4.3349,
+      "learning_rate": 3.913849958779885e-05,
+      "loss": 2.7039,
       "step": 3453
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003917971970321517,
-      "loss": 4.7756,
+      "learning_rate": 3.9117889530090684e-05,
+      "loss": 3.0128,
       "step": 3454
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00039159109645507006,
-      "loss": 4.5968,
+      "learning_rate": 3.909727947238253e-05,
+      "loss": 3.0176,
       "step": 3455
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00039138499587798846,
-      "loss": 4.4312,
+      "learning_rate": 3.907666941467436e-05,
+      "loss": 2.3582,
       "step": 3456
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003911788953009068,
-      "loss": 4.8211,
+      "learning_rate": 3.90560593569662e-05,
+      "loss": 2.4583,
       "step": 3457
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003909727947238252,
-      "loss": 4.9525,
+      "learning_rate": 3.9035449299258044e-05,
+      "loss": 2.9678,
       "step": 3458
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003907666941467436,
-      "loss": 4.8558,
+      "learning_rate": 3.9014839241549874e-05,
+      "loss": 2.7522,
       "step": 3459
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.000390560593569662,
-      "loss": 4.5851,
+      "learning_rate": 3.899422918384172e-05,
+      "loss": 2.5758,
       "step": 3460
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003903544929925804,
-      "loss": 4.4795,
+      "learning_rate": 3.8973619126133554e-05,
+      "loss": 2.0862,
       "step": 3461
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00039014839241549874,
-      "loss": 4.2778,
+      "learning_rate": 3.89530090684254e-05,
+      "loss": 2.1651,
       "step": 3462
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.00038994229183841715,
-      "loss": 4.8323,
+      "learning_rate": 3.893239901071723e-05,
+      "loss": 2.5925,
       "step": 3463
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003897361912613355,
-      "loss": 4.5137,
+      "learning_rate": 3.891178895300907e-05,
+      "loss": 2.1555,
       "step": 3464
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003895300906842539,
-      "loss": 4.9396,
+      "learning_rate": 3.889117889530091e-05,
+      "loss": 2.9565,
       "step": 3465
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003893239901071723,
-      "loss": 5.0272,
+      "learning_rate": 3.8870568837592744e-05,
+      "loss": 3.2902,
       "step": 3466
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003891178895300907,
-      "loss": 4.7508,
+      "learning_rate": 3.884995877988459e-05,
+      "loss": 2.989,
       "step": 3467
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003889117889530091,
-      "loss": 4.433,
+      "learning_rate": 3.8829348722176424e-05,
+      "loss": 2.6123,
       "step": 3468
     },
     {
       "epoch": 1.94,
-      "learning_rate": 0.0003887056883759275,
-      "loss": 4.4877,
+      "learning_rate": 3.880873866446826e-05,
+      "loss": 1.9073,
       "step": 3469
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0003884995877988459,
-      "loss": 5.4642,
+      "learning_rate": 3.87881286067601e-05,
+      "loss": 3.0128,
       "step": 3470
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038829348722176424,
-      "loss": 4.4374,
+      "learning_rate": 3.876751854905194e-05,
+      "loss": 1.9756,
       "step": 3471
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038808738664468265,
-      "loss": 4.611,
+      "learning_rate": 3.874690849134378e-05,
+      "loss": 1.7938,
       "step": 3472
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.000387881286067601,
-      "loss": 5.8008,
+      "learning_rate": 3.8726298433635614e-05,
+      "loss": 2.745,
       "step": 3473
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0003876751854905194,
-      "loss": 5.458,
+      "learning_rate": 3.870568837592746e-05,
+      "loss": 2.7922,
       "step": 3474
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038746908491343776,
-      "loss": 4.496,
+      "learning_rate": 3.8685078318219294e-05,
+      "loss": 1.8327,
       "step": 3475
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038726298433635616,
-      "loss": 4.5265,
+      "learning_rate": 3.866446826051113e-05,
+      "loss": 2.0764,
       "step": 3476
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038705688375927457,
-      "loss": 4.8058,
+      "learning_rate": 3.864385820280297e-05,
+      "loss": 1.8988,
       "step": 3477
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0003868507831821929,
-      "loss": 4.735,
+      "learning_rate": 3.862324814509481e-05,
+      "loss": 1.8484,
       "step": 3478
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038664468260511133,
-      "loss": 5.0505,
+      "learning_rate": 3.860263808738665e-05,
+      "loss": 2.3245,
       "step": 3479
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0003864385820280297,
-      "loss": 4.7965,
+      "learning_rate": 3.8582028029678484e-05,
+      "loss": 1.5965,
       "step": 3480
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0003862324814509481,
-      "loss": 4.9624,
+      "learning_rate": 3.856141797197033e-05,
+      "loss": 1.931,
       "step": 3481
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038602638087386644,
-      "loss": 4.3876,
+      "learning_rate": 3.8540807914262164e-05,
+      "loss": 1.4004,
       "step": 3482
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038582028029678485,
-      "loss": 4.846,
+      "learning_rate": 3.8520197856554e-05,
+      "loss": 1.987,
       "step": 3483
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038561417971970326,
-      "loss": 4.651,
+      "learning_rate": 3.849958779884584e-05,
+      "loss": 1.4458,
       "step": 3484
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.0003854080791426216,
-      "loss": 4.9595,
+      "learning_rate": 3.847897774113768e-05,
+      "loss": 2.9637,
       "step": 3485
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038520197856554,
-      "loss": 4.6494,
+      "learning_rate": 3.845836768342952e-05,
+      "loss": 2.7212,
       "step": 3486
     },
     {
       "epoch": 1.95,
-      "learning_rate": 0.00038499587798845837,
-      "loss": 4.8342,
+      "learning_rate": 3.843775762572135e-05,
+      "loss": 2.5523,
       "step": 3487
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003847897774113768,
-      "loss": 4.728,
+      "learning_rate": 3.841714756801319e-05,
+      "loss": 2.5999,
       "step": 3488
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003845836768342951,
-      "loss": 4.9452,
+      "learning_rate": 3.839653751030503e-05,
+      "loss": 2.9462,
       "step": 3489
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00038437757625721353,
-      "loss": 5.6787,
+      "learning_rate": 3.837592745259687e-05,
+      "loss": 3.5366,
       "step": 3490
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003841714756801319,
-      "loss": 4.6806,
+      "learning_rate": 3.8355317394888707e-05,
+      "loss": 2.5996,
       "step": 3491
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003839653751030503,
-      "loss": 4.8563,
+      "learning_rate": 3.833470733718055e-05,
+      "loss": 2.6982,
       "step": 3492
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003837592745259687,
-      "loss": 4.8826,
+      "learning_rate": 3.831409727947238e-05,
+      "loss": 2.8804,
       "step": 3493
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00038355317394888705,
-      "loss": 4.7766,
+      "learning_rate": 3.829348722176422e-05,
+      "loss": 2.8606,
       "step": 3494
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00038334707337180546,
-      "loss": 4.8236,
+      "learning_rate": 3.827287716405606e-05,
+      "loss": 2.5649,
       "step": 3495
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003831409727947238,
-      "loss": 4.9607,
+      "learning_rate": 3.82522671063479e-05,
+      "loss": 2.6038,
       "step": 3496
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003829348722176422,
-      "loss": 4.7433,
+      "learning_rate": 3.823165704863974e-05,
+      "loss": 2.2936,
       "step": 3497
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00038272877164056057,
-      "loss": 4.6359,
+      "learning_rate": 3.8211046990931576e-05,
+      "loss": 2.366,
       "step": 3498
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.000382522671063479,
-      "loss": 4.7336,
+      "learning_rate": 3.819043693322342e-05,
+      "loss": 2.445,
       "step": 3499
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003823165704863974,
-      "loss": 4.566,
+      "learning_rate": 3.816982687551525e-05,
+      "loss": 2.2759,
       "step": 3500
     },
     {
       "epoch": 1.96,
-      "eval_loss": 5.284274101257324,
-      "eval_runtime": 1249.3998,
+      "eval_loss": 6.309370040893555,
+      "eval_runtime": 1248.8846,
       "eval_samples_per_second": 2.115,
       "eval_steps_per_second": 0.265,
-      "eval_wer": 1.9802261007536692,
+      "eval_wer": 1.7494248314161047,
       "step": 3500
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00038211046990931574,
-      "loss": 4.8114,
+      "learning_rate": 3.814921681780709e-05,
+      "loss": 2.401,
       "step": 3501
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00038190436933223414,
-      "loss": 6.0071,
+      "learning_rate": 3.812860676009893e-05,
+      "loss": 3.2168,
       "step": 3502
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003816982687551525,
-      "loss": 4.852,
+      "learning_rate": 3.8107996702390766e-05,
+      "loss": 2.5309,
       "step": 3503
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.0003814921681780709,
-      "loss": 4.3603,
+      "learning_rate": 3.80873866446826e-05,
+      "loss": 2.2873,
       "step": 3504
     },
     {
       "epoch": 1.96,
-      "learning_rate": 0.00038128606760098925,
-      "loss": 4.451,
+      "learning_rate": 3.8066776586974446e-05,
+      "loss": 2.1887,
       "step": 3505
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00038107996702390766,
-      "loss": 4.8503,
+      "learning_rate": 3.804616652926629e-05,
+      "loss": 2.5726,
       "step": 3506
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.000380873866446826,
-      "loss": 4.6641,
+      "learning_rate": 3.802555647155812e-05,
+      "loss": 2.3932,
       "step": 3507
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003806677658697444,
-      "loss": 4.9771,
+      "learning_rate": 3.800494641384996e-05,
+      "loss": 2.8668,
       "step": 3508
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003804616652926628,
-      "loss": 4.9483,
+      "learning_rate": 3.79843363561418e-05,
+      "loss": 2.4369,
       "step": 3509
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003802555647155812,
-      "loss": 5.0402,
+      "learning_rate": 3.7963726298433636e-05,
+      "loss": 3.0359,
       "step": 3510
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003800494641384996,
-      "loss": 4.2775,
+      "learning_rate": 3.794311624072547e-05,
+      "loss": 2.2576,
       "step": 3511
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00037984336356141794,
-      "loss": 4.7486,
+      "learning_rate": 3.7922506183017316e-05,
+      "loss": 2.3818,
       "step": 3512
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003796372629843364,
-      "loss": 4.8403,
+      "learning_rate": 3.790189612530915e-05,
+      "loss": 2.9305,
       "step": 3513
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00037943116240725475,
-      "loss": 4.4698,
+      "learning_rate": 3.788128606760099e-05,
+      "loss": 2.1396,
       "step": 3514
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00037922506183017316,
-      "loss": 4.7063,
+      "learning_rate": 3.786067600989283e-05,
+      "loss": 2.2274,
       "step": 3515
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00037901896125309157,
-      "loss": 4.5009,
+      "learning_rate": 3.784006595218467e-05,
+      "loss": 2.0323,
       "step": 3516
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003788128606760099,
-      "loss": 4.4723,
+      "learning_rate": 3.7819455894476506e-05,
+      "loss": 2.0801,
       "step": 3517
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003786067600989283,
-      "loss": 4.8936,
+      "learning_rate": 3.779884583676834e-05,
+      "loss": 2.8402,
       "step": 3518
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003784006595218467,
-      "loss": 4.5395,
+      "learning_rate": 3.7778235779060186e-05,
+      "loss": 2.1351,
       "step": 3519
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003781945589447651,
-      "loss": 4.5365,
+      "learning_rate": 3.775762572135202e-05,
+      "loss": 2.047,
       "step": 3520
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00037798845836768344,
-      "loss": 4.6528,
+      "learning_rate": 3.773701566364386e-05,
+      "loss": 2.263,
       "step": 3521
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.00037778235779060184,
-      "loss": 4.8759,
+      "learning_rate": 3.77164056059357e-05,
+      "loss": 2.1401,
       "step": 3522
     },
     {
       "epoch": 1.97,
-      "learning_rate": 0.0003775762572135202,
-      "loss": 4.8592,
+      "learning_rate": 3.769579554822754e-05,
+      "loss": 2.4023,
       "step": 3523
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003773701566364386,
-      "loss": 5.1568,
+      "learning_rate": 3.7675185490519375e-05,
+      "loss": 3.2837,
       "step": 3524
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.000377164056059357,
-      "loss": 4.769,
+      "learning_rate": 3.765457543281121e-05,
+      "loss": 1.8719,
       "step": 3525
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037695795548227536,
-      "loss": 4.8461,
+      "learning_rate": 3.7633965375103055e-05,
+      "loss": 2.1283,
       "step": 3526
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037675185490519377,
-      "loss": 4.8091,
+      "learning_rate": 3.7613355317394885e-05,
+      "loss": 1.9624,
       "step": 3527
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003765457543281121,
-      "loss": 4.93,
+      "learning_rate": 3.759274525968673e-05,
+      "loss": 1.7317,
       "step": 3528
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003763396537510305,
-      "loss": 4.5999,
+      "learning_rate": 3.757213520197857e-05,
+      "loss": 1.5915,
       "step": 3529
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003761335531739489,
-      "loss": 4.5166,
+      "learning_rate": 3.755152514427041e-05,
+      "loss": 1.7033,
       "step": 3530
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003759274525968673,
-      "loss": 4.8218,
+      "learning_rate": 3.7530915086562245e-05,
+      "loss": 1.6335,
       "step": 3531
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003757213520197857,
-      "loss": 4.739,
+      "learning_rate": 3.751030502885408e-05,
+      "loss": 1.5016,
       "step": 3532
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037551525144270405,
-      "loss": 4.2466,
+      "learning_rate": 3.7489694971145925e-05,
+      "loss": 1.4779,
       "step": 3533
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037530915086562245,
-      "loss": 4.2381,
+      "learning_rate": 3.7469084913437755e-05,
+      "loss": 1.5341,
       "step": 3534
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003751030502885408,
-      "loss": 4.7059,
+      "learning_rate": 3.74484748557296e-05,
+      "loss": 2.6322,
       "step": 3535
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003748969497114592,
-      "loss": 4.7721,
+      "learning_rate": 3.7427864798021435e-05,
+      "loss": 3.4075,
       "step": 3536
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037469084913437756,
-      "loss": 4.7028,
+      "learning_rate": 3.740725474031327e-05,
+      "loss": 2.4885,
       "step": 3537
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037448474855729597,
-      "loss": 4.5897,
+      "learning_rate": 3.7386644682605115e-05,
+      "loss": 2.4961,
       "step": 3538
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.0003742786479802143,
-      "loss": 4.6412,
+      "learning_rate": 3.736603462489695e-05,
+      "loss": 2.4181,
       "step": 3539
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037407254740313273,
-      "loss": 4.6575,
+      "learning_rate": 3.734542456718879e-05,
+      "loss": 2.8901,
       "step": 3540
     },
     {
       "epoch": 1.98,
-      "learning_rate": 0.00037386644682605114,
-      "loss": 5.1197,
+      "learning_rate": 3.7324814509480625e-05,
+      "loss": 3.104,
       "step": 3541
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003736603462489695,
-      "loss": 4.4617,
+      "learning_rate": 3.730420445177247e-05,
+      "loss": 2.3141,
       "step": 3542
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003734542456718879,
-      "loss": 5.1106,
+      "learning_rate": 3.7283594394064305e-05,
+      "loss": 3.3087,
       "step": 3543
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037324814509480625,
-      "loss": 4.5566,
+      "learning_rate": 3.726298433635614e-05,
+      "loss": 2.4229,
       "step": 3544
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037304204451772465,
-      "loss": 4.4198,
+      "learning_rate": 3.7242374278647985e-05,
+      "loss": 2.2154,
       "step": 3545
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.000372835943940643,
-      "loss": 4.6055,
+      "learning_rate": 3.722176422093982e-05,
+      "loss": 2.492,
       "step": 3546
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003726298433635614,
-      "loss": 5.1071,
+      "learning_rate": 3.720115416323166e-05,
+      "loss": 2.8221,
       "step": 3547
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003724237427864798,
-      "loss": 4.9423,
+      "learning_rate": 3.7180544105523495e-05,
+      "loss": 2.9635,
       "step": 3548
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037221764220939817,
-      "loss": 4.5571,
+      "learning_rate": 3.715993404781534e-05,
+      "loss": 2.3267,
       "step": 3549
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003720115416323166,
-      "loss": 4.6288,
+      "learning_rate": 3.7139323990107175e-05,
+      "loss": 2.1839,
       "step": 3550
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037180544105523493,
-      "loss": 4.9229,
+      "learning_rate": 3.711871393239901e-05,
+      "loss": 2.5399,
       "step": 3551
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037159934047815334,
-      "loss": 4.4566,
+      "learning_rate": 3.709810387469085e-05,
+      "loss": 2.1042,
       "step": 3552
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003713932399010717,
-      "loss": 4.8379,
+      "learning_rate": 3.707749381698269e-05,
+      "loss": 2.9674,
       "step": 3553
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003711871393239901,
-      "loss": 5.0964,
+      "learning_rate": 3.705688375927453e-05,
+      "loss": 2.6598,
       "step": 3554
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037098103874690845,
-      "loss": 5.0211,
+      "learning_rate": 3.7036273701566364e-05,
+      "loss": 2.7575,
       "step": 3555
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037077493816982686,
-      "loss": 5.2793,
+      "learning_rate": 3.701566364385821e-05,
+      "loss": 2.6727,
       "step": 3556
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.00037056883759274526,
-      "loss": 4.647,
+      "learning_rate": 3.6995053586150044e-05,
+      "loss": 2.0124,
       "step": 3557
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003703627370156636,
-      "loss": 4.7218,
+      "learning_rate": 3.697444352844188e-05,
+      "loss": 2.3189,
       "step": 3558
     },
     {
       "epoch": 1.99,
-      "learning_rate": 0.0003701566364385821,
-      "loss": 4.9586,
+      "learning_rate": 3.695383347073372e-05,
+      "loss": 2.5291,
       "step": 3559
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036995053586150043,
-      "loss": 4.3742,
+      "learning_rate": 3.693322341302556e-05,
+      "loss": 2.1168,
       "step": 3560
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036974443528441884,
-      "loss": 4.5387,
+      "learning_rate": 3.69126133553174e-05,
+      "loss": 2.1073,
       "step": 3561
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003695383347073372,
-      "loss": 4.7167,
+      "learning_rate": 3.6892003297609234e-05,
+      "loss": 1.9174,
       "step": 3562
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003693322341302556,
-      "loss": 4.7,
+      "learning_rate": 3.687139323990108e-05,
+      "loss": 2.0986,
       "step": 3563
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.000369126133553174,
-      "loss": 5.0233,
+      "learning_rate": 3.685078318219291e-05,
+      "loss": 1.8672,
       "step": 3564
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036892003297609235,
-      "loss": 4.5137,
+      "learning_rate": 3.683017312448475e-05,
+      "loss": 1.5825,
       "step": 3565
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036871393239901076,
-      "loss": 5.0293,
+      "learning_rate": 3.680956306677659e-05,
+      "loss": 1.9076,
       "step": 3566
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003685078318219291,
-      "loss": 5.1896,
+      "learning_rate": 3.678895300906843e-05,
+      "loss": 1.7297,
       "step": 3567
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003683017312448475,
-      "loss": 3.5698,
+      "learning_rate": 3.678895300906843e-05,
+      "loss": 1.422,
       "step": 3568
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003680956306677659,
-      "loss": 4.3443,
+      "learning_rate": 3.676834295136026e-05,
+      "loss": 2.5333,
       "step": 3569
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003678895300906843,
-      "loss": 4.1845,
+      "learning_rate": 3.6747732893652104e-05,
+      "loss": 2.1597,
       "step": 3570
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036768342951360263,
-      "loss": 4.2753,
+      "learning_rate": 3.672712283594395e-05,
+      "loss": 2.4194,
       "step": 3571
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036747732893652104,
-      "loss": 4.1534,
+      "learning_rate": 3.670651277823578e-05,
+      "loss": 2.0196,
       "step": 3572
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036727122835943945,
-      "loss": 4.1776,
+      "learning_rate": 3.668590272052762e-05,
+      "loss": 2.2929,
       "step": 3573
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003670651277823578,
-      "loss": 4.0363,
+      "learning_rate": 3.666529266281946e-05,
+      "loss": 2.0931,
       "step": 3574
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.0003668590272052762,
-      "loss": 4.3003,
+      "learning_rate": 3.6644682605111294e-05,
+      "loss": 2.2114,
       "step": 3575
     },
     {
       "epoch": 2.0,
-      "learning_rate": 0.00036665292662819456,
-      "loss": 4.1835,
+      "learning_rate": 3.662407254740313e-05,
+      "loss": 1.9893,
       "step": 3576
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036644682605111296,
-      "loss": 4.5585,
+      "learning_rate": 3.6603462489694974e-05,
+      "loss": 2.6441,
       "step": 3577
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003662407254740313,
-      "loss": 4.1136,
+      "learning_rate": 3.658285243198682e-05,
+      "loss": 1.9169,
       "step": 3578
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003660346248969497,
-      "loss": 4.4215,
+      "learning_rate": 3.656224237427865e-05,
+      "loss": 2.7692,
       "step": 3579
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036582852431986813,
-      "loss": 4.1625,
+      "learning_rate": 3.654163231657049e-05,
+      "loss": 1.9499,
       "step": 3580
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003656224237427865,
-      "loss": 4.3837,
+      "learning_rate": 3.652102225886233e-05,
+      "loss": 2.3584,
       "step": 3581
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003654163231657049,
-      "loss": 4.1832,
+      "learning_rate": 3.6500412201154163e-05,
+      "loss": 2.0065,
       "step": 3582
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036521022258862324,
-      "loss": 4.1196,
+      "learning_rate": 3.6479802143446e-05,
+      "loss": 2.2329,
       "step": 3583
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036500412201154165,
-      "loss": 4.435,
+      "learning_rate": 3.6459192085737843e-05,
+      "loss": 2.7839,
       "step": 3584
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036479802143446,
-      "loss": 4.296,
+      "learning_rate": 3.643858202802968e-05,
+      "loss": 2.0239,
       "step": 3585
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003645919208573784,
-      "loss": 4.3019,
+      "learning_rate": 3.641797197032152e-05,
+      "loss": 1.9634,
       "step": 3586
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036438582028029676,
-      "loss": 4.3063,
+      "learning_rate": 3.639736191261336e-05,
+      "loss": 2.2681,
       "step": 3587
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036417971970321517,
-      "loss": 4.079,
+      "learning_rate": 3.63767518549052e-05,
+      "loss": 1.8198,
       "step": 3588
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003639736191261336,
-      "loss": 4.239,
+      "learning_rate": 3.635614179719703e-05,
+      "loss": 1.9621,
       "step": 3589
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003637675185490519,
-      "loss": 3.9906,
+      "learning_rate": 3.633553173948887e-05,
+      "loss": 2.1901,
       "step": 3590
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036356141797197033,
-      "loss": 4.4964,
+      "learning_rate": 3.631492168178071e-05,
+      "loss": 2.6732,
       "step": 3591
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003633553173948887,
-      "loss": 4.4445,
+      "learning_rate": 3.629431162407255e-05,
+      "loss": 2.3036,
       "step": 3592
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.0003631492168178071,
-      "loss": 4.166,
+      "learning_rate": 3.6273701566364386e-05,
+      "loss": 1.7702,
       "step": 3593
     },
     {
       "epoch": 2.01,
-      "learning_rate": 0.00036294311624072544,
-      "loss": 4.1784,
+      "learning_rate": 3.625309150865623e-05,
+      "loss": 1.7698,
       "step": 3594
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00036273701566364385,
-      "loss": 4.0959,
+      "learning_rate": 3.6232481450948066e-05,
+      "loss": 1.7313,
       "step": 3595
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00036253091508656226,
-      "loss": 4.2517,
+      "learning_rate": 3.62118713932399e-05,
+      "loss": 1.8319,
       "step": 3596
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003623248145094806,
-      "loss": 4.0979,
+      "learning_rate": 3.619126133553174e-05,
+      "loss": 1.6439,
       "step": 3597
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.000362118713932399,
-      "loss": 4.257,
+      "learning_rate": 3.617065127782358e-05,
+      "loss": 1.761,
       "step": 3598
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00036191261335531737,
-      "loss": 4.0202,
+      "learning_rate": 3.615004122011541e-05,
+      "loss": 1.6386,
       "step": 3599
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003617065127782358,
-      "loss": 4.2162,
+      "learning_rate": 3.6129431162407256e-05,
+      "loss": 1.9617,
       "step": 3600
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00036150041220115413,
-      "loss": 4.0788,
+      "learning_rate": 3.61088211046991e-05,
+      "loss": 1.606,
       "step": 3601
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00036129431162407253,
-      "loss": 4.0035,
+      "learning_rate": 3.6088211046990936e-05,
+      "loss": 1.5855,
       "step": 3602
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00036108821104699094,
-      "loss": 3.9209,
+      "learning_rate": 3.606760098928277e-05,
+      "loss": 1.7563,
       "step": 3603
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003608821104699093,
-      "loss": 4.0792,
+      "learning_rate": 3.604699093157461e-05,
+      "loss": 2.1119,
       "step": 3604
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003606760098928277,
-      "loss": 4.2221,
+      "learning_rate": 3.602638087386645e-05,
+      "loss": 1.9781,
       "step": 3605
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003604699093157461,
-      "loss": 4.2138,
+      "learning_rate": 3.600577081615828e-05,
+      "loss": 1.9485,
       "step": 3606
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003602638087386645,
-      "loss": 3.9808,
+      "learning_rate": 3.5985160758450126e-05,
+      "loss": 1.9687,
       "step": 3607
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00036005770816158287,
-      "loss": 4.286,
+      "learning_rate": 3.596455070074196e-05,
+      "loss": 1.7899,
       "step": 3608
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003598516075845013,
-      "loss": 4.5028,
+      "learning_rate": 3.59439406430338e-05,
+      "loss": 1.9039,
       "step": 3609
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.0003596455070074196,
-      "loss": 4.2353,
+      "learning_rate": 3.592333058532564e-05,
+      "loss": 1.388,
       "step": 3610
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00035943940643033803,
-      "loss": 3.8068,
+      "learning_rate": 3.590272052761748e-05,
+      "loss": 1.1476,
       "step": 3611
     },
     {
       "epoch": 2.02,
-      "learning_rate": 0.00035923330585325644,
-      "loss": 3.7772,
+      "learning_rate": 3.5882110469909316e-05,
+      "loss": 1.3324,
       "step": 3612
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003590272052761748,
-      "loss": 3.7642,
+      "learning_rate": 3.586150041220115e-05,
+      "loss": 1.0436,
       "step": 3613
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003588211046990932,
-      "loss": 3.945,
+      "learning_rate": 3.5840890354492996e-05,
+      "loss": 1.3885,
       "step": 3614
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00035861500412201155,
-      "loss": 3.143,
+      "learning_rate": 3.582028029678483e-05,
+      "loss": 0.9031,
       "step": 3615
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00035840890354492996,
-      "loss": 3.5583,
+      "learning_rate": 3.579967023907667e-05,
+      "loss": 1.3295,
       "step": 3616
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003582028029678483,
-      "loss": 3.5837,
+      "learning_rate": 3.577906018136851e-05,
+      "loss": 1.2231,
       "step": 3617
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003579967023907667,
-      "loss": 3.359,
+      "learning_rate": 3.575845012366035e-05,
+      "loss": 1.3857,
       "step": 3618
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003577906018136851,
-      "loss": 4.4598,
+      "learning_rate": 3.5737840065952186e-05,
+      "loss": 2.733,
       "step": 3619
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003575845012366035,
-      "loss": 4.1168,
+      "learning_rate": 3.571723000824402e-05,
+      "loss": 2.2372,
       "step": 3620
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003573784006595219,
-      "loss": 4.1852,
+      "learning_rate": 3.5696619950535866e-05,
+      "loss": 2.4001,
       "step": 3621
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00035717230008244024,
-      "loss": 4.433,
+      "learning_rate": 3.56760098928277e-05,
+      "loss": 2.3141,
       "step": 3622
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00035696619950535864,
-      "loss": 4.1676,
+      "learning_rate": 3.565539983511954e-05,
+      "loss": 1.8945,
       "step": 3623
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.000356760098928277,
-      "loss": 4.1231,
+      "learning_rate": 3.5634789777411375e-05,
+      "loss": 1.8482,
       "step": 3624
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003565539983511954,
-      "loss": 4.4675,
+      "learning_rate": 3.561417971970322e-05,
+      "loss": 2.6001,
       "step": 3625
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00035634789777411375,
-      "loss": 4.467,
+      "learning_rate": 3.5593569661995055e-05,
+      "loss": 2.2642,
       "step": 3626
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00035614179719703216,
-      "loss": 4.369,
+      "learning_rate": 3.557295960428689e-05,
+      "loss": 1.844,
       "step": 3627
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.00035593569661995057,
-      "loss": 4.6344,
+      "learning_rate": 3.5552349546578735e-05,
+      "loss": 2.7377,
       "step": 3628
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003557295960428689,
-      "loss": 4.3623,
+      "learning_rate": 3.553173948887057e-05,
+      "loss": 2.0737,
       "step": 3629
     },
     {
       "epoch": 2.03,
-      "learning_rate": 0.0003555234954657873,
-      "loss": 4.0455,
+      "learning_rate": 3.551112943116241e-05,
+      "loss": 2.1002,
       "step": 3630
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003553173948887057,
-      "loss": 4.0276,
+      "learning_rate": 3.5490519373454245e-05,
+      "loss": 1.7289,
       "step": 3631
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003551112943116241,
-      "loss": 4.1344,
+      "learning_rate": 3.546990931574609e-05,
+      "loss": 1.8915,
       "step": 3632
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035490519373454244,
-      "loss": 4.9253,
+      "learning_rate": 3.5449299258037925e-05,
+      "loss": 2.894,
       "step": 3633
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035469909315746084,
-      "loss": 4.1643,
+      "learning_rate": 3.542868920032976e-05,
+      "loss": 1.8074,
       "step": 3634
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035449299258037925,
-      "loss": 4.3861,
+      "learning_rate": 3.5408079142621605e-05,
+      "loss": 1.9272,
       "step": 3635
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003542868920032976,
-      "loss": 4.3731,
+      "learning_rate": 3.5387469084913435e-05,
+      "loss": 2.0177,
       "step": 3636
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.000354080791426216,
-      "loss": 4.3804,
+      "learning_rate": 3.536685902720528e-05,
+      "loss": 1.8291,
       "step": 3637
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035387469084913436,
-      "loss": 4.2411,
+      "learning_rate": 3.5346248969497115e-05,
+      "loss": 2.0499,
       "step": 3638
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035366859027205277,
-      "loss": 4.0534,
+      "learning_rate": 3.532563891178896e-05,
+      "loss": 1.7957,
       "step": 3639
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003534624896949711,
-      "loss": 4.1232,
+      "learning_rate": 3.530502885408079e-05,
+      "loss": 1.7133,
       "step": 3640
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035325638911788953,
-      "loss": 3.9911,
+      "learning_rate": 3.528441879637263e-05,
+      "loss": 1.5558,
       "step": 3641
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003530502885408079,
-      "loss": 3.9408,
+      "learning_rate": 3.5263808738664475e-05,
+      "loss": 1.8391,
       "step": 3642
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003528441879637263,
-      "loss": 4.178,
+      "learning_rate": 3.5243198680956305e-05,
+      "loss": 1.7067,
       "step": 3643
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003526380873866447,
-      "loss": 4.2293,
+      "learning_rate": 3.522258862324815e-05,
+      "loss": 2.0074,
       "step": 3644
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035243198680956305,
-      "loss": 4.1888,
+      "learning_rate": 3.5201978565539985e-05,
+      "loss": 2.3192,
       "step": 3645
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.00035222588623248145,
-      "loss": 4.1129,
+      "learning_rate": 3.518136850783182e-05,
+      "loss": 2.0332,
       "step": 3646
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003520197856553998,
-      "loss": 4.1418,
+      "learning_rate": 3.516075845012366e-05,
+      "loss": 1.6754,
       "step": 3647
     },
     {
       "epoch": 2.04,
-      "learning_rate": 0.0003518136850783182,
-      "loss": 3.975,
+      "learning_rate": 3.51401483924155e-05,
+      "loss": 1.4448,
       "step": 3648
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00035160758450123657,
-      "loss": 4.4257,
+      "learning_rate": 3.5119538334707345e-05,
+      "loss": 2.0632,
       "step": 3649
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00035140148392415497,
-      "loss": 3.9152,
+      "learning_rate": 3.5098928276999174e-05,
+      "loss": 1.2994,
       "step": 3650
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0003511953833470734,
-      "loss": 4.6934,
+      "learning_rate": 3.507831821929102e-05,
+      "loss": 2.3034,
       "step": 3651
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00035098928276999173,
-      "loss": 4.3512,
+      "learning_rate": 3.5057708161582854e-05,
+      "loss": 2.3247,
       "step": 3652
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0003507831821929102,
-      "loss": 4.2817,
+      "learning_rate": 3.503709810387469e-05,
+      "loss": 1.9066,
       "step": 3653
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00035057708161582854,
-      "loss": 3.7464,
+      "learning_rate": 3.501648804616653e-05,
+      "loss": 1.4853,
       "step": 3654
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00035037098103874695,
-      "loss": 3.9491,
+      "learning_rate": 3.499587798845837e-05,
+      "loss": 1.5061,
       "step": 3655
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0003501648804616653,
-      "loss": 4.0878,
+      "learning_rate": 3.497526793075021e-05,
+      "loss": 1.416,
       "step": 3656
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0003499587798845837,
-      "loss": 4.1872,
+      "learning_rate": 3.4954657873042044e-05,
+      "loss": 1.3931,
       "step": 3657
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00034975267930750206,
-      "loss": 3.8181,
+      "learning_rate": 3.493404781533389e-05,
+      "loss": 1.3671,
       "step": 3658
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00034954657873042047,
-      "loss": 3.9885,
+      "learning_rate": 3.4913437757625724e-05,
+      "loss": 1.4612,
       "step": 3659
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0003493404781533389,
-      "loss": 4.2296,
+      "learning_rate": 3.489282769991756e-05,
+      "loss": 1.5603,
       "step": 3660
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00034913437757625723,
-      "loss": 4.1406,
+      "learning_rate": 3.48722176422094e-05,
+      "loss": 1.6145,
       "step": 3661
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00034892827699917564,
-      "loss": 3.8549,
+      "learning_rate": 3.485160758450124e-05,
+      "loss": 1.2323,
       "step": 3662
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.000348722176422094,
-      "loss": 3.8693,
+      "learning_rate": 3.483099752679308e-05,
+      "loss": 1.2984,
       "step": 3663
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.0003485160758450124,
-      "loss": 3.7375,
+      "learning_rate": 3.4810387469084914e-05,
+      "loss": 1.0512,
       "step": 3664
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00034830997526793075,
-      "loss": 4.0972,
+      "learning_rate": 3.478977741137676e-05,
+      "loss": 1.0716,
       "step": 3665
     },
     {
       "epoch": 2.05,
-      "learning_rate": 0.00034810387469084915,
-      "loss": 3.646,
+      "learning_rate": 3.4769167353668594e-05,
+      "loss": 1.0318,
       "step": 3666
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00034789777411376756,
-      "loss": 3.8753,
+      "learning_rate": 3.474855729596043e-05,
+      "loss": 1.3118,
       "step": 3667
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003476916735366859,
-      "loss": 3.5197,
+      "learning_rate": 3.472794723825227e-05,
+      "loss": 1.2522,
       "step": 3668
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003474855729596043,
-      "loss": 4.305,
+      "learning_rate": 3.470733718054411e-05,
+      "loss": 2.0544,
       "step": 3669
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00034727947238252267,
-      "loss": 4.5513,
+      "learning_rate": 3.468672712283594e-05,
+      "loss": 2.3669,
       "step": 3670
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003470733718054411,
-      "loss": 4.0203,
+      "learning_rate": 3.4666117065127784e-05,
+      "loss": 1.7602,
       "step": 3671
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00034686727122835943,
-      "loss": 4.0658,
+      "learning_rate": 3.464550700741962e-05,
+      "loss": 1.6758,
       "step": 3672
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00034666117065127784,
-      "loss": 4.0953,
+      "learning_rate": 3.4624896949711464e-05,
+      "loss": 1.8647,
       "step": 3673
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003464550700741962,
-      "loss": 4.2018,
+      "learning_rate": 3.46042868920033e-05,
+      "loss": 2.2035,
       "step": 3674
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003462489694971146,
-      "loss": 4.8505,
+      "learning_rate": 3.458367683429514e-05,
+      "loss": 3.1765,
       "step": 3675
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.000346042868920033,
-      "loss": 4.1872,
+      "learning_rate": 3.456306677658698e-05,
+      "loss": 1.7157,
       "step": 3676
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00034583676834295136,
-      "loss": 4.406,
+      "learning_rate": 3.454245671887881e-05,
+      "loss": 1.8856,
       "step": 3677
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.00034563066776586976,
-      "loss": 4.1894,
+      "learning_rate": 3.4521846661170654e-05,
+      "loss": 2.32,
       "step": 3678
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003454245671887881,
-      "loss": 4.0363,
+      "learning_rate": 3.450123660346249e-05,
+      "loss": 1.7955,
       "step": 3679
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003452184666117065,
-      "loss": 4.238,
+      "learning_rate": 3.448062654575433e-05,
+      "loss": 2.0955,
       "step": 3680
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003450123660346249,
-      "loss": 4.2913,
+      "learning_rate": 3.446001648804617e-05,
+      "loss": 2.2383,
       "step": 3681
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003448062654575433,
-      "loss": 4.1583,
+      "learning_rate": 3.443940643033801e-05,
+      "loss": 1.7512,
       "step": 3682
     },
     {
       "epoch": 2.06,
-      "learning_rate": 0.0003446001648804617,
-      "loss": 4.2548,
+      "learning_rate": 3.441879637262985e-05,
+      "loss": 2.4249,
       "step": 3683
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034439406430338004,
-      "loss": 4.3861,
+      "learning_rate": 3.439818631492168e-05,
+      "loss": 2.6975,
       "step": 3684
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034418796372629845,
-      "loss": 4.3047,
+      "learning_rate": 3.437757625721352e-05,
+      "loss": 2.6307,
       "step": 3685
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003439818631492168,
-      "loss": 4.4504,
+      "learning_rate": 3.435696619950536e-05,
+      "loss": 2.0661,
       "step": 3686
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003437757625721352,
-      "loss": 4.3347,
+      "learning_rate": 3.4336356141797197e-05,
+      "loss": 2.3148,
       "step": 3687
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034356966199505356,
-      "loss": 4.3435,
+      "learning_rate": 3.431574608408903e-05,
+      "loss": 1.7212,
       "step": 3688
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034336356141797197,
-      "loss": 4.0876,
+      "learning_rate": 3.4295136026380877e-05,
+      "loss": 1.7815,
       "step": 3689
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003431574608408903,
-      "loss": 4.1959,
+      "learning_rate": 3.427452596867271e-05,
+      "loss": 1.9633,
       "step": 3690
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003429513602638087,
-      "loss": 4.0557,
+      "learning_rate": 3.425391591096455e-05,
+      "loss": 1.6484,
       "step": 3691
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034274525968672713,
-      "loss": 4.1933,
+      "learning_rate": 3.423330585325639e-05,
+      "loss": 1.8562,
       "step": 3692
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003425391591096455,
-      "loss": 3.9794,
+      "learning_rate": 3.421269579554823e-05,
+      "loss": 1.7794,
       "step": 3693
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003423330585325639,
-      "loss": 4.2046,
+      "learning_rate": 3.4192085737840066e-05,
+      "loss": 1.8231,
       "step": 3694
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034212695795548224,
-      "loss": 4.257,
+      "learning_rate": 3.41714756801319e-05,
+      "loss": 1.6328,
       "step": 3695
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034192085737840065,
-      "loss": 3.9712,
+      "learning_rate": 3.4150865622423746e-05,
+      "loss": 1.6304,
       "step": 3696
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.000341714756801319,
-      "loss": 4.3585,
+      "learning_rate": 3.413025556471558e-05,
+      "loss": 2.0494,
       "step": 3697
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003415086562242374,
-      "loss": 4.0659,
+      "learning_rate": 3.410964550700742e-05,
+      "loss": 1.4954,
       "step": 3698
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034130255564715587,
-      "loss": 4.0655,
+      "learning_rate": 3.408903544929926e-05,
+      "loss": 1.66,
       "step": 3699
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.0003410964550700742,
-      "loss": 4.2117,
+      "learning_rate": 3.40684253915911e-05,
+      "loss": 2.1312,
       "step": 3700
     },
     {
       "epoch": 2.07,
-      "learning_rate": 0.00034089035449299263,
-      "loss": 4.484,
+      "learning_rate": 3.4047815333882936e-05,
+      "loss": 2.1933,
       "step": 3701
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.000340684253915911,
-      "loss": 3.9193,
+      "learning_rate": 3.402720527617477e-05,
+      "loss": 1.5397,
       "step": 3702
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003404781533388294,
-      "loss": 4.0548,
+      "learning_rate": 3.4006595218466616e-05,
+      "loss": 1.4747,
       "step": 3703
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00034027205276174774,
-      "loss": 4.1632,
+      "learning_rate": 3.398598516075845e-05,
+      "loss": 1.5549,
       "step": 3704
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00034006595218466615,
-      "loss": 4.0142,
+      "learning_rate": 3.396537510305029e-05,
+      "loss": 1.3224,
       "step": 3705
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00033985985160758455,
-      "loss": 4.0628,
+      "learning_rate": 3.394476504534213e-05,
+      "loss": 1.6666,
       "step": 3706
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003396537510305029,
-      "loss": 4.2197,
+      "learning_rate": 3.392415498763397e-05,
+      "loss": 2.1122,
       "step": 3707
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003394476504534213,
-      "loss": 4.0214,
+      "learning_rate": 3.3903544929925806e-05,
+      "loss": 1.599,
       "step": 3708
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00033924154987633967,
-      "loss": 4.1325,
+      "learning_rate": 3.388293487221764e-05,
+      "loss": 1.4339,
       "step": 3709
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00033903544929925807,
-      "loss": 4.1346,
+      "learning_rate": 3.3862324814509486e-05,
+      "loss": 1.5029,
       "step": 3710
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003388293487221764,
-      "loss": 4.122,
+      "learning_rate": 3.3841714756801316e-05,
+      "loss": 1.4044,
       "step": 3711
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00033862324814509483,
-      "loss": 5.1289,
+      "learning_rate": 3.382110469909316e-05,
+      "loss": 2.3451,
       "step": 3712
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003384171475680132,
-      "loss": 3.9975,
+      "learning_rate": 3.3800494641385e-05,
+      "loss": 1.1671,
       "step": 3713
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003382110469909316,
-      "loss": 3.9169,
+      "learning_rate": 3.377988458367683e-05,
+      "loss": 1.1371,
       "step": 3714
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00033800494641385,
-      "loss": 4.2691,
+      "learning_rate": 3.3759274525968676e-05,
+      "loss": 1.565,
       "step": 3715
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00033779884583676835,
-      "loss": 3.5229,
+      "learning_rate": 3.373866446826051e-05,
+      "loss": 1.1395,
       "step": 3716
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.00033759274525968676,
-      "loss": 3.5943,
+      "learning_rate": 3.371805441055235e-05,
+      "loss": 1.0511,
       "step": 3717
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003373866446826051,
-      "loss": 3.3199,
+      "learning_rate": 3.3697444352844185e-05,
+      "loss": 1.2396,
       "step": 3718
     },
     {
       "epoch": 2.08,
-      "learning_rate": 0.0003371805441055235,
-      "loss": 4.5585,
+      "learning_rate": 3.367683429513603e-05,
+      "loss": 2.8596,
       "step": 3719
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00033697444352844187,
-      "loss": 4.6449,
+      "learning_rate": 3.365622423742787e-05,
+      "loss": 2.6995,
       "step": 3720
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003367683429513603,
-      "loss": 4.0208,
+      "learning_rate": 3.36356141797197e-05,
+      "loss": 1.7117,
       "step": 3721
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003365622423742787,
-      "loss": 4.1108,
+      "learning_rate": 3.3615004122011545e-05,
+      "loss": 2.1307,
       "step": 3722
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00033635614179719703,
-      "loss": 4.9957,
+      "learning_rate": 3.359439406430338e-05,
+      "loss": 3.109,
       "step": 3723
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00033615004122011544,
-      "loss": 4.1105,
+      "learning_rate": 3.357378400659522e-05,
+      "loss": 1.6505,
       "step": 3724
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003359439406430338,
-      "loss": 4.023,
+      "learning_rate": 3.3553173948887055e-05,
+      "loss": 2.4618,
       "step": 3725
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003357378400659522,
-      "loss": 4.2332,
+      "learning_rate": 3.35325638911789e-05,
+      "loss": 3.2794,
       "step": 3726
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00033553173948887055,
-      "loss": 4.1649,
+      "learning_rate": 3.3511953833470735e-05,
+      "loss": 2.3178,
       "step": 3727
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00033532563891178896,
-      "loss": 4.0826,
+      "learning_rate": 3.349134377576257e-05,
+      "loss": 1.8105,
       "step": 3728
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003351195383347073,
-      "loss": 4.3062,
+      "learning_rate": 3.3470733718054415e-05,
+      "loss": 1.9348,
       "step": 3729
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003349134377576257,
-      "loss": 4.1424,
+      "learning_rate": 3.345012366034625e-05,
+      "loss": 2.6311,
       "step": 3730
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003347073371805441,
-      "loss": 3.8809,
+      "learning_rate": 3.342951360263809e-05,
+      "loss": 1.639,
       "step": 3731
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003345012366034625,
-      "loss": 4.1399,
+      "learning_rate": 3.3408903544929925e-05,
+      "loss": 1.7201,
       "step": 3732
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003342951360263809,
-      "loss": 4.1754,
+      "learning_rate": 3.338829348722177e-05,
+      "loss": 1.7561,
       "step": 3733
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00033408903544929924,
-      "loss": 4.3943,
+      "learning_rate": 3.3367683429513605e-05,
+      "loss": 2.0078,
       "step": 3734
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.00033388293487221764,
-      "loss": 4.0475,
+      "learning_rate": 3.334707337180544e-05,
+      "loss": 2.0132,
       "step": 3735
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.000333676834295136,
-      "loss": 4.2241,
+      "learning_rate": 3.3326463314097285e-05,
+      "loss": 1.85,
       "step": 3736
     },
     {
       "epoch": 2.09,
-      "learning_rate": 0.0003334707337180544,
-      "loss": 4.1113,
+      "learning_rate": 3.330585325638912e-05,
+      "loss": 1.6569,
       "step": 3737
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003332646331409728,
-      "loss": 4.2762,
+      "learning_rate": 3.328524319868096e-05,
+      "loss": 1.8126,
       "step": 3738
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00033305853256389116,
-      "loss": 4.274,
+      "learning_rate": 3.3264633140972795e-05,
+      "loss": 1.9144,
       "step": 3739
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00033285243198680957,
-      "loss": 4.2894,
+      "learning_rate": 3.324402308326464e-05,
+      "loss": 1.678,
       "step": 3740
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003326463314097279,
-      "loss": 4.1549,
+      "learning_rate": 3.322341302555647e-05,
+      "loss": 1.9912,
       "step": 3741
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00033244023083264633,
-      "loss": 4.2425,
+      "learning_rate": 3.320280296784831e-05,
+      "loss": 1.7625,
       "step": 3742
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003322341302555647,
-      "loss": 4.1323,
+      "learning_rate": 3.318219291014015e-05,
+      "loss": 1.9813,
       "step": 3743
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003320280296784831,
-      "loss": 4.278,
+      "learning_rate": 3.316158285243199e-05,
+      "loss": 1.9955,
       "step": 3744
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00033182192910140144,
-      "loss": 4.2883,
+      "learning_rate": 3.314097279472383e-05,
+      "loss": 1.68,
       "step": 3745
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003316158285243199,
-      "loss": 4.3033,
+      "learning_rate": 3.3120362737015665e-05,
+      "loss": 1.6859,
       "step": 3746
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003314097279472383,
-      "loss": 3.8966,
+      "learning_rate": 3.309975267930751e-05,
+      "loss": 1.433,
       "step": 3747
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00033120362737015666,
-      "loss": 4.07,
+      "learning_rate": 3.307914262159934e-05,
+      "loss": 1.4129,
       "step": 3748
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00033099752679307507,
-      "loss": 4.0953,
+      "learning_rate": 3.305853256389118e-05,
+      "loss": 1.8077,
       "step": 3749
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003307914262159934,
-      "loss": 4.3498,
+      "learning_rate": 3.303792250618302e-05,
+      "loss": 1.9802,
       "step": 3750
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003305853256389118,
-      "loss": 4.2417,
+      "learning_rate": 3.3017312448474854e-05,
+      "loss": 1.7373,
       "step": 3751
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003303792250618302,
-      "loss": 4.1761,
+      "learning_rate": 3.29967023907667e-05,
+      "loss": 2.0471,
       "step": 3752
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.0003301731244847486,
-      "loss": 4.017,
+      "learning_rate": 3.2976092333058534e-05,
+      "loss": 1.916,
       "step": 3753
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.000329967023907667,
-      "loss": 3.9064,
+      "learning_rate": 3.295548227535038e-05,
+      "loss": 1.6662,
       "step": 3754
     },
     {
       "epoch": 2.1,
-      "learning_rate": 0.00032976092333058534,
-      "loss": 4.352,
+      "learning_rate": 3.293487221764221e-05,
+      "loss": 1.4737,
       "step": 3755
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032955482275350375,
-      "loss": 4.0681,
+      "learning_rate": 3.291426215993405e-05,
+      "loss": 2.0426,
       "step": 3756
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003293487221764221,
-      "loss": 4.1873,
+      "learning_rate": 3.289365210222589e-05,
+      "loss": 1.8901,
       "step": 3757
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003291426215993405,
-      "loss": 4.3458,
+      "learning_rate": 3.2873042044517724e-05,
+      "loss": 1.9637,
       "step": 3758
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032893652102225886,
-      "loss": 4.5754,
+      "learning_rate": 3.285243198680956e-05,
+      "loss": 1.8498,
       "step": 3759
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032873042044517727,
-      "loss": 4.1492,
+      "learning_rate": 3.2831821929101404e-05,
+      "loss": 1.8199,
       "step": 3760
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003285243198680956,
-      "loss": 3.9552,
+      "learning_rate": 3.281121187139324e-05,
+      "loss": 1.0573,
       "step": 3761
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032831821929101403,
-      "loss": 3.9674,
+      "learning_rate": 3.279060181368508e-05,
+      "loss": 1.2325,
       "step": 3762
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032811211871393243,
-      "loss": 3.8928,
+      "learning_rate": 3.276999175597692e-05,
+      "loss": 1.1963,
       "step": 3763
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003279060181368508,
-      "loss": 3.7404,
+      "learning_rate": 3.274938169826876e-05,
+      "loss": 1.2635,
       "step": 3764
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003276999175597692,
-      "loss": 3.686,
+      "learning_rate": 3.2728771640560594e-05,
+      "loss": 1.0214,
       "step": 3765
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032749381698268755,
-      "loss": 3.8057,
+      "learning_rate": 3.270816158285243e-05,
+      "loss": 1.0842,
       "step": 3766
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032728771640560595,
-      "loss": 4.0004,
+      "learning_rate": 3.2687551525144274e-05,
+      "loss": 1.2913,
       "step": 3767
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003270816158285243,
-      "loss": 3.5113,
+      "learning_rate": 3.266694146743611e-05,
+      "loss": 1.1382,
       "step": 3768
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003268755152514427,
-      "loss": 4.5058,
+      "learning_rate": 3.264633140972795e-05,
+      "loss": 2.3634,
       "step": 3769
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003266694146743611,
-      "loss": 4.0737,
+      "learning_rate": 3.262572135201979e-05,
+      "loss": 1.8256,
       "step": 3770
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032646331409727947,
-      "loss": 4.2785,
+      "learning_rate": 3.260511129431163e-05,
+      "loss": 1.8188,
       "step": 3771
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.0003262572135201979,
-      "loss": 4.4197,
+      "learning_rate": 3.2584501236603464e-05,
+      "loss": 2.1902,
       "step": 3772
     },
     {
       "epoch": 2.11,
-      "learning_rate": 0.00032605111294311623,
-      "loss": 4.1764,
+      "learning_rate": 3.25638911788953e-05,
+      "loss": 1.8946,
       "step": 3773
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032584501236603464,
-      "loss": 4.003,
+      "learning_rate": 3.2543281121187144e-05,
+      "loss": 1.7156,
       "step": 3774
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.000325638911788953,
-      "loss": 4.0184,
+      "learning_rate": 3.2522671063478974e-05,
+      "loss": 1.8395,
       "step": 3775
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0003254328112118714,
-      "loss": 4.2844,
+      "learning_rate": 3.250206100577082e-05,
+      "loss": 2.0671,
       "step": 3776
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032522671063478975,
-      "loss": 4.2929,
+      "learning_rate": 3.248145094806266e-05,
+      "loss": 1.7672,
       "step": 3777
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032502061005770816,
-      "loss": 4.3295,
+      "learning_rate": 3.24608408903545e-05,
+      "loss": 1.8625,
       "step": 3778
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032481450948062656,
-      "loss": 3.911,
+      "learning_rate": 3.2440230832646333e-05,
+      "loss": 1.4769,
       "step": 3779
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0003246084089035449,
-      "loss": 3.95,
+      "learning_rate": 3.241962077493817e-05,
+      "loss": 1.7081,
       "step": 3780
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0003244023083264633,
-      "loss": 4.0873,
+      "learning_rate": 3.2399010717230013e-05,
+      "loss": 1.6927,
       "step": 3781
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0003241962077493817,
-      "loss": 4.7243,
+      "learning_rate": 3.237840065952184e-05,
+      "loss": 2.8409,
       "step": 3782
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0003239901071723001,
-      "loss": 3.7764,
+      "learning_rate": 3.235779060181369e-05,
+      "loss": 1.8483,
       "step": 3783
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032378400659521843,
-      "loss": 4.0955,
+      "learning_rate": 3.233718054410553e-05,
+      "loss": 1.986,
       "step": 3784
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032357790601813684,
-      "loss": 4.3724,
+      "learning_rate": 3.231657048639736e-05,
+      "loss": 2.0396,
       "step": 3785
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032337180544105525,
-      "loss": 4.1733,
+      "learning_rate": 3.22959604286892e-05,
+      "loss": 1.9133,
       "step": 3786
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0003231657048639736,
-      "loss": 4.2521,
+      "learning_rate": 3.227535037098104e-05,
+      "loss": 1.7958,
       "step": 3787
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.000322959604286892,
-      "loss": 4.3336,
+      "learning_rate": 3.2254740313272876e-05,
+      "loss": 2.0908,
       "step": 3788
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032275350370981036,
-      "loss": 3.9601,
+      "learning_rate": 3.223413025556471e-05,
+      "loss": 1.5067,
       "step": 3789
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.00032254740313272876,
-      "loss": 4.5063,
+      "learning_rate": 3.2213520197856556e-05,
+      "loss": 2.203,
       "step": 3790
     },
     {
       "epoch": 2.12,
-      "learning_rate": 0.0003223413025556471,
-      "loss": 4.1958,
+      "learning_rate": 3.219291014014839e-05,
+      "loss": 1.7331,
       "step": 3791
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003221352019785655,
-      "loss": 4.6148,
+      "learning_rate": 3.217230008244023e-05,
+      "loss": 2.8389,
       "step": 3792
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00032192910140148393,
-      "loss": 4.2743,
+      "learning_rate": 3.215169002473207e-05,
+      "loss": 2.038,
       "step": 3793
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00032172300082440234,
-      "loss": 4.1043,
+      "learning_rate": 3.213107996702391e-05,
+      "loss": 1.671,
       "step": 3794
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00032151690024732074,
-      "loss": 3.9861,
+      "learning_rate": 3.2110469909315746e-05,
+      "loss": 1.4847,
       "step": 3795
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003213107996702391,
-      "loss": 4.5033,
+      "learning_rate": 3.208985985160758e-05,
+      "loss": 2.1759,
       "step": 3796
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003211046990931575,
-      "loss": 4.3829,
+      "learning_rate": 3.2069249793899426e-05,
+      "loss": 1.9684,
       "step": 3797
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00032089859851607586,
-      "loss": 4.158,
+      "learning_rate": 3.204863973619126e-05,
+      "loss": 2.2373,
       "step": 3798
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00032069249793899426,
-      "loss": 4.2529,
+      "learning_rate": 3.20280296784831e-05,
+      "loss": 1.7044,
       "step": 3799
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003204863973619126,
-      "loss": 4.2216,
+      "learning_rate": 3.200741962077494e-05,
+      "loss": 2.0709,
       "step": 3800
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.000320280296784831,
-      "loss": 3.9512,
+      "learning_rate": 3.198680956306678e-05,
+      "loss": 1.4353,
       "step": 3801
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00032007419620774943,
-      "loss": 4.2462,
+      "learning_rate": 3.1966199505358616e-05,
+      "loss": 1.5708,
       "step": 3802
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003198680956306678,
-      "loss": 4.0295,
+      "learning_rate": 3.194558944765045e-05,
+      "loss": 1.6038,
       "step": 3803
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003196619950535862,
-      "loss": 4.1878,
+      "learning_rate": 3.1924979389942296e-05,
+      "loss": 1.7494,
       "step": 3804
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00031945589447650454,
-      "loss": 4.0144,
+      "learning_rate": 3.190436933223413e-05,
+      "loss": 1.7746,
       "step": 3805
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.00031924979389942295,
-      "loss": 4.2387,
+      "learning_rate": 3.188375927452597e-05,
+      "loss": 2.2944,
       "step": 3806
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003190436933223413,
-      "loss": 3.769,
+      "learning_rate": 3.1863149216817806e-05,
+      "loss": 1.1918,
       "step": 3807
     },
     {
       "epoch": 2.13,
-      "learning_rate": 0.0003188375927452597,
-      "loss": 3.8161,
+      "learning_rate": 3.184253915910965e-05,
+      "loss": 1.2084,
       "step": 3808
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031863149216817806,
-      "loss": 3.7658,
+      "learning_rate": 3.1821929101401486e-05,
+      "loss": 1.1201,
       "step": 3809
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031842539159109646,
-      "loss": 3.9501,
+      "learning_rate": 3.180131904369332e-05,
+      "loss": 1.1452,
       "step": 3810
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031821929101401487,
-      "loss": 3.8618,
+      "learning_rate": 3.1780708985985166e-05,
+      "loss": 1.1911,
       "step": 3811
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0003180131904369332,
-      "loss": 3.3782,
+      "learning_rate": 3.1760098928276996e-05,
+      "loss": 1.0219,
       "step": 3812
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031780708985985163,
-      "loss": 4.4266,
+      "learning_rate": 3.173948887056884e-05,
+      "loss": 1.7457,
       "step": 3813
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031760098928277,
-      "loss": 3.6482,
+      "learning_rate": 3.1718878812860676e-05,
+      "loss": 1.0193,
       "step": 3814
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0003173948887056884,
-      "loss": 3.4938,
+      "learning_rate": 3.169826875515252e-05,
+      "loss": 0.9709,
       "step": 3815
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031718878812860674,
-      "loss": 3.5221,
+      "learning_rate": 3.1677658697444356e-05,
+      "loss": 1.3172,
       "step": 3816
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031698268755152515,
-      "loss": 3.2281,
+      "learning_rate": 3.165704863973619e-05,
+      "loss": 1.0766,
       "step": 3817
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031677658697444356,
-      "loss": 3.5781,
+      "learning_rate": 3.1636438582028036e-05,
+      "loss": 1.5233,
       "step": 3818
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0003165704863973619,
-      "loss": 4.3927,
+      "learning_rate": 3.1615828524319865e-05,
+      "loss": 2.3424,
       "step": 3819
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0003163643858202803,
-      "loss": 4.317,
+      "learning_rate": 3.159521846661171e-05,
+      "loss": 1.941,
       "step": 3820
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031615828524319867,
-      "loss": 4.4545,
+      "learning_rate": 3.1574608408903545e-05,
+      "loss": 2.5411,
       "step": 3821
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0003159521846661171,
-      "loss": 4.1876,
+      "learning_rate": 3.155399835119538e-05,
+      "loss": 1.7795,
       "step": 3822
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0003157460840890354,
-      "loss": 4.4242,
+      "learning_rate": 3.1533388293487225e-05,
+      "loss": 1.8743,
       "step": 3823
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031553998351195383,
-      "loss": 4.2528,
+      "learning_rate": 3.151277823577906e-05,
+      "loss": 2.6759,
       "step": 3824
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.00031533388293487224,
-      "loss": 4.3391,
+      "learning_rate": 3.1492168178070905e-05,
+      "loss": 2.1207,
       "step": 3825
     },
     {
       "epoch": 2.14,
-      "learning_rate": 0.0003151277823577906,
-      "loss": 4.0592,
+      "learning_rate": 3.1471558120362735e-05,
+      "loss": 1.712,
       "step": 3826
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.000314921681780709,
-      "loss": 4.736,
+      "learning_rate": 3.145094806265458e-05,
+      "loss": 3.0567,
       "step": 3827
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00031471558120362735,
-      "loss": 4.1938,
+      "learning_rate": 3.1430338004946415e-05,
+      "loss": 2.1058,
       "step": 3828
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00031450948062654576,
-      "loss": 4.2883,
+      "learning_rate": 3.140972794723825e-05,
+      "loss": 2.0925,
       "step": 3829
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003143033800494641,
-      "loss": 4.1106,
+      "learning_rate": 3.138911788953009e-05,
+      "loss": 1.7119,
       "step": 3830
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003140972794723825,
-      "loss": 4.4614,
+      "learning_rate": 3.136850783182193e-05,
+      "loss": 2.3943,
       "step": 3831
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00031389117889530087,
-      "loss": 3.9693,
+      "learning_rate": 3.134789777411377e-05,
+      "loss": 1.7712,
       "step": 3832
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003136850783182193,
-      "loss": 4.0741,
+      "learning_rate": 3.1327287716405605e-05,
+      "loss": 1.917,
       "step": 3833
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003134789777411377,
-      "loss": 4.1548,
+      "learning_rate": 3.130667765869745e-05,
+      "loss": 2.4659,
       "step": 3834
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00031327287716405604,
-      "loss": 4.186,
+      "learning_rate": 3.1286067600989285e-05,
+      "loss": 1.8953,
       "step": 3835
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00031306677658697444,
-      "loss": 4.0764,
+      "learning_rate": 3.126545754328112e-05,
+      "loss": 1.6294,
       "step": 3836
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003128606760098928,
-      "loss": 4.0459,
+      "learning_rate": 3.124484748557296e-05,
+      "loss": 1.6442,
       "step": 3837
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003126545754328112,
-      "loss": 4.1743,
+      "learning_rate": 3.12242374278648e-05,
+      "loss": 1.6782,
       "step": 3838
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00031244847485572955,
-      "loss": 3.9454,
+      "learning_rate": 3.120362737015664e-05,
+      "loss": 1.5222,
       "step": 3839
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.000312242374278648,
-      "loss": 4.226,
+      "learning_rate": 3.1183017312448475e-05,
+      "loss": 2.2116,
       "step": 3840
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003120362737015664,
-      "loss": 4.0656,
+      "learning_rate": 3.116240725474032e-05,
+      "loss": 1.8138,
       "step": 3841
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003118301731244848,
-      "loss": 4.1275,
+      "learning_rate": 3.1141797197032155e-05,
+      "loss": 1.6519,
       "step": 3842
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.0003116240725474032,
-      "loss": 4.1272,
+      "learning_rate": 3.112118713932399e-05,
+      "loss": 1.5285,
       "step": 3843
     },
     {
       "epoch": 2.15,
-      "learning_rate": 0.00031141797197032153,
-      "loss": 4.0491,
+      "learning_rate": 3.110057708161583e-05,
+      "loss": 1.522,
       "step": 3844
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00031121187139323994,
-      "loss": 3.8922,
+      "learning_rate": 3.107996702390767e-05,
+      "loss": 1.8269,
       "step": 3845
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003110057708161583,
-      "loss": 4.1425,
+      "learning_rate": 3.10593569661995e-05,
+      "loss": 1.6728,
       "step": 3846
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003107996702390767,
-      "loss": 4.2908,
+      "learning_rate": 3.1038746908491344e-05,
+      "loss": 1.6514,
       "step": 3847
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00031059356966199505,
-      "loss": 4.4668,
+      "learning_rate": 3.101813685078319e-05,
+      "loss": 2.006,
       "step": 3848
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00031038746908491346,
-      "loss": 3.9677,
+      "learning_rate": 3.0997526793075024e-05,
+      "loss": 1.7016,
       "step": 3849
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00031018136850783186,
-      "loss": 3.9712,
+      "learning_rate": 3.097691673536686e-05,
+      "loss": 1.4469,
       "step": 3850
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003099752679307502,
-      "loss": 3.9558,
+      "learning_rate": 3.09563066776587e-05,
+      "loss": 1.333,
       "step": 3851
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003097691673536686,
-      "loss": 4.0367,
+      "learning_rate": 3.093569661995054e-05,
+      "loss": 1.4678,
       "step": 3852
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.000309563066776587,
-      "loss": 3.8838,
+      "learning_rate": 3.091508656224237e-05,
+      "loss": 1.3634,
       "step": 3853
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003093569661995054,
-      "loss": 3.8805,
+      "learning_rate": 3.0894476504534214e-05,
+      "loss": 1.3404,
       "step": 3854
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00030915086562242374,
-      "loss": 4.2344,
+      "learning_rate": 3.087386644682606e-05,
+      "loss": 1.7788,
       "step": 3855
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00030894476504534214,
-      "loss": 3.9473,
+      "learning_rate": 3.085325638911789e-05,
+      "loss": 1.296,
       "step": 3856
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00030873866446826055,
-      "loss": 4.0824,
+      "learning_rate": 3.083264633140973e-05,
+      "loss": 1.4134,
       "step": 3857
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003085325638911789,
-      "loss": 3.9773,
+      "learning_rate": 3.081203627370157e-05,
+      "loss": 1.426,
       "step": 3858
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003083264633140973,
-      "loss": 4.2702,
+      "learning_rate": 3.079142621599341e-05,
+      "loss": 1.9104,
       "step": 3859
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00030812036273701566,
-      "loss": 4.087,
+      "learning_rate": 3.077081615828524e-05,
+      "loss": 1.6123,
       "step": 3860
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.00030791426215993407,
-      "loss": 3.8625,
+      "learning_rate": 3.0750206100577084e-05,
+      "loss": 1.2552,
       "step": 3861
     },
     {
       "epoch": 2.16,
-      "learning_rate": 0.0003077081615828524,
-      "loss": 3.2908,
+      "learning_rate": 3.072959604286892e-05,
+      "loss": 0.8694,
       "step": 3862
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003075020610057708,
-      "loss": 3.4741,
+      "learning_rate": 3.070898598516076e-05,
+      "loss": 0.9311,
       "step": 3863
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003072959604286892,
-      "loss": 3.8052,
+      "learning_rate": 3.06883759274526e-05,
+      "loss": 1.1078,
       "step": 3864
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003070898598516076,
-      "loss": 3.6967,
+      "learning_rate": 3.066776586974444e-05,
+      "loss": 1.3128,
       "step": 3865
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.000306883759274526,
-      "loss": 3.5192,
+      "learning_rate": 3.0647155812036274e-05,
+      "loss": 0.9829,
       "step": 3866
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030667765869744435,
-      "loss": 4.0731,
+      "learning_rate": 3.062654575432811e-05,
+      "loss": 1.1359,
       "step": 3867
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030647155812036275,
-      "loss": 3.9215,
+      "learning_rate": 3.0605935696619954e-05,
+      "loss": 1.5057,
       "step": 3868
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003062654575432811,
-      "loss": 4.4112,
+      "learning_rate": 3.058532563891179e-05,
+      "loss": 2.0319,
       "step": 3869
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003060593569661995,
-      "loss": 4.559,
+      "learning_rate": 3.056471558120363e-05,
+      "loss": 2.6168,
       "step": 3870
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030585325638911786,
-      "loss": 4.2582,
+      "learning_rate": 3.054410552349547e-05,
+      "loss": 1.8395,
       "step": 3871
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030564715581203627,
-      "loss": 4.3368,
+      "learning_rate": 3.052349546578731e-05,
+      "loss": 2.1881,
       "step": 3872
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003054410552349547,
-      "loss": 4.2757,
+      "learning_rate": 3.0502885408079147e-05,
+      "loss": 1.7884,
       "step": 3873
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030523495465787303,
-      "loss": 4.3322,
+      "learning_rate": 3.048227535037098e-05,
+      "loss": 1.8681,
       "step": 3874
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030502885408079144,
-      "loss": 4.2184,
+      "learning_rate": 3.0461665292662824e-05,
+      "loss": 1.859,
       "step": 3875
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003048227535037098,
-      "loss": 4.3487,
+      "learning_rate": 3.0441055234954657e-05,
+      "loss": 2.0176,
       "step": 3876
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003046166529266282,
-      "loss": 4.1783,
+      "learning_rate": 3.0420445177246497e-05,
+      "loss": 1.764,
       "step": 3877
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030441055234954655,
-      "loss": 4.0487,
+      "learning_rate": 3.0399835119538333e-05,
+      "loss": 1.5285,
       "step": 3878
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.00030420445177246495,
-      "loss": 4.2616,
+      "learning_rate": 3.0379225061830173e-05,
+      "loss": 1.9944,
       "step": 3879
     },
     {
       "epoch": 2.17,
-      "learning_rate": 0.0003039983511953833,
-      "loss": 4.5808,
+      "learning_rate": 3.0358615004122017e-05,
+      "loss": 2.2367,
       "step": 3880
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003037922506183017,
-      "loss": 4.0817,
+      "learning_rate": 3.033800494641385e-05,
+      "loss": 2.241,
       "step": 3881
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003035861500412201,
-      "loss": 4.2913,
+      "learning_rate": 3.031739488870569e-05,
+      "loss": 2.0711,
       "step": 3882
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030338004946413847,
-      "loss": 4.2897,
+      "learning_rate": 3.0296784830997527e-05,
+      "loss": 1.9515,
       "step": 3883
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003031739488870569,
-      "loss": 4.3845,
+      "learning_rate": 3.0276174773289367e-05,
+      "loss": 1.8353,
       "step": 3884
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030296784830997523,
-      "loss": 4.0265,
+      "learning_rate": 3.0255564715581203e-05,
+      "loss": 1.7561,
       "step": 3885
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003027617477328937,
-      "loss": 4.0364,
+      "learning_rate": 3.0234954657873043e-05,
+      "loss": 1.6324,
       "step": 3886
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030255564715581205,
-      "loss": 4.1108,
+      "learning_rate": 3.0214344600164883e-05,
+      "loss": 2.0107,
       "step": 3887
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030234954657873045,
-      "loss": 4.1764,
+      "learning_rate": 3.019373454245672e-05,
+      "loss": 1.7178,
       "step": 3888
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030214344600164886,
-      "loss": 3.988,
+      "learning_rate": 3.017312448474856e-05,
+      "loss": 2.0687,
       "step": 3889
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003019373454245672,
-      "loss": 4.2859,
+      "learning_rate": 3.0152514427040396e-05,
+      "loss": 1.952,
       "step": 3890
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003017312448474856,
-      "loss": 3.9113,
+      "learning_rate": 3.0131904369332236e-05,
+      "loss": 1.5619,
       "step": 3891
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030152514427040397,
-      "loss": 4.0745,
+      "learning_rate": 3.0111294311624073e-05,
+      "loss": 1.5323,
       "step": 3892
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003013190436933224,
-      "loss": 4.0313,
+      "learning_rate": 3.0090684253915913e-05,
+      "loss": 1.5943,
       "step": 3893
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030111294311624073,
-      "loss": 4.1866,
+      "learning_rate": 3.007007419620775e-05,
+      "loss": 2.088,
       "step": 3894
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.00030090684253915914,
-      "loss": 3.8588,
+      "learning_rate": 3.004946413849959e-05,
+      "loss": 1.4934,
       "step": 3895
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003007007419620775,
-      "loss": 4.1562,
+      "learning_rate": 3.002885408079143e-05,
+      "loss": 1.9309,
       "step": 3896
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003004946413849959,
-      "loss": 4.1709,
+      "learning_rate": 3.0008244023083266e-05,
+      "loss": 2.5394,
       "step": 3897
     },
     {
       "epoch": 2.18,
-      "learning_rate": 0.0003002885408079143,
-      "loss": 4.0835,
+      "learning_rate": 2.9987633965375106e-05,
+      "loss": 1.5625,
       "step": 3898
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00030008244023083265,
-      "loss": 3.7711,
+      "learning_rate": 2.9967023907666943e-05,
+      "loss": 1.6003,
       "step": 3899
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029987633965375106,
-      "loss": 3.9256,
+      "learning_rate": 2.9946413849958783e-05,
+      "loss": 1.3802,
       "step": 3900
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002996702390766694,
-      "loss": 3.9953,
+      "learning_rate": 2.9925803792250616e-05,
+      "loss": 1.5009,
       "step": 3901
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002994641384995878,
-      "loss": 4.0807,
+      "learning_rate": 2.990519373454246e-05,
+      "loss": 1.9854,
       "step": 3902
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029925803792250617,
-      "loss": 4.0887,
+      "learning_rate": 2.98845836768343e-05,
+      "loss": 1.8073,
       "step": 3903
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002990519373454246,
-      "loss": 4.3249,
+      "learning_rate": 2.9863973619126136e-05,
+      "loss": 2.4037,
       "step": 3904
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.000298845836768343,
-      "loss": 4.0313,
+      "learning_rate": 2.9843363561417976e-05,
+      "loss": 1.6989,
       "step": 3905
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029863973619126134,
-      "loss": 4.0025,
+      "learning_rate": 2.982275350370981e-05,
+      "loss": 1.3917,
       "step": 3906
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029843363561417975,
-      "loss": 4.0359,
+      "learning_rate": 2.9802143446001652e-05,
+      "loss": 1.8603,
       "step": 3907
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002982275350370981,
-      "loss": 4.0512,
+      "learning_rate": 2.9781533388293486e-05,
+      "loss": 1.5243,
       "step": 3908
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002980214344600165,
-      "loss": 4.3657,
+      "learning_rate": 2.976092333058533e-05,
+      "loss": 1.3851,
       "step": 3909
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029781533388293486,
-      "loss": 4.6396,
+      "learning_rate": 2.9740313272877162e-05,
+      "loss": 2.0409,
       "step": 3910
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029760923330585326,
-      "loss": 4.1532,
+      "learning_rate": 2.9719703215169002e-05,
+      "loss": 1.9229,
       "step": 3911
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002974031327287716,
-      "loss": 3.8258,
+      "learning_rate": 2.9699093157460846e-05,
+      "loss": 1.4231,
       "step": 3912
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029719703215169,
-      "loss": 3.8341,
+      "learning_rate": 2.967848309975268e-05,
+      "loss": 1.3001,
       "step": 3913
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.00029699093157460843,
-      "loss": 3.8641,
+      "learning_rate": 2.965787304204452e-05,
+      "loss": 1.0457,
       "step": 3914
     },
     {
       "epoch": 2.19,
-      "learning_rate": 0.0002967848309975268,
-      "loss": 3.8049,
+      "learning_rate": 2.9637262984336355e-05,
+      "loss": 1.1153,
       "step": 3915
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002965787304204452,
-      "loss": 3.3339,
+      "learning_rate": 2.9616652926628195e-05,
+      "loss": 1.01,
       "step": 3916
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029637262984336354,
-      "loss": 3.7607,
+      "learning_rate": 2.9596042868920032e-05,
+      "loss": 1.0339,
       "step": 3917
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029616652926628195,
-      "loss": 3.3526,
+      "learning_rate": 2.9575432811211872e-05,
+      "loss": 1.277,
       "step": 3918
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002959604286892003,
-      "loss": 4.6614,
+      "learning_rate": 2.9554822753503712e-05,
+      "loss": 2.5749,
       "step": 3919
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002957543281121187,
-      "loss": 4.3389,
+      "learning_rate": 2.953421269579555e-05,
+      "loss": 2.1099,
       "step": 3920
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002955482275350371,
-      "loss": 4.0561,
+      "learning_rate": 2.951360263808739e-05,
+      "loss": 2.2838,
       "step": 3921
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029534212695795547,
-      "loss": 4.0836,
+      "learning_rate": 2.9492992580379225e-05,
+      "loss": 1.9575,
       "step": 3922
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002951360263808739,
-      "loss": 4.4105,
+      "learning_rate": 2.9472382522671065e-05,
+      "loss": 2.0117,
       "step": 3923
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002949299258037922,
-      "loss": 4.1422,
+      "learning_rate": 2.9451772464962902e-05,
+      "loss": 1.9166,
       "step": 3924
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029472382522671063,
-      "loss": 4.1495,
+      "learning_rate": 2.9431162407254742e-05,
+      "loss": 2.0831,
       "step": 3925
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.000294517724649629,
-      "loss": 4.1073,
+      "learning_rate": 2.9410552349546582e-05,
+      "loss": 1.6486,
       "step": 3926
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002943116240725474,
-      "loss": 4.1018,
+      "learning_rate": 2.938994229183842e-05,
+      "loss": 1.7729,
       "step": 3927
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002941055234954658,
-      "loss": 3.9543,
+      "learning_rate": 2.936933223413026e-05,
+      "loss": 1.6077,
       "step": 3928
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029389942291838415,
-      "loss": 4.2391,
+      "learning_rate": 2.9348722176422095e-05,
+      "loss": 2.0974,
       "step": 3929
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029369332234130256,
-      "loss": 4.7043,
+      "learning_rate": 2.9328112118713935e-05,
+      "loss": 3.1629,
       "step": 3930
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002934872217642209,
-      "loss": 4.084,
+      "learning_rate": 2.930750206100577e-05,
+      "loss": 1.6192,
       "step": 3931
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.00029328112118713937,
-      "loss": 4.185,
+      "learning_rate": 2.928689200329761e-05,
+      "loss": 1.9541,
       "step": 3932
     },
     {
       "epoch": 2.2,
-      "learning_rate": 0.0002930750206100577,
-      "loss": 4.1114,
+      "learning_rate": 2.9266281945589448e-05,
+      "loss": 1.6713,
       "step": 3933
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029286892003297613,
-      "loss": 4.0511,
+      "learning_rate": 2.9245671887881288e-05,
+      "loss": 1.599,
       "step": 3934
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002926628194558945,
-      "loss": 4.2042,
+      "learning_rate": 2.9225061830173128e-05,
+      "loss": 1.8941,
       "step": 3935
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002924567188788129,
-      "loss": 4.0384,
+      "learning_rate": 2.9204451772464965e-05,
+      "loss": 1.7227,
       "step": 3936
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002922506183017313,
-      "loss": 4.2764,
+      "learning_rate": 2.9183841714756805e-05,
+      "loss": 1.8317,
       "step": 3937
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029204451772464965,
-      "loss": 4.192,
+      "learning_rate": 2.9163231657048638e-05,
+      "loss": 1.6387,
       "step": 3938
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029183841714756805,
-      "loss": 4.1381,
+      "learning_rate": 2.914262159934048e-05,
+      "loss": 2.067,
       "step": 3939
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002916323165704864,
-      "loss": 3.9465,
+      "learning_rate": 2.9122011541632315e-05,
+      "loss": 1.6806,
       "step": 3940
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002914262159934048,
-      "loss": 4.0165,
+      "learning_rate": 2.9101401483924158e-05,
+      "loss": 1.8944,
       "step": 3941
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029122011541632317,
-      "loss": 3.8843,
+      "learning_rate": 2.9080791426215998e-05,
+      "loss": 1.706,
       "step": 3942
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002910140148392416,
-      "loss": 4.3323,
+      "learning_rate": 2.906018136850783e-05,
+      "loss": 1.9573,
       "step": 3943
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029080791426216,
-      "loss": 4.069,
+      "learning_rate": 2.9039571310799675e-05,
+      "loss": 1.5217,
       "step": 3944
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029060181368507833,
-      "loss": 4.1575,
+      "learning_rate": 2.9018961253091508e-05,
+      "loss": 1.6721,
       "step": 3945
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00029039571310799674,
-      "loss": 4.3389,
+      "learning_rate": 2.899835119538335e-05,
+      "loss": 2.4399,
       "step": 3946
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002901896125309151,
-      "loss": 4.178,
+      "learning_rate": 2.8977741137675184e-05,
+      "loss": 1.5293,
       "step": 3947
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002899835119538335,
-      "loss": 4.3657,
+      "learning_rate": 2.8957131079967024e-05,
+      "loss": 2.3684,
       "step": 3948
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00028977741137675185,
-      "loss": 4.0921,
+      "learning_rate": 2.893652102225886e-05,
+      "loss": 1.705,
       "step": 3949
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.00028957131079967026,
-      "loss": 3.72,
+      "learning_rate": 2.89159109645507e-05,
+      "loss": 1.3657,
       "step": 3950
     },
     {
       "epoch": 2.21,
-      "learning_rate": 0.0002893652102225886,
-      "loss": 4.0078,
+      "learning_rate": 2.8895300906842544e-05,
+      "loss": 2.2916,
       "step": 3951
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.000289159109645507,
-      "loss": 4.3387,
+      "learning_rate": 2.8874690849134378e-05,
+      "loss": 2.7511,
       "step": 3952
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002889530090684254,
-      "loss": 3.9529,
+      "learning_rate": 2.8854080791426218e-05,
+      "loss": 1.6401,
       "step": 3953
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002887469084913438,
-      "loss": 3.7621,
+      "learning_rate": 2.8833470733718054e-05,
+      "loss": 1.3772,
       "step": 3954
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002885408079142622,
-      "loss": 3.802,
+      "learning_rate": 2.8812860676009894e-05,
+      "loss": 1.2958,
       "step": 3955
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00028833470733718053,
-      "loss": 4.0281,
+      "learning_rate": 2.879225061830173e-05,
+      "loss": 1.7652,
       "step": 3956
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00028812860676009894,
-      "loss": 3.9139,
+      "learning_rate": 2.877164056059357e-05,
+      "loss": 1.3648,
       "step": 3957
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002879225061830173,
-      "loss": 3.8896,
+      "learning_rate": 2.875103050288541e-05,
+      "loss": 1.2371,
       "step": 3958
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002877164056059357,
-      "loss": 3.8029,
+      "learning_rate": 2.8730420445177247e-05,
+      "loss": 1.2343,
       "step": 3959
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002875103050288541,
-      "loss": 3.9813,
+      "learning_rate": 2.8709810387469087e-05,
+      "loss": 1.1683,
       "step": 3960
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00028730420445177246,
-      "loss": 3.7209,
+      "learning_rate": 2.8689200329760924e-05,
+      "loss": 1.1994,
       "step": 3961
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00028709810387469087,
-      "loss": 3.9258,
+      "learning_rate": 2.8668590272052764e-05,
+      "loss": 1.2619,
       "step": 3962
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002868920032976092,
-      "loss": 4.4414,
+      "learning_rate": 2.86479802143446e-05,
+      "loss": 1.6221,
       "step": 3963
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002866859027205276,
-      "loss": 3.9385,
+      "learning_rate": 2.862737015663644e-05,
+      "loss": 1.2201,
       "step": 3964
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.000286479802143446,
-      "loss": 3.5593,
+      "learning_rate": 2.8606760098928277e-05,
+      "loss": 0.9092,
       "step": 3965
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.0002862737015663644,
-      "loss": 3.8846,
+      "learning_rate": 2.8586150041220117e-05,
+      "loss": 1.0663,
       "step": 3966
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00028606760098928274,
-      "loss": 4.0206,
+      "learning_rate": 2.8565539983511957e-05,
+      "loss": 1.149,
       "step": 3967
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00028586150041220114,
-      "loss": 3.4617,
+      "learning_rate": 2.8544929925803794e-05,
+      "loss": 1.4353,
       "step": 3968
     },
     {
       "epoch": 2.22,
-      "learning_rate": 0.00028565539983511955,
-      "loss": 4.4926,
+      "learning_rate": 2.8524319868095634e-05,
+      "loss": 2.3239,
       "step": 3969
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002854492992580379,
-      "loss": 4.3168,
+      "learning_rate": 2.850370981038747e-05,
+      "loss": 2.5191,
       "step": 3970
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002852431986809563,
-      "loss": 4.428,
+      "learning_rate": 2.848309975267931e-05,
+      "loss": 2.4498,
       "step": 3971
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00028503709810387466,
-      "loss": 4.0235,
+      "learning_rate": 2.8462489694971144e-05,
+      "loss": 2.3238,
       "step": 3972
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00028483099752679307,
-      "loss": 4.6313,
+      "learning_rate": 2.8441879637262987e-05,
+      "loss": 3.357,
       "step": 3973
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002846248969497114,
-      "loss": 4.0336,
+      "learning_rate": 2.8421269579554827e-05,
+      "loss": 1.982,
       "step": 3974
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00028441879637262983,
-      "loss": 4.27,
+      "learning_rate": 2.8400659521846663e-05,
+      "loss": 1.9772,
       "step": 3975
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00028421269579554823,
-      "loss": 4.343,
+      "learning_rate": 2.8380049464138503e-05,
+      "loss": 2.0625,
       "step": 3976
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002840065952184666,
-      "loss": 4.0177,
+      "learning_rate": 2.8359439406430337e-05,
+      "loss": 2.5513,
       "step": 3977
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.000283800494641385,
-      "loss": 4.0206,
+      "learning_rate": 2.833882934872218e-05,
+      "loss": 1.9177,
       "step": 3978
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002835943940643034,
-      "loss": 4.4611,
+      "learning_rate": 2.8318219291014013e-05,
+      "loss": 2.4832,
       "step": 3979
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002833882934872218,
-      "loss": 4.2425,
+      "learning_rate": 2.8297609233305857e-05,
+      "loss": 2.2579,
       "step": 3980
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00028318219291014016,
-      "loss": 3.8993,
+      "learning_rate": 2.827699917559769e-05,
+      "loss": 1.9215,
       "step": 3981
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00028297609233305857,
-      "loss": 4.0273,
+      "learning_rate": 2.825638911788953e-05,
+      "loss": 1.9381,
       "step": 3982
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002827699917559769,
-      "loss": 4.1378,
+      "learning_rate": 2.8235779060181373e-05,
+      "loss": 2.2498,
       "step": 3983
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002825638911788953,
-      "loss": 4.0979,
+      "learning_rate": 2.8215169002473206e-05,
+      "loss": 2.24,
       "step": 3984
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.00028235779060181373,
-      "loss": 4.19,
+      "learning_rate": 2.819455894476505e-05,
+      "loss": 1.7442,
       "step": 3985
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002821516900247321,
-      "loss": 4.0824,
+      "learning_rate": 2.8173948887056883e-05,
+      "loss": 2.3828,
       "step": 3986
     },
     {
       "epoch": 2.23,
-      "learning_rate": 0.0002819455894476505,
-      "loss": 4.1026,
+      "learning_rate": 2.8153338829348723e-05,
+      "loss": 1.8797,
       "step": 3987
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00028173948887056884,
-      "loss": 4.1953,
+      "learning_rate": 2.813272877164056e-05,
+      "loss": 1.6651,
       "step": 3988
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00028153338829348725,
-      "loss": 3.9746,
+      "learning_rate": 2.81121187139324e-05,
+      "loss": 1.5834,
       "step": 3989
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002813272877164056,
-      "loss": 4.3866,
+      "learning_rate": 2.809150865622424e-05,
+      "loss": 1.9786,
       "step": 3990
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.000281121187139324,
-      "loss": 4.1055,
+      "learning_rate": 2.8070898598516076e-05,
+      "loss": 1.7601,
       "step": 3991
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002809150865622424,
-      "loss": 4.2328,
+      "learning_rate": 2.8050288540807916e-05,
+      "loss": 2.0586,
       "step": 3992
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00028070898598516077,
-      "loss": 4.0071,
+      "learning_rate": 2.8029678483099753e-05,
+      "loss": 1.6999,
       "step": 3993
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002805028854080792,
-      "loss": 4.1887,
+      "learning_rate": 2.8009068425391593e-05,
+      "loss": 1.8167,
       "step": 3994
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00028029678483099753,
-      "loss": 4.26,
+      "learning_rate": 2.798845836768343e-05,
+      "loss": 1.6345,
       "step": 3995
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00028009068425391594,
-      "loss": 3.9181,
+      "learning_rate": 2.796784830997527e-05,
+      "loss": 1.4678,
       "step": 3996
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002798845836768343,
-      "loss": 4.6314,
+      "learning_rate": 2.7947238252267106e-05,
+      "loss": 2.8052,
       "step": 3997
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002796784830997527,
-      "loss": 4.187,
+      "learning_rate": 2.7926628194558946e-05,
+      "loss": 1.7185,
       "step": 3998
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00027947238252267105,
-      "loss": 4.0675,
+      "learning_rate": 2.7906018136850786e-05,
+      "loss": 1.5869,
       "step": 3999
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00027926628194558945,
-      "loss": 4.4488,
+      "learning_rate": 2.7885408079142623e-05,
+      "loss": 2.2989,
       "step": 4000
     },
     {
       "epoch": 2.24,
-      "eval_loss": 5.381552219390869,
-      "eval_runtime": 1248.3108,
-      "eval_samples_per_second": 2.116,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.9657080523601744,
+      "eval_loss": 6.989109039306641,
+      "eval_runtime": 1252.8987,
+      "eval_samples_per_second": 2.109,
+      "eval_steps_per_second": 0.264,
+      "eval_wer": 1.9115033716779055,
       "step": 4000
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00027906018136850786,
-      "loss": 4.1331,
+      "learning_rate": 2.7864798021434463e-05,
+      "loss": 1.895,
       "step": 4001
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002788540807914262,
-      "loss": 4.2401,
+      "learning_rate": 2.78441879637263e-05,
+      "loss": 1.8824,
       "step": 4002
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002786479802143446,
-      "loss": 4.1776,
+      "learning_rate": 2.782357790601814e-05,
+      "loss": 1.6363,
       "step": 4003
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.00027844187963726297,
-      "loss": 3.8531,
+      "learning_rate": 2.7802967848309976e-05,
+      "loss": 1.4736,
       "step": 4004
     },
     {
       "epoch": 2.24,
-      "learning_rate": 0.0002782357790601814,
-      "loss": 4.4502,
+      "learning_rate": 2.7782357790601816e-05,
+      "loss": 2.3416,
       "step": 4005
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027802967848309973,
-      "loss": 4.0371,
+      "learning_rate": 2.7761747732893656e-05,
+      "loss": 1.4714,
       "step": 4006
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027782357790601814,
-      "loss": 4.0091,
+      "learning_rate": 2.7741137675185492e-05,
+      "loss": 1.8853,
       "step": 4007
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027761747732893654,
-      "loss": 4.079,
+      "learning_rate": 2.7720527617477332e-05,
+      "loss": 1.6053,
       "step": 4008
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002774113767518549,
-      "loss": 4.2397,
+      "learning_rate": 2.769991755976917e-05,
+      "loss": 1.9316,
       "step": 4009
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002772052761747733,
-      "loss": 4.0205,
+      "learning_rate": 2.767930750206101e-05,
+      "loss": 1.7797,
       "step": 4010
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027699917559769166,
-      "loss": 4.4451,
+      "learning_rate": 2.7658697444352842e-05,
+      "loss": 2.059,
       "step": 4011
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027679307502061006,
-      "loss": 3.8258,
+      "learning_rate": 2.7638087386644686e-05,
+      "loss": 1.2318,
       "step": 4012
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002765869744435284,
-      "loss": 3.5809,
+      "learning_rate": 2.761747732893652e-05,
+      "loss": 0.8817,
       "step": 4013
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002763808738664468,
-      "loss": 3.6749,
+      "learning_rate": 2.759686727122836e-05,
+      "loss": 1.1407,
       "step": 4014
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002761747732893652,
-      "loss": 3.2793,
+      "learning_rate": 2.7576257213520202e-05,
+      "loss": 0.8695,
       "step": 4015
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002759686727122836,
-      "loss": 3.9043,
+      "learning_rate": 2.7555647155812035e-05,
+      "loss": 1.1968,
       "step": 4016
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.000275762572135202,
-      "loss": 3.6897,
+      "learning_rate": 2.753503709810388e-05,
+      "loss": 1.1514,
       "step": 4017
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027555647155812034,
-      "loss": 3.6636,
+      "learning_rate": 2.7514427040395712e-05,
+      "loss": 1.3519,
       "step": 4018
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027535037098103875,
-      "loss": 4.5099,
+      "learning_rate": 2.7493816982687552e-05,
+      "loss": 2.9821,
       "step": 4019
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002751442704039571,
-      "loss": 4.3459,
+      "learning_rate": 2.747320692497939e-05,
+      "loss": 2.0069,
       "step": 4020
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.0002749381698268755,
-      "loss": 4.1424,
+      "learning_rate": 2.745259686727123e-05,
+      "loss": 2.3182,
       "step": 4021
     },
     {
       "epoch": 2.25,
-      "learning_rate": 0.00027473206924979386,
-      "loss": 4.1823,
+      "learning_rate": 2.7431986809563072e-05,
+      "loss": 1.8567,
       "step": 4022
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027452596867271227,
-      "loss": 4.1436,
+      "learning_rate": 2.7411376751854905e-05,
+      "loss": 2.0152,
       "step": 4023
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027431986809563067,
-      "loss": 4.2751,
+      "learning_rate": 2.7390766694146745e-05,
+      "loss": 2.5202,
       "step": 4024
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.000274113767518549,
-      "loss": 4.7626,
+      "learning_rate": 2.7370156636438582e-05,
+      "loss": 2.9863,
       "step": 4025
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.0002739076669414675,
-      "loss": 4.3093,
+      "learning_rate": 2.7349546578730422e-05,
+      "loss": 2.012,
       "step": 4026
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027370156636438584,
-      "loss": 4.3633,
+      "learning_rate": 2.732893652102226e-05,
+      "loss": 2.72,
       "step": 4027
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027349546578730424,
-      "loss": 4.3353,
+      "learning_rate": 2.73083264633141e-05,
+      "loss": 1.8658,
       "step": 4028
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.0002732893652102226,
-      "loss": 4.1317,
+      "learning_rate": 2.7287716405605935e-05,
+      "loss": 1.7873,
       "step": 4029
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.000273083264633141,
-      "loss": 3.9832,
+      "learning_rate": 2.7267106347897775e-05,
+      "loss": 2.0064,
       "step": 4030
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027287716405605936,
-      "loss": 3.9999,
+      "learning_rate": 2.7246496290189615e-05,
+      "loss": 1.6848,
       "step": 4031
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027267106347897776,
-      "loss": 4.0836,
+      "learning_rate": 2.722588623248145e-05,
+      "loss": 1.7583,
       "step": 4032
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027246496290189617,
-      "loss": 4.3445,
+      "learning_rate": 2.720527617477329e-05,
+      "loss": 2.184,
       "step": 4033
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.0002722588623248145,
-      "loss": 4.0807,
+      "learning_rate": 2.7184666117065128e-05,
+      "loss": 1.758,
       "step": 4034
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027205276174773293,
-      "loss": 4.3269,
+      "learning_rate": 2.7164056059356968e-05,
+      "loss": 2.2583,
       "step": 4035
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.0002718466611706513,
-      "loss": 3.9357,
+      "learning_rate": 2.7143446001648805e-05,
+      "loss": 1.6325,
       "step": 4036
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.0002716405605935697,
-      "loss": 4.1915,
+      "learning_rate": 2.7122835943940645e-05,
+      "loss": 2.1482,
       "step": 4037
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027143446001648804,
-      "loss": 4.2524,
+      "learning_rate": 2.7102225886232485e-05,
+      "loss": 2.2487,
       "step": 4038
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027122835943940645,
-      "loss": 4.0003,
+      "learning_rate": 2.708161582852432e-05,
+      "loss": 2.2175,
       "step": 4039
     },
     {
       "epoch": 2.26,
-      "learning_rate": 0.00027102225886232485,
-      "loss": 4.3482,
+      "learning_rate": 2.706100577081616e-05,
+      "loss": 2.4593,
       "step": 4040
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002708161582852432,
-      "loss": 3.9382,
+      "learning_rate": 2.7040395713107998e-05,
+      "loss": 1.5986,
       "step": 4041
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002706100577081616,
-      "loss": 4.1501,
+      "learning_rate": 2.7019785655399838e-05,
+      "loss": 2.5544,
       "step": 4042
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00027040395713107997,
-      "loss": 4.3174,
+      "learning_rate": 2.699917559769167e-05,
+      "loss": 1.7434,
       "step": 4043
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00027019785655399837,
-      "loss": 4.1654,
+      "learning_rate": 2.6978565539983514e-05,
+      "loss": 1.9351,
       "step": 4044
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002699917559769167,
-      "loss": 4.0054,
+      "learning_rate": 2.6957955482275354e-05,
+      "loss": 1.8202,
       "step": 4045
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00026978565539983513,
-      "loss": 4.3263,
+      "learning_rate": 2.693734542456719e-05,
+      "loss": 2.2346,
       "step": 4046
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00026957955482275354,
-      "loss": 3.8431,
+      "learning_rate": 2.691673536685903e-05,
+      "loss": 1.5585,
       "step": 4047
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002693734542456719,
-      "loss": 4.1205,
+      "learning_rate": 2.6896125309150864e-05,
+      "loss": 1.5718,
       "step": 4048
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002691673536685903,
-      "loss": 3.9594,
+      "learning_rate": 2.6875515251442708e-05,
+      "loss": 1.6298,
       "step": 4049
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00026896125309150865,
-      "loss": 4.4257,
+      "learning_rate": 2.685490519373454e-05,
+      "loss": 2.1078,
       "step": 4050
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00026875515251442706,
-      "loss": 4.0251,
+      "learning_rate": 2.6834295136026384e-05,
+      "loss": 1.6044,
       "step": 4051
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002685490519373454,
-      "loss": 4.0489,
+      "learning_rate": 2.6813685078318217e-05,
+      "loss": 1.5112,
       "step": 4052
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002683429513602638,
-      "loss": 4.5972,
+      "learning_rate": 2.6793075020610057e-05,
+      "loss": 2.443,
       "step": 4053
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00026813685078318217,
-      "loss": 4.1315,
+      "learning_rate": 2.67724649629019e-05,
+      "loss": 1.6501,
       "step": 4054
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.0002679307502061006,
-      "loss": 3.922,
+      "learning_rate": 2.6751854905193734e-05,
+      "loss": 1.6832,
       "step": 4055
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.000267724649629019,
-      "loss": 3.9014,
+      "learning_rate": 2.6731244847485577e-05,
+      "loss": 1.4202,
       "step": 4056
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00026751854905193733,
-      "loss": 4.3813,
+      "learning_rate": 2.671063478977741e-05,
+      "loss": 1.7046,
       "step": 4057
     },
     {
       "epoch": 2.27,
-      "learning_rate": 0.00026731244847485574,
-      "loss": 4.0994,
+      "learning_rate": 2.669002473206925e-05,
+      "loss": 1.7218,
       "step": 4058
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002671063478977741,
-      "loss": 4.0838,
+      "learning_rate": 2.6669414674361087e-05,
+      "loss": 1.45,
       "step": 4059
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002669002473206925,
-      "loss": 4.1131,
+      "learning_rate": 2.6648804616652927e-05,
+      "loss": 1.3636,
       "step": 4060
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00026669414674361085,
-      "loss": 4.0381,
+      "learning_rate": 2.662819455894477e-05,
+      "loss": 1.6198,
       "step": 4061
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00026648804616652926,
-      "loss": 3.5563,
+      "learning_rate": 2.6607584501236604e-05,
+      "loss": 0.991,
       "step": 4062
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00026628194558944767,
-      "loss": 4.059,
+      "learning_rate": 2.6586974443528444e-05,
+      "loss": 1.2623,
       "step": 4063
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.000266075845012366,
-      "loss": 4.3212,
+      "learning_rate": 2.656636438582028e-05,
+      "loss": 1.3199,
       "step": 4064
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002658697444352844,
-      "loss": 3.7728,
+      "learning_rate": 2.654575432811212e-05,
+      "loss": 1.3835,
       "step": 4065
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002656636438582028,
-      "loss": 3.7103,
+      "learning_rate": 2.6525144270403957e-05,
+      "loss": 1.3746,
       "step": 4066
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002654575432811212,
-      "loss": 3.4302,
+      "learning_rate": 2.6504534212695797e-05,
+      "loss": 1.064,
       "step": 4067
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00026525144270403954,
-      "loss": 3.9319,
+      "learning_rate": 2.6483924154987634e-05,
+      "loss": 1.3109,
       "step": 4068
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00026504534212695794,
-      "loss": 4.3872,
+      "learning_rate": 2.6463314097279474e-05,
+      "loss": 2.2041,
       "step": 4069
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002648392415498763,
-      "loss": 4.2557,
+      "learning_rate": 2.6442704039571314e-05,
+      "loss": 2.4934,
       "step": 4070
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002646331409727947,
-      "loss": 4.4009,
+      "learning_rate": 2.642209398186315e-05,
+      "loss": 2.6537,
       "step": 4071
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.00026442704039571316,
-      "loss": 4.1808,
+      "learning_rate": 2.640148392415499e-05,
+      "loss": 2.4036,
       "step": 4072
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002642209398186315,
-      "loss": 4.0731,
+      "learning_rate": 2.6380873866446827e-05,
+      "loss": 2.1137,
       "step": 4073
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002640148392415499,
-      "loss": 4.3833,
+      "learning_rate": 2.6360263808738667e-05,
+      "loss": 2.2644,
       "step": 4074
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002638087386644683,
-      "loss": 4.3395,
+      "learning_rate": 2.6339653751030503e-05,
+      "loss": 2.0549,
       "step": 4075
     },
     {
       "epoch": 2.28,
-      "learning_rate": 0.0002636026380873867,
-      "loss": 4.499,
+      "learning_rate": 2.6319043693322343e-05,
+      "loss": 2.7904,
       "step": 4076
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026339653751030503,
-      "loss": 4.4247,
+      "learning_rate": 2.6298433635614183e-05,
+      "loss": 2.3752,
       "step": 4077
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026319043693322344,
-      "loss": 4.0909,
+      "learning_rate": 2.627782357790602e-05,
+      "loss": 1.7648,
       "step": 4078
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026298433635614185,
-      "loss": 4.6453,
+      "learning_rate": 2.625721352019786e-05,
+      "loss": 3.1727,
       "step": 4079
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002627782357790602,
-      "loss": 4.0222,
+      "learning_rate": 2.6236603462489697e-05,
+      "loss": 1.7678,
       "step": 4080
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002625721352019786,
-      "loss": 4.0939,
+      "learning_rate": 2.6215993404781537e-05,
+      "loss": 2.5593,
       "step": 4081
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026236603462489696,
-      "loss": 3.9484,
+      "learning_rate": 2.619538334707337e-05,
+      "loss": 1.5931,
       "step": 4082
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026215993404781537,
-      "loss": 4.1714,
+      "learning_rate": 2.6174773289365213e-05,
+      "loss": 1.8648,
       "step": 4083
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002619538334707337,
-      "loss": 3.9365,
+      "learning_rate": 2.6154163231657046e-05,
+      "loss": 1.6468,
       "step": 4084
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002617477328936521,
-      "loss": 4.2443,
+      "learning_rate": 2.613355317394889e-05,
+      "loss": 1.8313,
       "step": 4085
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002615416323165705,
-      "loss": 4.2667,
+      "learning_rate": 2.611294311624073e-05,
+      "loss": 2.1829,
       "step": 4086
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002613355317394889,
-      "loss": 4.1833,
+      "learning_rate": 2.6092333058532563e-05,
+      "loss": 1.8775,
       "step": 4087
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002611294311624073,
-      "loss": 4.382,
+      "learning_rate": 2.6071723000824406e-05,
+      "loss": 2.0743,
       "step": 4088
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026092333058532564,
-      "loss": 4.1521,
+      "learning_rate": 2.605111294311624e-05,
+      "loss": 1.7778,
       "step": 4089
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026071723000824405,
-      "loss": 4.0904,
+      "learning_rate": 2.603050288540808e-05,
+      "loss": 1.8255,
       "step": 4090
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002605111294311624,
-      "loss": 4.0451,
+      "learning_rate": 2.6009892827699916e-05,
+      "loss": 1.919,
       "step": 4091
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.0002603050288540808,
-      "loss": 4.1446,
+      "learning_rate": 2.5989282769991756e-05,
+      "loss": 2.1194,
       "step": 4092
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00026009892827699916,
-      "loss": 4.1433,
+      "learning_rate": 2.59686727122836e-05,
+      "loss": 2.1113,
       "step": 4093
     },
     {
       "epoch": 2.29,
-      "learning_rate": 0.00025989282769991757,
-      "loss": 4.2448,
+      "learning_rate": 2.5948062654575433e-05,
+      "loss": 1.6282,
       "step": 4094
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.000259686727122836,
-      "loss": 4.0876,
+      "learning_rate": 2.5927452596867273e-05,
+      "loss": 1.7169,
       "step": 4095
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025948062654575433,
-      "loss": 4.1934,
+      "learning_rate": 2.590684253915911e-05,
+      "loss": 1.7379,
       "step": 4096
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025927452596867273,
-      "loss": 4.3251,
+      "learning_rate": 2.588623248145095e-05,
+      "loss": 2.0054,
       "step": 4097
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002590684253915911,
-      "loss": 4.1628,
+      "learning_rate": 2.5865622423742786e-05,
+      "loss": 1.7527,
       "step": 4098
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002588623248145095,
-      "loss": 4.1732,
+      "learning_rate": 2.5845012366034626e-05,
+      "loss": 2.0758,
       "step": 4099
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025865622423742785,
-      "loss": 4.3772,
+      "learning_rate": 2.5824402308326463e-05,
+      "loss": 2.4052,
       "step": 4100
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025845012366034625,
-      "loss": 4.0576,
+      "learning_rate": 2.5803792250618303e-05,
+      "loss": 1.5949,
       "step": 4101
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002582440230832646,
-      "loss": 4.0362,
+      "learning_rate": 2.5783182192910143e-05,
+      "loss": 1.6159,
       "step": 4102
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.000258037922506183,
-      "loss": 3.7747,
+      "learning_rate": 2.576257213520198e-05,
+      "loss": 1.5305,
       "step": 4103
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002578318219291014,
-      "loss": 3.9429,
+      "learning_rate": 2.574196207749382e-05,
+      "loss": 1.4105,
       "step": 4104
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025762572135201977,
-      "loss": 3.9268,
+      "learning_rate": 2.5721352019785656e-05,
+      "loss": 1.3995,
       "step": 4105
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002574196207749382,
-      "loss": 3.829,
+      "learning_rate": 2.5700741962077496e-05,
+      "loss": 1.3949,
       "step": 4106
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025721352019785653,
-      "loss": 4.036,
+      "learning_rate": 2.5680131904369332e-05,
+      "loss": 1.3954,
       "step": 4107
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025700741962077494,
-      "loss": 3.9983,
+      "learning_rate": 2.5659521846661172e-05,
+      "loss": 1.6199,
       "step": 4108
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002568013190436933,
-      "loss": 4.2549,
+      "learning_rate": 2.5638911788953012e-05,
+      "loss": 1.8624,
       "step": 4109
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002565952184666117,
-      "loss": 4.0171,
+      "learning_rate": 2.561830173124485e-05,
+      "loss": 1.6204,
       "step": 4110
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.0002563891178895301,
-      "loss": 3.9687,
+      "learning_rate": 2.559769167353669e-05,
+      "loss": 1.1534,
       "step": 4111
     },
     {
       "epoch": 2.3,
-      "learning_rate": 0.00025618301731244845,
-      "loss": 3.6023,
+      "learning_rate": 2.5577081615828525e-05,
+      "loss": 1.0291,
       "step": 4112
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00025597691673536686,
-      "loss": 3.2781,
+      "learning_rate": 2.5556471558120365e-05,
+      "loss": 0.9095,
       "step": 4113
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002557708161582852,
-      "loss": 4.6506,
+      "learning_rate": 2.55358615004122e-05,
+      "loss": 1.7771,
       "step": 4114
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002555647155812036,
-      "loss": 3.5175,
+      "learning_rate": 2.5515251442704042e-05,
+      "loss": 1.078,
       "step": 4115
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.000255358615004122,
-      "loss": 3.6853,
+      "learning_rate": 2.5494641384995875e-05,
+      "loss": 1.1197,
       "step": 4116
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002551525144270404,
-      "loss": 3.5823,
+      "learning_rate": 2.547403132728772e-05,
+      "loss": 1.244,
       "step": 4117
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00025494641384995873,
-      "loss": 3.4434,
+      "learning_rate": 2.545342126957956e-05,
+      "loss": 1.5429,
       "step": 4118
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002547403132728772,
-      "loss": 4.2603,
+      "learning_rate": 2.5432811211871392e-05,
+      "loss": 2.3036,
       "step": 4119
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002545342126957956,
-      "loss": 4.3738,
+      "learning_rate": 2.5412201154163235e-05,
+      "loss": 2.4945,
       "step": 4120
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00025432811211871395,
-      "loss": 4.4777,
+      "learning_rate": 2.539159109645507e-05,
+      "loss": 2.1708,
       "step": 4121
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00025412201154163236,
-      "loss": 4.0555,
+      "learning_rate": 2.5370981038746912e-05,
+      "loss": 1.7878,
       "step": 4122
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002539159109645507,
-      "loss": 3.9854,
+      "learning_rate": 2.5350370981038745e-05,
+      "loss": 1.8051,
       "step": 4123
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002537098103874691,
-      "loss": 4.2689,
+      "learning_rate": 2.5329760923330585e-05,
+      "loss": 2.0307,
       "step": 4124
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00025350370981038747,
-      "loss": 4.0145,
+      "learning_rate": 2.530915086562243e-05,
+      "loss": 2.097,
       "step": 4125
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002532976092333059,
-      "loss": 4.0647,
+      "learning_rate": 2.528854080791426e-05,
+      "loss": 1.7075,
       "step": 4126
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.0002530915086562243,
-      "loss": 4.1744,
+      "learning_rate": 2.5267930750206105e-05,
+      "loss": 2.4709,
       "step": 4127
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00025288540807914264,
-      "loss": 4.3321,
+      "learning_rate": 2.5247320692497938e-05,
+      "loss": 1.952,
       "step": 4128
     },
     {
       "epoch": 2.31,
-      "learning_rate": 0.00025267930750206104,
-      "loss": 4.1915,
+      "learning_rate": 2.5226710634789778e-05,
+      "loss": 1.7348,
       "step": 4129
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002524732069249794,
-      "loss": 4.2373,
+      "learning_rate": 2.5206100577081615e-05,
+      "loss": 2.7339,
       "step": 4130
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002522671063478978,
-      "loss": 4.5036,
+      "learning_rate": 2.5185490519373455e-05,
+      "loss": 2.2301,
       "step": 4131
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00025206100577081616,
-      "loss": 4.1453,
+      "learning_rate": 2.516488046166529e-05,
+      "loss": 1.7353,
       "step": 4132
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00025185490519373456,
-      "loss": 4.7558,
+      "learning_rate": 2.514427040395713e-05,
+      "loss": 2.7905,
       "step": 4133
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002516488046166529,
-      "loss": 4.2711,
+      "learning_rate": 2.512366034624897e-05,
+      "loss": 2.406,
       "step": 4134
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002514427040395713,
-      "loss": 4.1406,
+      "learning_rate": 2.5103050288540808e-05,
+      "loss": 2.0317,
       "step": 4135
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00025123660346248973,
-      "loss": 4.1755,
+      "learning_rate": 2.5082440230832648e-05,
+      "loss": 1.8808,
       "step": 4136
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002510305028854081,
-      "loss": 4.2562,
+      "learning_rate": 2.5061830173124485e-05,
+      "loss": 1.8899,
       "step": 4137
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002508244023083265,
-      "loss": 3.9805,
+      "learning_rate": 2.5041220115416325e-05,
+      "loss": 2.5013,
       "step": 4138
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00025061830173124484,
-      "loss": 4.0209,
+      "learning_rate": 2.502061005770816e-05,
+      "loss": 1.7708,
       "step": 4139
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00025041220115416325,
-      "loss": 4.1379,
+      "learning_rate": 2.5e-05,
+      "loss": 1.6512,
       "step": 4140
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002502061005770816,
-      "loss": 3.8318,
+      "learning_rate": 2.4979389942291838e-05,
+      "loss": 1.5709,
       "step": 4141
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00025,
-      "loss": 4.0634,
+      "learning_rate": 2.4958779884583678e-05,
+      "loss": 2.0283,
       "step": 4142
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002497938994229184,
-      "loss": 3.9837,
+      "learning_rate": 2.4938169826875514e-05,
+      "loss": 1.8931,
       "step": 4143
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00024958779884583676,
-      "loss": 4.3871,
+      "learning_rate": 2.4917559769167358e-05,
+      "loss": 2.0135,
       "step": 4144
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00024938169826875517,
-      "loss": 4.1231,
+      "learning_rate": 2.4896949711459194e-05,
+      "loss": 2.372,
       "step": 4145
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.0002491755976916736,
-      "loss": 4.1318,
+      "learning_rate": 2.487633965375103e-05,
+      "loss": 1.9465,
       "step": 4146
     },
     {
       "epoch": 2.32,
-      "learning_rate": 0.00024896949711459193,
-      "loss": 4.1949,
+      "learning_rate": 2.485572959604287e-05,
+      "loss": 1.9954,
       "step": 4147
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00024876339653751034,
-      "loss": 4.3088,
+      "learning_rate": 2.4835119538334708e-05,
+      "loss": 1.8194,
       "step": 4148
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002485572959604287,
-      "loss": 4.2365,
+      "learning_rate": 2.4814509480626548e-05,
+      "loss": 2.1269,
       "step": 4149
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002483511953833471,
-      "loss": 3.9402,
+      "learning_rate": 2.4793899422918384e-05,
+      "loss": 2.086,
       "step": 4150
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00024814509480626545,
-      "loss": 3.9767,
+      "learning_rate": 2.4773289365210224e-05,
+      "loss": 2.0574,
       "step": 4151
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00024793899422918386,
-      "loss": 4.0358,
+      "learning_rate": 2.4752679307502064e-05,
+      "loss": 1.8904,
       "step": 4152
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002477328936521022,
-      "loss": 4.1359,
+      "learning_rate": 2.47320692497939e-05,
+      "loss": 2.1875,
       "step": 4153
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002475267930750206,
-      "loss": 4.0114,
+      "learning_rate": 2.471145919208574e-05,
+      "loss": 1.7801,
       "step": 4154
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.000247320692497939,
-      "loss": 3.7886,
+      "learning_rate": 2.4690849134377577e-05,
+      "loss": 1.2165,
       "step": 4155
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002471145919208574,
-      "loss": 4.1232,
+      "learning_rate": 2.4670239076669417e-05,
+      "loss": 1.4691,
       "step": 4156
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002469084913437758,
-      "loss": 3.8082,
+      "learning_rate": 2.4649629018961254e-05,
+      "loss": 1.5161,
       "step": 4157
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00024670239076669413,
-      "loss": 3.9314,
+      "learning_rate": 2.462901896125309e-05,
+      "loss": 1.7035,
       "step": 4158
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00024649629018961254,
-      "loss": 4.0557,
+      "learning_rate": 2.460840890354493e-05,
+      "loss": 1.4618,
       "step": 4159
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002462901896125309,
-      "loss": 4.1488,
+      "learning_rate": 2.458779884583677e-05,
+      "loss": 1.9286,
       "step": 4160
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002460840890354493,
-      "loss": 4.229,
+      "learning_rate": 2.456718878812861e-05,
+      "loss": 1.4467,
       "step": 4161
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002458779884583677,
-      "loss": 3.7991,
+      "learning_rate": 2.4546578730420447e-05,
+      "loss": 1.6174,
       "step": 4162
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00024567188788128606,
-      "loss": 3.8878,
+      "learning_rate": 2.4525968672712284e-05,
+      "loss": 1.3568,
       "step": 4163
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.00024546578730420446,
-      "loss": 3.834,
+      "learning_rate": 2.4505358615004124e-05,
+      "loss": 1.2466,
       "step": 4164
     },
     {
       "epoch": 2.33,
-      "learning_rate": 0.0002452596867271228,
-      "loss": 3.7531,
+      "learning_rate": 2.448474855729596e-05,
+      "loss": 1.0479,
       "step": 4165
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.0002450535861500412,
-      "loss": 3.6259,
+      "learning_rate": 2.44641384995878e-05,
+      "loss": 1.0127,
       "step": 4166
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024484748557295963,
-      "loss": 3.7643,
+      "learning_rate": 2.4443528441879637e-05,
+      "loss": 1.1653,
       "step": 4167
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.000244641384995878,
-      "loss": 3.3971,
+      "learning_rate": 2.4422918384171477e-05,
+      "loss": 1.2686,
       "step": 4168
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.0002444352844187964,
-      "loss": 4.3359,
+      "learning_rate": 2.4402308326463317e-05,
+      "loss": 2.8772,
       "step": 4169
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.0002442291838417148,
-      "loss": 4.215,
+      "learning_rate": 2.4381698268755154e-05,
+      "loss": 2.1561,
       "step": 4170
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024402308326463315,
-      "loss": 4.2213,
+      "learning_rate": 2.4361088211046994e-05,
+      "loss": 2.4777,
       "step": 4171
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024381698268755153,
-      "loss": 4.1195,
+      "learning_rate": 2.434047815333883e-05,
+      "loss": 2.2604,
       "step": 4172
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.0002436108821104699,
-      "loss": 4.2723,
+      "learning_rate": 2.431986809563067e-05,
+      "loss": 2.4498,
       "step": 4173
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.0002434047815333883,
-      "loss": 4.2158,
+      "learning_rate": 2.4299258037922507e-05,
+      "loss": 2.0952,
       "step": 4174
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024319868095630667,
-      "loss": 4.2159,
+      "learning_rate": 2.4278647980214343e-05,
+      "loss": 1.9992,
       "step": 4175
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024299258037922505,
-      "loss": 4.2421,
+      "learning_rate": 2.4258037922506187e-05,
+      "loss": 1.939,
       "step": 4176
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024278647980214343,
-      "loss": 4.8157,
+      "learning_rate": 2.4237427864798023e-05,
+      "loss": 3.0597,
       "step": 4177
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024258037922506186,
-      "loss": 4.4018,
+      "learning_rate": 2.4216817807089863e-05,
+      "loss": 2.0694,
       "step": 4178
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024237427864798024,
-      "loss": 3.9332,
+      "learning_rate": 2.41962077493817e-05,
+      "loss": 2.2893,
       "step": 4179
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024216817807089862,
-      "loss": 4.3388,
+      "learning_rate": 2.4175597691673536e-05,
+      "loss": 1.9562,
       "step": 4180
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.000241962077493817,
-      "loss": 4.4778,
+      "learning_rate": 2.4154987633965376e-05,
+      "loss": 2.2223,
       "step": 4181
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024175597691673538,
-      "loss": 4.2407,
+      "learning_rate": 2.4134377576257213e-05,
+      "loss": 2.0968,
       "step": 4182
     },
     {
       "epoch": 2.34,
-      "learning_rate": 0.00024154987633965376,
-      "loss": 4.2373,
+      "learning_rate": 2.4113767518549053e-05,
+      "loss": 1.8622,
       "step": 4183
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00024134377576257214,
-      "loss": 4.2351,
+      "learning_rate": 2.4093157460840893e-05,
+      "loss": 2.2898,
       "step": 4184
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00024113767518549052,
-      "loss": 3.8392,
+      "learning_rate": 2.407254740313273e-05,
+      "loss": 1.8056,
       "step": 4185
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00024093157460840892,
-      "loss": 4.1394,
+      "learning_rate": 2.405193734542457e-05,
+      "loss": 2.0134,
       "step": 4186
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.0002407254740313273,
-      "loss": 3.9883,
+      "learning_rate": 2.4031327287716406e-05,
+      "loss": 1.7983,
       "step": 4187
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00024051937345424568,
-      "loss": 3.9671,
+      "learning_rate": 2.4010717230008246e-05,
+      "loss": 1.8557,
       "step": 4188
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00024031327287716406,
-      "loss": 4.5236,
+      "learning_rate": 2.3990107172300083e-05,
+      "loss": 2.9047,
       "step": 4189
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00024010717230008244,
-      "loss": 4.0749,
+      "learning_rate": 2.3969497114591923e-05,
+      "loss": 2.0202,
       "step": 4190
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023990107172300082,
-      "loss": 4.0904,
+      "learning_rate": 2.394888705688376e-05,
+      "loss": 2.3733,
       "step": 4191
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.0002396949711459192,
-      "loss": 3.903,
+      "learning_rate": 2.39282769991756e-05,
+      "loss": 1.6945,
       "step": 4192
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023948887056883758,
-      "loss": 4.2202,
+      "learning_rate": 2.390766694146744e-05,
+      "loss": 1.7968,
       "step": 4193
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.000239282769991756,
-      "loss": 4.2079,
+      "learning_rate": 2.3887056883759276e-05,
+      "loss": 1.8172,
       "step": 4194
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023907666941467437,
-      "loss": 4.1324,
+      "learning_rate": 2.3866446826051113e-05,
+      "loss": 1.7203,
       "step": 4195
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023887056883759275,
-      "loss": 4.038,
+      "learning_rate": 2.3845836768342953e-05,
+      "loss": 2.0292,
       "step": 4196
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023866446826051113,
-      "loss": 4.0589,
+      "learning_rate": 2.382522671063479e-05,
+      "loss": 1.8282,
       "step": 4197
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.0002384583676834295,
-      "loss": 4.2291,
+      "learning_rate": 2.380461665292663e-05,
+      "loss": 2.2744,
       "step": 4198
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023825226710634789,
-      "loss": 3.9721,
+      "learning_rate": 2.3784006595218466e-05,
+      "loss": 1.6453,
       "step": 4199
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023804616652926627,
-      "loss": 3.859,
+      "learning_rate": 2.3763396537510306e-05,
+      "loss": 1.4779,
       "step": 4200
     },
     {
       "epoch": 2.35,
-      "learning_rate": 0.00023784006595218467,
-      "loss": 3.9392,
+      "learning_rate": 2.3742786479802146e-05,
+      "loss": 1.6972,
       "step": 4201
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023763396537510308,
-      "loss": 4.3039,
+      "learning_rate": 2.3722176422093982e-05,
+      "loss": 2.2489,
       "step": 4202
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023742786479802146,
-      "loss": 4.1007,
+      "learning_rate": 2.3701566364385822e-05,
+      "loss": 1.6703,
       "step": 4203
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023722176422093984,
-      "loss": 4.0879,
+      "learning_rate": 2.368095630667766e-05,
+      "loss": 1.946,
       "step": 4204
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023701566364385822,
-      "loss": 4.2285,
+      "learning_rate": 2.36603462489695e-05,
+      "loss": 1.9797,
       "step": 4205
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.0002368095630667766,
-      "loss": 4.1627,
+      "learning_rate": 2.3639736191261336e-05,
+      "loss": 1.7016,
       "step": 4206
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023660346248969498,
-      "loss": 4.1993,
+      "learning_rate": 2.3619126133553172e-05,
+      "loss": 2.0445,
       "step": 4207
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023639736191261336,
-      "loss": 4.0148,
+      "learning_rate": 2.3598516075845016e-05,
+      "loss": 1.6937,
       "step": 4208
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023619126133553174,
-      "loss": 4.1544,
+      "learning_rate": 2.3577906018136852e-05,
+      "loss": 1.9806,
       "step": 4209
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023598516075845014,
-      "loss": 4.0744,
+      "learning_rate": 2.3557295960428692e-05,
+      "loss": 1.4781,
       "step": 4210
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023577906018136852,
-      "loss": 4.105,
+      "learning_rate": 2.353668590272053e-05,
+      "loss": 1.4613,
       "step": 4211
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.0002355729596042869,
-      "loss": 3.5753,
+      "learning_rate": 2.3516075845012365e-05,
+      "loss": 1.2611,
       "step": 4212
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023536685902720528,
-      "loss": 3.9634,
+      "learning_rate": 2.3495465787304205e-05,
+      "loss": 1.202,
       "step": 4213
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023516075845012366,
-      "loss": 3.4128,
+      "learning_rate": 2.3474855729596042e-05,
+      "loss": 1.0566,
       "step": 4214
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023495465787304204,
-      "loss": 3.8764,
+      "learning_rate": 2.3454245671887882e-05,
+      "loss": 1.555,
       "step": 4215
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023474855729596042,
-      "loss": 3.7209,
+      "learning_rate": 2.3433635614179722e-05,
+      "loss": 1.1086,
       "step": 4216
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.0002345424567188788,
-      "loss": 3.7795,
+      "learning_rate": 2.341302555647156e-05,
+      "loss": 1.1697,
       "step": 4217
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.0002343363561417972,
-      "loss": 3.546,
+      "learning_rate": 2.33924154987634e-05,
+      "loss": 1.4044,
       "step": 4218
     },
     {
       "epoch": 2.36,
-      "learning_rate": 0.00023413025556471559,
-      "loss": 4.1219,
+      "learning_rate": 2.3371805441055235e-05,
+      "loss": 2.0796,
       "step": 4219
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023392415498763397,
-      "loss": 4.6844,
+      "learning_rate": 2.3351195383347075e-05,
+      "loss": 3.0244,
       "step": 4220
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023371805441055234,
-      "loss": 4.4344,
+      "learning_rate": 2.3330585325638912e-05,
+      "loss": 2.5023,
       "step": 4221
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023351195383347072,
-      "loss": 4.2514,
+      "learning_rate": 2.3309975267930752e-05,
+      "loss": 2.1875,
       "step": 4222
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.0002333058532563891,
-      "loss": 3.8876,
+      "learning_rate": 2.328936521022259e-05,
+      "loss": 1.8063,
       "step": 4223
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.0002330997526793075,
-      "loss": 3.9969,
+      "learning_rate": 2.326875515251443e-05,
+      "loss": 2.0445,
       "step": 4224
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.0002328936521022259,
-      "loss": 3.9539,
+      "learning_rate": 2.324814509480627e-05,
+      "loss": 1.8865,
       "step": 4225
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.0002326875515251443,
-      "loss": 4.2602,
+      "learning_rate": 2.3227535037098105e-05,
+      "loss": 2.5202,
       "step": 4226
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023248145094806268,
-      "loss": 4.1161,
+      "learning_rate": 2.3206924979389945e-05,
+      "loss": 2.2993,
       "step": 4227
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023227535037098106,
-      "loss": 4.4657,
+      "learning_rate": 2.318631492168178e-05,
+      "loss": 2.5088,
       "step": 4228
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023206924979389944,
-      "loss": 4.1667,
+      "learning_rate": 2.3165704863973618e-05,
+      "loss": 2.4949,
       "step": 4229
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023186314921681782,
-      "loss": 4.3854,
+      "learning_rate": 2.3145094806265458e-05,
+      "loss": 2.5256,
       "step": 4230
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.0002316570486397362,
-      "loss": 4.3664,
+      "learning_rate": 2.3124484748557295e-05,
+      "loss": 3.0619,
       "step": 4231
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023145094806265457,
-      "loss": 4.0751,
+      "learning_rate": 2.3103874690849138e-05,
+      "loss": 1.9442,
       "step": 4232
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023124484748557295,
-      "loss": 3.9203,
+      "learning_rate": 2.3083264633140975e-05,
+      "loss": 1.5474,
       "step": 4233
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023103874690849136,
-      "loss": 4.2466,
+      "learning_rate": 2.306265457543281e-05,
+      "loss": 1.8954,
       "step": 4234
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023083264633140974,
-      "loss": 4.1337,
+      "learning_rate": 2.304204451772465e-05,
+      "loss": 2.5269,
       "step": 4235
     },
     {
       "epoch": 2.37,
-      "learning_rate": 0.00023062654575432812,
-      "loss": 4.1854,
+      "learning_rate": 2.3021434460016488e-05,
+      "loss": 2.1125,
       "step": 4236
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.0002304204451772465,
-      "loss": 4.2772,
+      "learning_rate": 2.3000824402308328e-05,
+      "loss": 1.9354,
       "step": 4237
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00023021434460016488,
-      "loss": 4.1271,
+      "learning_rate": 2.2980214344600165e-05,
+      "loss": 2.1542,
       "step": 4238
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00023000824402308326,
-      "loss": 4.0891,
+      "learning_rate": 2.2959604286892005e-05,
+      "loss": 1.81,
       "step": 4239
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022980214344600164,
-      "loss": 4.1034,
+      "learning_rate": 2.2938994229183844e-05,
+      "loss": 2.2077,
       "step": 4240
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022959604286892002,
-      "loss": 3.9906,
+      "learning_rate": 2.291838417147568e-05,
+      "loss": 1.8944,
       "step": 4241
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022938994229183842,
-      "loss": 4.3241,
+      "learning_rate": 2.289777411376752e-05,
+      "loss": 2.1661,
       "step": 4242
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.0002291838417147568,
-      "loss": 4.0666,
+      "learning_rate": 2.2877164056059358e-05,
+      "loss": 1.7782,
       "step": 4243
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022897774113767518,
-      "loss": 4.1986,
+      "learning_rate": 2.2856553998351198e-05,
+      "loss": 1.9171,
       "step": 4244
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022877164056059356,
-      "loss": 4.1445,
+      "learning_rate": 2.2835943940643034e-05,
+      "loss": 2.1286,
       "step": 4245
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022856553998351194,
-      "loss": 3.9616,
+      "learning_rate": 2.281533388293487e-05,
+      "loss": 1.6715,
       "step": 4246
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022835943940643032,
-      "loss": 4.2208,
+      "learning_rate": 2.279472382522671e-05,
+      "loss": 2.2035,
       "step": 4247
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022815333882934873,
-      "loss": 3.8156,
+      "learning_rate": 2.277411376751855e-05,
+      "loss": 1.5827,
       "step": 4248
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.0002279472382522671,
-      "loss": 4.0649,
+      "learning_rate": 2.275350370981039e-05,
+      "loss": 1.5826,
       "step": 4249
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022774113767518552,
-      "loss": 4.0061,
+      "learning_rate": 2.2732893652102227e-05,
+      "loss": 2.0344,
       "step": 4250
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.0002275350370981039,
-      "loss": 3.9628,
+      "learning_rate": 2.2712283594394064e-05,
+      "loss": 1.6277,
       "step": 4251
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022732893652102227,
-      "loss": 3.8302,
+      "learning_rate": 2.2691673536685904e-05,
+      "loss": 1.6443,
       "step": 4252
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022712283594394065,
-      "loss": 3.9246,
+      "learning_rate": 2.267106347897774e-05,
+      "loss": 1.6544,
       "step": 4253
     },
     {
       "epoch": 2.38,
-      "learning_rate": 0.00022691673536685903,
-      "loss": 4.0111,
+      "learning_rate": 2.265045342126958e-05,
+      "loss": 1.6311,
       "step": 4254
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0002267106347897774,
-      "loss": 4.0364,
+      "learning_rate": 2.262984336356142e-05,
+      "loss": 2.1493,
       "step": 4255
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0002265045342126958,
-      "loss": 3.6394,
+      "learning_rate": 2.2609233305853257e-05,
+      "loss": 2.1725,
       "step": 4256
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0002262984336356142,
-      "loss": 3.7725,
+      "learning_rate": 2.2588623248145097e-05,
+      "loss": 1.7109,
       "step": 4257
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022609233305853258,
-      "loss": 4.0937,
+      "learning_rate": 2.2568013190436934e-05,
+      "loss": 1.6394,
       "step": 4258
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022588623248145096,
-      "loss": 4.0166,
+      "learning_rate": 2.2547403132728774e-05,
+      "loss": 1.5542,
       "step": 4259
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022568013190436934,
-      "loss": 3.659,
+      "learning_rate": 2.252679307502061e-05,
+      "loss": 1.2829,
       "step": 4260
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022547403132728772,
-      "loss": 3.853,
+      "learning_rate": 2.250618301731245e-05,
+      "loss": 1.5369,
       "step": 4261
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0002252679307502061,
-      "loss": 4.0426,
+      "learning_rate": 2.2485572959604287e-05,
+      "loss": 1.1848,
       "step": 4262
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022506183017312448,
-      "loss": 3.7204,
+      "learning_rate": 2.2464962901896127e-05,
+      "loss": 1.2803,
       "step": 4263
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022485572959604286,
-      "loss": 3.6215,
+      "learning_rate": 2.2444352844187967e-05,
+      "loss": 1.0699,
       "step": 4264
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022464962901896126,
-      "loss": 3.4008,
+      "learning_rate": 2.2423742786479804e-05,
+      "loss": 0.8959,
       "step": 4265
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022444352844187964,
-      "loss": 3.4636,
+      "learning_rate": 2.2403132728771644e-05,
+      "loss": 1.1409,
       "step": 4266
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022423742786479802,
-      "loss": 3.3569,
+      "learning_rate": 2.238252267106348e-05,
+      "loss": 1.0743,
       "step": 4267
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.0002240313272877164,
-      "loss": 3.873,
+      "learning_rate": 2.2361912613355317e-05,
+      "loss": 1.42,
       "step": 4268
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022382522671063478,
-      "loss": 4.3543,
+      "learning_rate": 2.2341302555647157e-05,
+      "loss": 2.337,
       "step": 4269
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022361912613355316,
-      "loss": 4.4158,
+      "learning_rate": 2.2320692497938993e-05,
+      "loss": 2.3188,
       "step": 4270
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022341302555647157,
-      "loss": 4.2406,
+      "learning_rate": 2.2300082440230833e-05,
+      "loss": 2.1428,
       "step": 4271
     },
     {
       "epoch": 2.39,
-      "learning_rate": 0.00022320692497938995,
-      "loss": 3.8841,
+      "learning_rate": 2.2279472382522673e-05,
+      "loss": 1.7699,
       "step": 4272
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022300082440230835,
-      "loss": 4.4527,
+      "learning_rate": 2.225886232481451e-05,
+      "loss": 2.7778,
       "step": 4273
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022279472382522673,
-      "loss": 4.1433,
+      "learning_rate": 2.223825226710635e-05,
+      "loss": 1.9377,
       "step": 4274
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022258862324814511,
-      "loss": 4.0397,
+      "learning_rate": 2.2217642209398187e-05,
+      "loss": 1.7203,
       "step": 4275
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.0002223825226710635,
-      "loss": 4.4524,
+      "learning_rate": 2.2197032151690027e-05,
+      "loss": 2.2463,
       "step": 4276
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022217642209398187,
-      "loss": 4.1937,
+      "learning_rate": 2.2176422093981863e-05,
+      "loss": 2.4791,
       "step": 4277
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022197032151690025,
-      "loss": 4.2321,
+      "learning_rate": 2.2155812036273703e-05,
+      "loss": 2.6132,
       "step": 4278
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022176422093981863,
-      "loss": 4.1546,
+      "learning_rate": 2.2135201978565543e-05,
+      "loss": 1.8918,
       "step": 4279
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.000221558120362737,
-      "loss": 4.2091,
+      "learning_rate": 2.211459192085738e-05,
+      "loss": 2.1618,
       "step": 4280
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022135201978565542,
-      "loss": 4.2479,
+      "learning_rate": 2.209398186314922e-05,
+      "loss": 2.9924,
       "step": 4281
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.0002211459192085738,
-      "loss": 4.1404,
+      "learning_rate": 2.2073371805441056e-05,
+      "loss": 2.0941,
       "step": 4282
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022093981863149218,
-      "loss": 4.2063,
+      "learning_rate": 2.2052761747732893e-05,
+      "loss": 2.0075,
       "step": 4283
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022073371805441056,
-      "loss": 4.2816,
+      "learning_rate": 2.2032151690024733e-05,
+      "loss": 2.054,
       "step": 4284
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022052761747732894,
-      "loss": 4.4483,
+      "learning_rate": 2.201154163231657e-05,
+      "loss": 2.6934,
       "step": 4285
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00022032151690024732,
-      "loss": 3.9143,
+      "learning_rate": 2.199093157460841e-05,
+      "loss": 2.1163,
       "step": 4286
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.0002201154163231657,
-      "loss": 4.0612,
+      "learning_rate": 2.197032151690025e-05,
+      "loss": 2.0163,
       "step": 4287
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00021990931574608408,
-      "loss": 4.0047,
+      "learning_rate": 2.1949711459192086e-05,
+      "loss": 1.9309,
       "step": 4288
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00021970321516900248,
-      "loss": 4.3881,
+      "learning_rate": 2.1929101401483926e-05,
+      "loss": 2.4642,
       "step": 4289
     },
     {
       "epoch": 2.4,
-      "learning_rate": 0.00021949711459192086,
-      "loss": 4.1476,
+      "learning_rate": 2.1908491343775763e-05,
+      "loss": 1.9763,
       "step": 4290
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021929101401483924,
-      "loss": 4.2333,
+      "learning_rate": 2.1887881286067603e-05,
+      "loss": 2.1958,
       "step": 4291
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021908491343775762,
-      "loss": 3.8476,
+      "learning_rate": 2.186727122835944e-05,
+      "loss": 1.8168,
       "step": 4292
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.000218878812860676,
-      "loss": 3.9771,
+      "learning_rate": 2.184666117065128e-05,
+      "loss": 1.7234,
       "step": 4293
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.0002186727122835944,
-      "loss": 4.364,
+      "learning_rate": 2.1826051112943116e-05,
+      "loss": 2.7914,
       "step": 4294
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.0002184666117065128,
-      "loss": 4.0412,
+      "learning_rate": 2.1805441055234956e-05,
+      "loss": 1.9713,
       "step": 4295
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021826051112943117,
-      "loss": 4.0513,
+      "learning_rate": 2.1784830997526796e-05,
+      "loss": 1.7551,
       "step": 4296
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021805441055234957,
-      "loss": 3.9818,
+      "learning_rate": 2.1764220939818633e-05,
+      "loss": 1.8,
       "step": 4297
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021784830997526795,
-      "loss": 4.434,
+      "learning_rate": 2.1743610882110473e-05,
+      "loss": 2.7988,
       "step": 4298
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021764220939818633,
-      "loss": 3.962,
+      "learning_rate": 2.172300082440231e-05,
+      "loss": 1.8184,
       "step": 4299
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.0002174361088211047,
-      "loss": 4.1632,
+      "learning_rate": 2.1702390766694146e-05,
+      "loss": 1.9479,
       "step": 4300
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.0002172300082440231,
-      "loss": 4.0821,
+      "learning_rate": 2.1681780708985986e-05,
+      "loss": 1.7029,
       "step": 4301
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021702390766694147,
-      "loss": 4.0697,
+      "learning_rate": 2.1661170651277822e-05,
+      "loss": 1.8094,
       "step": 4302
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021681780708985985,
-      "loss": 4.4036,
+      "learning_rate": 2.1640560593569666e-05,
+      "loss": 1.9251,
       "step": 4303
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021661170651277823,
-      "loss": 3.9768,
+      "learning_rate": 2.1619950535861502e-05,
+      "loss": 1.5201,
       "step": 4304
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021640560593569664,
-      "loss": 4.1046,
+      "learning_rate": 2.159934047815334e-05,
+      "loss": 2.0926,
       "step": 4305
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021619950535861502,
-      "loss": 3.835,
+      "learning_rate": 2.157873042044518e-05,
+      "loss": 1.6682,
       "step": 4306
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.0002159934047815334,
-      "loss": 3.9192,
+      "learning_rate": 2.1558120362737016e-05,
+      "loss": 1.4861,
       "step": 4307
     },
     {
       "epoch": 2.41,
-      "learning_rate": 0.00021578730420445178,
-      "loss": 4.0563,
+      "learning_rate": 2.1537510305028855e-05,
+      "loss": 1.6772,
       "step": 4308
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021558120362737016,
-      "loss": 4.0672,
+      "learning_rate": 2.1516900247320692e-05,
+      "loss": 1.5653,
       "step": 4309
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021537510305028853,
-      "loss": 4.1773,
+      "learning_rate": 2.1496290189612532e-05,
+      "loss": 1.5368,
       "step": 4310
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021516900247320691,
-      "loss": 4.252,
+      "learning_rate": 2.1475680131904372e-05,
+      "loss": 1.5103,
       "step": 4311
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.0002149629018961253,
-      "loss": 4.163,
+      "learning_rate": 2.145507007419621e-05,
+      "loss": 1.7233,
       "step": 4312
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.0002147568013190437,
-      "loss": 4.0419,
+      "learning_rate": 2.143446001648805e-05,
+      "loss": 1.633,
       "step": 4313
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021455070074196208,
-      "loss": 3.884,
+      "learning_rate": 2.1413849958779885e-05,
+      "loss": 1.6287,
       "step": 4314
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021434460016488046,
-      "loss": 4.0691,
+      "learning_rate": 2.1393239901071725e-05,
+      "loss": 1.1541,
       "step": 4315
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021413849958779884,
-      "loss": 3.6671,
+      "learning_rate": 2.1372629843363562e-05,
+      "loss": 1.1021,
       "step": 4316
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021393239901071725,
-      "loss": 3.4863,
+      "learning_rate": 2.13520197856554e-05,
+      "loss": 1.2065,
       "step": 4317
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021372629843363563,
-      "loss": 3.971,
+      "learning_rate": 2.133140972794724e-05,
+      "loss": 1.9903,
       "step": 4318
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.000213520197856554,
-      "loss": 4.4743,
+      "learning_rate": 2.131079967023908e-05,
+      "loss": 3.3567,
       "step": 4319
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021331409727947238,
-      "loss": 4.534,
+      "learning_rate": 2.129018961253092e-05,
+      "loss": 2.3175,
       "step": 4320
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.0002131079967023908,
-      "loss": 4.0873,
+      "learning_rate": 2.1269579554822755e-05,
+      "loss": 2.0101,
       "step": 4321
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021290189612530917,
-      "loss": 4.4119,
+      "learning_rate": 2.124896949711459e-05,
+      "loss": 3.1063,
       "step": 4322
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021269579554822755,
-      "loss": 4.189,
+      "learning_rate": 2.122835943940643e-05,
+      "loss": 2.0418,
       "step": 4323
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.00021248969497114593,
-      "loss": 4.1494,
+      "learning_rate": 2.1207749381698268e-05,
+      "loss": 1.9794,
       "step": 4324
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.0002122835943940643,
-      "loss": 3.9685,
+      "learning_rate": 2.1187139323990108e-05,
+      "loss": 2.3953,
       "step": 4325
     },
     {
       "epoch": 2.42,
-      "learning_rate": 0.0002120774938169827,
-      "loss": 4.4685,
+      "learning_rate": 2.1166529266281945e-05,
+      "loss": 2.7116,
       "step": 4326
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021187139323990107,
-      "loss": 4.4001,
+      "learning_rate": 2.1145919208573785e-05,
+      "loss": 2.1733,
       "step": 4327
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021166529266281945,
-      "loss": 3.8996,
+      "learning_rate": 2.1125309150865625e-05,
+      "loss": 1.8868,
       "step": 4328
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021145919208573786,
-      "loss": 3.9122,
+      "learning_rate": 2.110469909315746e-05,
+      "loss": 1.8032,
       "step": 4329
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021125309150865623,
-      "loss": 3.8971,
+      "learning_rate": 2.10840890354493e-05,
+      "loss": 1.9019,
       "step": 4330
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021104699093157461,
-      "loss": 4.2987,
+      "learning_rate": 2.1063478977741138e-05,
+      "loss": 2.551,
       "step": 4331
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.000210840890354493,
-      "loss": 4.3142,
+      "learning_rate": 2.1042868920032978e-05,
+      "loss": 2.546,
       "step": 4332
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021063478977741137,
-      "loss": 4.117,
+      "learning_rate": 2.1022258862324815e-05,
+      "loss": 2.2454,
       "step": 4333
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021042868920032975,
-      "loss": 4.1344,
+      "learning_rate": 2.100164880461665e-05,
+      "loss": 2.0863,
       "step": 4334
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00021022258862324813,
-      "loss": 4.1553,
+      "learning_rate": 2.0981038746908495e-05,
+      "loss": 2.0152,
       "step": 4335
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.0002100164880461665,
-      "loss": 4.3156,
+      "learning_rate": 2.096042868920033e-05,
+      "loss": 2.1251,
       "step": 4336
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00020981038746908492,
-      "loss": 3.9727,
+      "learning_rate": 2.093981863149217e-05,
+      "loss": 1.8478,
       "step": 4337
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.0002096042868920033,
-      "loss": 4.224,
+      "learning_rate": 2.0919208573784008e-05,
+      "loss": 2.2969,
       "step": 4338
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00020939818631492168,
-      "loss": 4.668,
+      "learning_rate": 2.0898598516075844e-05,
+      "loss": 2.6557,
       "step": 4339
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00020919208573784006,
-      "loss": 4.1099,
+      "learning_rate": 2.0877988458367684e-05,
+      "loss": 2.0659,
       "step": 4340
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00020898598516075846,
-      "loss": 4.0461,
+      "learning_rate": 2.085737840065952e-05,
+      "loss": 1.7566,
       "step": 4341
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00020877988458367684,
-      "loss": 4.0291,
+      "learning_rate": 2.083676834295136e-05,
+      "loss": 2.4043,
       "step": 4342
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.00020857378400659522,
-      "loss": 4.0483,
+      "learning_rate": 2.08161582852432e-05,
+      "loss": 1.7091,
       "step": 4343
     },
     {
       "epoch": 2.43,
-      "learning_rate": 0.0002083676834295136,
-      "loss": 4.3188,
+      "learning_rate": 2.0795548227535038e-05,
+      "loss": 2.0176,
       "step": 4344
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.000208161582852432,
-      "loss": 4.2807,
+      "learning_rate": 2.0774938169826878e-05,
+      "loss": 2.6009,
       "step": 4345
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.0002079554822753504,
-      "loss": 4.0702,
+      "learning_rate": 2.0754328112118714e-05,
+      "loss": 2.1197,
       "step": 4346
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020774938169826877,
-      "loss": 4.2158,
+      "learning_rate": 2.0733718054410554e-05,
+      "loss": 2.7992,
       "step": 4347
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020754328112118715,
-      "loss": 3.8323,
+      "learning_rate": 2.071310799670239e-05,
+      "loss": 2.074,
       "step": 4348
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020733718054410553,
-      "loss": 4.178,
+      "learning_rate": 2.069249793899423e-05,
+      "loss": 2.0822,
       "step": 4349
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.0002071310799670239,
-      "loss": 3.981,
+      "learning_rate": 2.0671887881286067e-05,
+      "loss": 2.3792,
       "step": 4350
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.0002069249793899423,
-      "loss": 3.7945,
+      "learning_rate": 2.0651277823577907e-05,
+      "loss": 1.4841,
       "step": 4351
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020671887881286067,
-      "loss": 3.8524,
+      "learning_rate": 2.0630667765869747e-05,
+      "loss": 1.8934,
       "step": 4352
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020651277823577907,
-      "loss": 4.0307,
+      "learning_rate": 2.0610057708161584e-05,
+      "loss": 2.1588,
       "step": 4353
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020630667765869745,
-      "loss": 3.9921,
+      "learning_rate": 2.0589447650453424e-05,
+      "loss": 2.0268,
       "step": 4354
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020610057708161583,
-      "loss": 3.9507,
+      "learning_rate": 2.056883759274526e-05,
+      "loss": 1.5362,
       "step": 4355
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.0002058944765045342,
-      "loss": 4.1041,
+      "learning_rate": 2.0548227535037097e-05,
+      "loss": 1.8591,
       "step": 4356
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.0002056883759274526,
-      "loss": 4.1293,
+      "learning_rate": 2.0527617477328937e-05,
+      "loss": 1.6963,
       "step": 4357
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020548227535037097,
-      "loss": 3.8735,
+      "learning_rate": 2.0507007419620774e-05,
+      "loss": 1.4131,
       "step": 4358
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020527617477328935,
-      "loss": 4.2201,
+      "learning_rate": 2.0486397361912614e-05,
+      "loss": 2.1846,
       "step": 4359
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020507007419620773,
-      "loss": 4.1878,
+      "learning_rate": 2.0465787304204454e-05,
+      "loss": 2.4062,
       "step": 4360
     },
     {
       "epoch": 2.44,
-      "learning_rate": 0.00020486397361912614,
-      "loss": 4.0328,
+      "learning_rate": 2.044517724649629e-05,
+      "loss": 1.8621,
       "step": 4361
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020465787304204452,
-      "loss": 4.0569,
+      "learning_rate": 2.042456718878813e-05,
+      "loss": 1.704,
       "step": 4362
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002044517724649629,
-      "loss": 3.8381,
+      "learning_rate": 2.0403957131079967e-05,
+      "loss": 1.6612,
       "step": 4363
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002042456718878813,
-      "loss": 3.9157,
+      "learning_rate": 2.0383347073371807e-05,
+      "loss": 1.656,
       "step": 4364
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020403957131079968,
-      "loss": 3.924,
+      "learning_rate": 2.0362737015663644e-05,
+      "loss": 1.2891,
       "step": 4365
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020383347073371806,
-      "loss": 3.4105,
+      "learning_rate": 2.0342126957955484e-05,
+      "loss": 1.101,
       "step": 4366
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020362737015663644,
-      "loss": 3.723,
+      "learning_rate": 2.0321516900247324e-05,
+      "loss": 1.2569,
       "step": 4367
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020342126957955485,
-      "loss": 2.9416,
+      "learning_rate": 2.030090684253916e-05,
+      "loss": 1.2962,
       "step": 4368
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020321516900247323,
-      "loss": 4.4254,
+      "learning_rate": 2.0280296784831e-05,
+      "loss": 2.7966,
       "step": 4369
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002030090684253916,
-      "loss": 4.253,
+      "learning_rate": 2.0259686727122837e-05,
+      "loss": 2.5276,
       "step": 4370
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020280296784831,
-      "loss": 4.1225,
+      "learning_rate": 2.0239076669414673e-05,
+      "loss": 2.1794,
       "step": 4371
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020259686727122837,
-      "loss": 4.199,
+      "learning_rate": 2.0218466611706513e-05,
+      "loss": 2.0913,
       "step": 4372
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020239076669414675,
-      "loss": 3.975,
+      "learning_rate": 2.019785655399835e-05,
+      "loss": 1.9101,
       "step": 4373
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020218466611706513,
-      "loss": 4.3805,
+      "learning_rate": 2.0177246496290193e-05,
+      "loss": 2.7592,
       "step": 4374
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002019785655399835,
-      "loss": 4.2384,
+      "learning_rate": 2.015663643858203e-05,
+      "loss": 2.3406,
       "step": 4375
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002017724649629019,
-      "loss": 4.4688,
+      "learning_rate": 2.0136026380873867e-05,
+      "loss": 2.6645,
       "step": 4376
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.0002015663643858203,
-      "loss": 4.1422,
+      "learning_rate": 2.0115416323165706e-05,
+      "loss": 2.0601,
       "step": 4377
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020136026380873867,
-      "loss": 4.2854,
+      "learning_rate": 2.0094806265457543e-05,
+      "loss": 2.0104,
       "step": 4378
     },
     {
       "epoch": 2.45,
-      "learning_rate": 0.00020115416323165705,
-      "loss": 4.4665,
+      "learning_rate": 2.0074196207749383e-05,
+      "loss": 2.7165,
       "step": 4379
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00020094806265457543,
-      "loss": 4.1536,
+      "learning_rate": 2.005358615004122e-05,
+      "loss": 2.2201,
       "step": 4380
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.0002007419620774938,
-      "loss": 3.9008,
+      "learning_rate": 2.003297609233306e-05,
+      "loss": 1.6738,
       "step": 4381
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.0002005358615004122,
-      "loss": 3.98,
+      "learning_rate": 2.00123660346249e-05,
+      "loss": 1.8875,
       "step": 4382
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00020032976092333057,
-      "loss": 4.0527,
+      "learning_rate": 1.9991755976916736e-05,
+      "loss": 1.7755,
       "step": 4383
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00020012366034624898,
-      "loss": 4.4258,
+      "learning_rate": 1.9971145919208576e-05,
+      "loss": 2.6623,
       "step": 4384
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019991755976916736,
-      "loss": 4.3031,
+      "learning_rate": 1.9950535861500413e-05,
+      "loss": 2.3205,
       "step": 4385
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019971145919208574,
-      "loss": 3.924,
+      "learning_rate": 1.9929925803792253e-05,
+      "loss": 1.788,
       "step": 4386
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019950535861500414,
-      "loss": 4.0057,
+      "learning_rate": 1.990931574608409e-05,
+      "loss": 2.2443,
       "step": 4387
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019929925803792252,
-      "loss": 4.1019,
+      "learning_rate": 1.9888705688375926e-05,
+      "loss": 2.0013,
       "step": 4388
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.0001990931574608409,
-      "loss": 4.1623,
+      "learning_rate": 1.9868095630667766e-05,
+      "loss": 1.8645,
       "step": 4389
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019888705688375928,
-      "loss": 4.0478,
+      "learning_rate": 1.9847485572959606e-05,
+      "loss": 1.9215,
       "step": 4390
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019868095630667766,
-      "loss": 3.8388,
+      "learning_rate": 1.9826875515251446e-05,
+      "loss": 2.1439,
       "step": 4391
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019847485572959607,
-      "loss": 4.2833,
+      "learning_rate": 1.9806265457543283e-05,
+      "loss": 2.2421,
       "step": 4392
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019826875515251445,
-      "loss": 4.0378,
+      "learning_rate": 1.978565539983512e-05,
+      "loss": 1.9533,
       "step": 4393
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019806265457543283,
-      "loss": 4.0154,
+      "learning_rate": 1.976504534212696e-05,
+      "loss": 2.3174,
       "step": 4394
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.0001978565539983512,
-      "loss": 4.1751,
+      "learning_rate": 1.9744435284418796e-05,
+      "loss": 2.3749,
       "step": 4395
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019765045342126959,
-      "loss": 3.7957,
+      "learning_rate": 1.9723825226710636e-05,
+      "loss": 2.0048,
       "step": 4396
     },
     {
       "epoch": 2.46,
-      "learning_rate": 0.00019744435284418797,
-      "loss": 4.3307,
+      "learning_rate": 1.9703215169002472e-05,
+      "loss": 2.488,
       "step": 4397
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019723825226710634,
-      "loss": 4.1971,
+      "learning_rate": 1.9682605111294312e-05,
+      "loss": 1.9218,
       "step": 4398
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019703215169002472,
-      "loss": 4.0775,
+      "learning_rate": 1.9661995053586152e-05,
+      "loss": 1.69,
       "step": 4399
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019682605111294313,
-      "loss": 3.9886,
+      "learning_rate": 1.964138499587799e-05,
+      "loss": 2.1378,
       "step": 4400
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0001966199505358615,
-      "loss": 3.9144,
+      "learning_rate": 1.962077493816983e-05,
+      "loss": 1.7249,
       "step": 4401
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0001964138499587799,
-      "loss": 3.9152,
+      "learning_rate": 1.9600164880461666e-05,
+      "loss": 1.6303,
       "step": 4402
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019620774938169827,
-      "loss": 3.8627,
+      "learning_rate": 1.9579554822753506e-05,
+      "loss": 1.7236,
       "step": 4403
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019600164880461665,
-      "loss": 4.3605,
+      "learning_rate": 1.9558944765045342e-05,
+      "loss": 2.4136,
       "step": 4404
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019579554822753503,
-      "loss": 3.942,
+      "learning_rate": 1.953833470733718e-05,
+      "loss": 1.5809,
       "step": 4405
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0001955894476504534,
-      "loss": 3.8692,
+      "learning_rate": 1.9517724649629022e-05,
+      "loss": 1.6878,
       "step": 4406
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0001953833470733718,
-      "loss": 4.1607,
+      "learning_rate": 1.949711459192086e-05,
+      "loss": 1.7359,
       "step": 4407
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0001951772464962902,
-      "loss": 3.7447,
+      "learning_rate": 1.94765045342127e-05,
+      "loss": 1.6604,
       "step": 4408
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019497114591920857,
-      "loss": 3.8476,
+      "learning_rate": 1.9455894476504535e-05,
+      "loss": 1.3235,
       "step": 4409
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019476504534212695,
-      "loss": 4.0081,
+      "learning_rate": 1.9435284418796372e-05,
+      "loss": 1.3772,
       "step": 4410
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019455894476504536,
-      "loss": 4.1192,
+      "learning_rate": 1.9414674361088212e-05,
+      "loss": 1.3674,
       "step": 4411
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019435284418796374,
-      "loss": 3.8437,
+      "learning_rate": 1.939406430338005e-05,
+      "loss": 1.3205,
       "step": 4412
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019414674361088212,
-      "loss": 3.9487,
+      "learning_rate": 1.937345424567189e-05,
+      "loss": 1.2979,
       "step": 4413
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.0001939406430338005,
-      "loss": 3.8404,
+      "learning_rate": 1.935284418796373e-05,
+      "loss": 1.2836,
       "step": 4414
     },
     {
       "epoch": 2.47,
-      "learning_rate": 0.00019373454245671888,
-      "loss": 3.7573,
+      "learning_rate": 1.9332234130255565e-05,
+      "loss": 1.1934,
       "step": 4415
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019352844187963729,
-      "loss": 3.5396,
+      "learning_rate": 1.9311624072547405e-05,
+      "loss": 1.1745,
       "step": 4416
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019332234130255567,
-      "loss": 3.7796,
+      "learning_rate": 1.9291014014839242e-05,
+      "loss": 1.6749,
       "step": 4417
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019311624072547404,
-      "loss": 3.6335,
+      "learning_rate": 1.9270403957131082e-05,
+      "loss": 1.6855,
       "step": 4418
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019291014014839242,
-      "loss": 4.1384,
+      "learning_rate": 1.924979389942292e-05,
+      "loss": 2.1572,
       "step": 4419
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.0001927040395713108,
-      "loss": 4.0137,
+      "learning_rate": 1.922918384171476e-05,
+      "loss": 1.928,
       "step": 4420
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019249793899422918,
-      "loss": 4.3262,
+      "learning_rate": 1.9208573784006595e-05,
+      "loss": 2.2311,
       "step": 4421
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019229183841714756,
-      "loss": 4.0526,
+      "learning_rate": 1.9187963726298435e-05,
+      "loss": 1.9636,
       "step": 4422
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019208573784006594,
-      "loss": 4.165,
+      "learning_rate": 1.9167353668590275e-05,
+      "loss": 2.2186,
       "step": 4423
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019187963726298435,
-      "loss": 3.9193,
+      "learning_rate": 1.914674361088211e-05,
+      "loss": 1.8668,
       "step": 4424
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019167353668590273,
-      "loss": 4.3552,
+      "learning_rate": 1.912613355317395e-05,
+      "loss": 2.4097,
       "step": 4425
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.0001914674361088211,
-      "loss": 4.1973,
+      "learning_rate": 1.9105523495465788e-05,
+      "loss": 2.2406,
       "step": 4426
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.0001912613355317395,
-      "loss": 4.2136,
+      "learning_rate": 1.9084913437757625e-05,
+      "loss": 2.5389,
       "step": 4427
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019105523495465787,
-      "loss": 4.1574,
+      "learning_rate": 1.9064303380049465e-05,
+      "loss": 2.373,
       "step": 4428
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019084913437757625,
-      "loss": 4.375,
+      "learning_rate": 1.90436933223413e-05,
+      "loss": 2.463,
       "step": 4429
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.00019064303380049463,
-      "loss": 4.0145,
+      "learning_rate": 1.9023083264633145e-05,
+      "loss": 1.9646,
       "step": 4430
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.000190436933223413,
-      "loss": 4.0281,
+      "learning_rate": 1.900247320692498e-05,
+      "loss": 2.1152,
       "step": 4431
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.0001902308326463314,
-      "loss": 3.9624,
+      "learning_rate": 1.8981863149216818e-05,
+      "loss": 1.9956,
       "step": 4432
     },
     {
       "epoch": 2.48,
-      "learning_rate": 0.0001900247320692498,
-      "loss": 4.1391,
+      "learning_rate": 1.8961253091508658e-05,
+      "loss": 2.3962,
       "step": 4433
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.0001898186314921682,
-      "loss": 4.3172,
+      "learning_rate": 1.8940643033800495e-05,
+      "loss": 2.5547,
       "step": 4434
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018961253091508658,
-      "loss": 4.17,
+      "learning_rate": 1.8920032976092335e-05,
+      "loss": 1.9338,
       "step": 4435
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018940643033800496,
-      "loss": 4.0972,
+      "learning_rate": 1.889942291838417e-05,
+      "loss": 1.7885,
       "step": 4436
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018920032976092334,
-      "loss": 3.8981,
+      "learning_rate": 1.887881286067601e-05,
+      "loss": 1.8209,
       "step": 4437
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018899422918384172,
-      "loss": 4.2449,
+      "learning_rate": 1.885820280296785e-05,
+      "loss": 2.3371,
       "step": 4438
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.0001887881286067601,
-      "loss": 4.3083,
+      "learning_rate": 1.8837592745259688e-05,
+      "loss": 2.6515,
       "step": 4439
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.0001885820280296785,
-      "loss": 4.2294,
+      "learning_rate": 1.8816982687551528e-05,
+      "loss": 2.2864,
       "step": 4440
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018837592745259688,
-      "loss": 4.1225,
+      "learning_rate": 1.8796372629843364e-05,
+      "loss": 2.1313,
       "step": 4441
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018816982687551526,
-      "loss": 3.9648,
+      "learning_rate": 1.8775762572135204e-05,
+      "loss": 1.8508,
       "step": 4442
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018796372629843364,
-      "loss": 4.2298,
+      "learning_rate": 1.875515251442704e-05,
+      "loss": 2.0271,
       "step": 4443
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018775762572135202,
-      "loss": 4.1792,
+      "learning_rate": 1.8734542456718878e-05,
+      "loss": 2.5785,
       "step": 4444
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.0001875515251442704,
-      "loss": 4.2069,
+      "learning_rate": 1.8713932399010717e-05,
+      "loss": 2.0144,
       "step": 4445
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018734542456718878,
-      "loss": 3.9598,
+      "learning_rate": 1.8693322341302557e-05,
+      "loss": 1.9395,
       "step": 4446
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018713932399010716,
-      "loss": 3.9503,
+      "learning_rate": 1.8672712283594394e-05,
+      "loss": 1.8713,
       "step": 4447
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018693322341302557,
-      "loss": 4.3953,
+      "learning_rate": 1.8652102225886234e-05,
+      "loss": 2.704,
       "step": 4448
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018672712283594395,
-      "loss": 4.2098,
+      "learning_rate": 1.863149216817807e-05,
+      "loss": 2.0534,
       "step": 4449
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.00018652102225886233,
-      "loss": 3.9062,
+      "learning_rate": 1.861088211046991e-05,
+      "loss": 2.0762,
       "step": 4450
     },
     {
       "epoch": 2.49,
-      "learning_rate": 0.0001863149216817807,
-      "loss": 4.1454,
+      "learning_rate": 1.8590272052761747e-05,
+      "loss": 1.947,
       "step": 4451
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018610882110469909,
-      "loss": 4.0466,
+      "learning_rate": 1.8569661995053587e-05,
+      "loss": 2.4107,
       "step": 4452
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018590272052761747,
-      "loss": 3.9341,
+      "learning_rate": 1.8549051937345424e-05,
+      "loss": 1.6933,
       "step": 4453
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018569661995053585,
-      "loss": 4.1494,
+      "learning_rate": 1.8528441879637264e-05,
+      "loss": 2.5017,
       "step": 4454
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018549051937345423,
-      "loss": 4.1228,
+      "learning_rate": 1.8507831821929104e-05,
+      "loss": 1.7859,
       "step": 4455
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018528441879637263,
-      "loss": 3.9948,
+      "learning_rate": 1.848722176422094e-05,
+      "loss": 2.4809,
       "step": 4456
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018507831821929104,
-      "loss": 3.9963,
+      "learning_rate": 1.846661170651278e-05,
+      "loss": 1.5954,
       "step": 4457
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018487221764220942,
-      "loss": 4.0268,
+      "learning_rate": 1.8446001648804617e-05,
+      "loss": 1.7383,
       "step": 4458
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.0001846661170651278,
-      "loss": 4.4171,
+      "learning_rate": 1.8425391591096454e-05,
+      "loss": 2.0864,
       "step": 4459
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018446001648804618,
-      "loss": 4.2467,
+      "learning_rate": 1.8404781533388294e-05,
+      "loss": 1.6972,
       "step": 4460
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018425391591096456,
-      "loss": 3.8718,
+      "learning_rate": 1.838417147568013e-05,
+      "loss": 1.3979,
       "step": 4461
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018404781533388294,
-      "loss": 3.6873,
+      "learning_rate": 1.8363561417971974e-05,
+      "loss": 1.5774,
       "step": 4462
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018384171475680132,
-      "loss": 3.7128,
+      "learning_rate": 1.834295136026381e-05,
+      "loss": 1.3288,
       "step": 4463
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018363561417971972,
-      "loss": 3.8094,
+      "learning_rate": 1.8322341302555647e-05,
+      "loss": 1.2576,
       "step": 4464
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.0001834295136026381,
-      "loss": 3.9145,
+      "learning_rate": 1.8301731244847487e-05,
+      "loss": 1.2284,
       "step": 4465
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018322341302555648,
-      "loss": 3.9553,
+      "learning_rate": 1.8281121187139323e-05,
+      "loss": 1.2326,
       "step": 4466
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018301731244847486,
-      "loss": 3.7035,
+      "learning_rate": 1.8260511129431163e-05,
+      "loss": 1.3153,
       "step": 4467
     },
     {
       "epoch": 2.5,
-      "learning_rate": 0.00018281121187139324,
-      "loss": 3.6201,
+      "learning_rate": 1.8239901071723e-05,
+      "loss": 1.6317,
       "step": 4468
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018260511129431162,
-      "loss": 4.1343,
+      "learning_rate": 1.821929101401484e-05,
+      "loss": 2.3805,
       "step": 4469
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018239901071723,
-      "loss": 4.4102,
+      "learning_rate": 1.819868095630668e-05,
+      "loss": 2.7878,
       "step": 4470
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018219291014014838,
-      "loss": 4.172,
+      "learning_rate": 1.8178070898598517e-05,
+      "loss": 2.3234,
       "step": 4471
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.0001819868095630668,
-      "loss": 4.4094,
+      "learning_rate": 1.8157460840890357e-05,
+      "loss": 2.4884,
       "step": 4472
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018178070898598517,
-      "loss": 3.9854,
+      "learning_rate": 1.8136850783182193e-05,
+      "loss": 1.888,
       "step": 4473
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018157460840890355,
-      "loss": 4.1397,
+      "learning_rate": 1.8116240725474033e-05,
+      "loss": 2.2146,
       "step": 4474
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018136850783182193,
-      "loss": 3.8177,
+      "learning_rate": 1.809563066776587e-05,
+      "loss": 1.9691,
       "step": 4475
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.0001811624072547403,
-      "loss": 3.941,
+      "learning_rate": 1.8075020610057706e-05,
+      "loss": 2.0052,
       "step": 4476
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018095630667765868,
-      "loss": 4.1557,
+      "learning_rate": 1.805441055234955e-05,
+      "loss": 2.2441,
       "step": 4477
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018075020610057706,
-      "loss": 4.1745,
+      "learning_rate": 1.8033800494641386e-05,
+      "loss": 2.0305,
       "step": 4478
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018054410552349547,
-      "loss": 4.4811,
+      "learning_rate": 1.8013190436933226e-05,
+      "loss": 2.6942,
       "step": 4479
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018033800494641385,
-      "loss": 4.0649,
+      "learning_rate": 1.7992580379225063e-05,
+      "loss": 1.9571,
       "step": 4480
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00018013190436933226,
-      "loss": 4.1412,
+      "learning_rate": 1.79719703215169e-05,
+      "loss": 2.0478,
       "step": 4481
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00017992580379225064,
-      "loss": 4.0778,
+      "learning_rate": 1.795136026380874e-05,
+      "loss": 2.0042,
       "step": 4482
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00017971970321516902,
-      "loss": 3.8618,
+      "learning_rate": 1.7930750206100576e-05,
+      "loss": 1.8883,
       "step": 4483
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.0001795136026380874,
-      "loss": 3.8971,
+      "learning_rate": 1.7910140148392416e-05,
+      "loss": 2.0261,
       "step": 4484
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00017930750206100578,
-      "loss": 3.9643,
+      "learning_rate": 1.7889530090684256e-05,
+      "loss": 1.878,
       "step": 4485
     },
     {
       "epoch": 2.51,
-      "learning_rate": 0.00017910140148392416,
-      "loss": 4.0749,
+      "learning_rate": 1.7868920032976093e-05,
+      "loss": 2.7478,
       "step": 4486
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017889530090684256,
-      "loss": 3.9006,
+      "learning_rate": 1.7848309975267933e-05,
+      "loss": 1.8674,
       "step": 4487
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017868920032976094,
-      "loss": 4.1171,
+      "learning_rate": 1.782769991755977e-05,
+      "loss": 2.0431,
       "step": 4488
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017848309975267932,
-      "loss": 4.0268,
+      "learning_rate": 1.780708985985161e-05,
+      "loss": 1.8906,
       "step": 4489
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.0001782769991755977,
-      "loss": 4.1329,
+      "learning_rate": 1.7786479802143446e-05,
+      "loss": 2.1223,
       "step": 4490
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017807089859851608,
-      "loss": 4.3143,
+      "learning_rate": 1.7765869744435286e-05,
+      "loss": 2.239,
       "step": 4491
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017786479802143446,
-      "loss": 4.2383,
+      "learning_rate": 1.7745259686727123e-05,
+      "loss": 2.2098,
       "step": 4492
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017765869744435284,
-      "loss": 4.0743,
+      "learning_rate": 1.7724649629018963e-05,
+      "loss": 2.5242,
       "step": 4493
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017745259686727122,
-      "loss": 4.1571,
+      "learning_rate": 1.7704039571310803e-05,
+      "loss": 1.9812,
       "step": 4494
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017724649629018963,
-      "loss": 3.9069,
+      "learning_rate": 1.768342951360264e-05,
+      "loss": 1.8675,
       "step": 4495
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.000177040395713108,
-      "loss": 4.1684,
+      "learning_rate": 1.766281945589448e-05,
+      "loss": 2.0739,
       "step": 4496
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017683429513602638,
-      "loss": 4.0489,
+      "learning_rate": 1.7642209398186316e-05,
+      "loss": 1.9777,
       "step": 4497
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017662819455894476,
-      "loss": 3.9246,
+      "learning_rate": 1.7621599340478152e-05,
+      "loss": 1.7158,
       "step": 4498
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017642209398186314,
-      "loss": 3.8547,
+      "learning_rate": 1.7600989282769992e-05,
+      "loss": 2.065,
       "step": 4499
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017621599340478152,
-      "loss": 3.9933,
+      "learning_rate": 1.758037922506183e-05,
+      "loss": 1.8814,
       "step": 4500
     },
     {
       "epoch": 2.52,
-      "eval_loss": 5.2800092697143555,
-      "eval_runtime": 1256.1512,
-      "eval_samples_per_second": 2.103,
-      "eval_steps_per_second": 0.264,
-      "eval_wer": 1.9741372471241572,
+      "eval_loss": 6.9818243980407715,
+      "eval_runtime": 1257.2152,
+      "eval_samples_per_second": 2.101,
+      "eval_steps_per_second": 0.263,
+      "eval_wer": 1.9831614438714795,
       "step": 4500
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.0001760098928276999,
-      "loss": 4.2388,
+      "learning_rate": 1.7559769167353672e-05,
+      "loss": 2.44,
       "step": 4501
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.00017580379225061828,
-      "loss": 4.0934,
+      "learning_rate": 1.753915910964551e-05,
+      "loss": 1.9161,
       "step": 4502
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.0001755976916735367,
-      "loss": 4.4195,
+      "learning_rate": 1.7518549051937346e-05,
+      "loss": 2.2991,
       "step": 4503
     },
     {
       "epoch": 2.52,
-      "learning_rate": 0.0001753915910964551,
-      "loss": 3.9851,
+      "learning_rate": 1.7497938994229186e-05,
+      "loss": 1.8165,
       "step": 4504
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017518549051937348,
-      "loss": 4.2827,
+      "learning_rate": 1.7477328936521022e-05,
+      "loss": 1.8859,
       "step": 4505
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017497938994229186,
-      "loss": 3.7112,
+      "learning_rate": 1.7456718878812862e-05,
+      "loss": 1.7076,
       "step": 4506
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017477328936521023,
-      "loss": 3.8002,
+      "learning_rate": 1.74361088211047e-05,
+      "loss": 1.6407,
       "step": 4507
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017456718878812861,
-      "loss": 4.237,
+      "learning_rate": 1.741549876339654e-05,
+      "loss": 1.9023,
       "step": 4508
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.000174361088211047,
-      "loss": 4.2165,
+      "learning_rate": 1.739488870568838e-05,
+      "loss": 2.106,
       "step": 4509
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017415498763396537,
-      "loss": 3.9822,
+      "learning_rate": 1.7374278647980215e-05,
+      "loss": 1.4409,
       "step": 4510
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017394888705688378,
-      "loss": 3.8946,
+      "learning_rate": 1.7353668590272055e-05,
+      "loss": 1.5518,
       "step": 4511
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017374278647980216,
-      "loss": 4.0408,
+      "learning_rate": 1.7333058532563892e-05,
+      "loss": 1.5434,
       "step": 4512
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017353668590272054,
-      "loss": 4.0769,
+      "learning_rate": 1.7312448474855732e-05,
+      "loss": 1.9354,
       "step": 4513
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017333058532563892,
-      "loss": 4.1706,
+      "learning_rate": 1.729183841714757e-05,
+      "loss": 1.6905,
       "step": 4514
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.0001731244847485573,
-      "loss": 3.707,
+      "learning_rate": 1.7271228359439405e-05,
+      "loss": 1.2417,
       "step": 4515
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017291838417147568,
-      "loss": 3.446,
+      "learning_rate": 1.7250618301731245e-05,
+      "loss": 1.0342,
       "step": 4516
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017271228359439406,
-      "loss": 3.3464,
+      "learning_rate": 1.7230008244023085e-05,
+      "loss": 1.0935,
       "step": 4517
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017250618301731244,
-      "loss": 3.0425,
+      "learning_rate": 1.7209398186314925e-05,
+      "loss": 1.2863,
       "step": 4518
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017230008244023084,
-      "loss": 4.6893,
+      "learning_rate": 1.718878812860676e-05,
+      "loss": 3.0449,
       "step": 4519
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017209398186314922,
-      "loss": 4.1628,
+      "learning_rate": 1.7168178070898598e-05,
+      "loss": 2.452,
       "step": 4520
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.0001718878812860676,
-      "loss": 4.3683,
+      "learning_rate": 1.7147568013190438e-05,
+      "loss": 2.8919,
       "step": 4521
     },
     {
       "epoch": 2.53,
-      "learning_rate": 0.00017168178070898598,
-      "loss": 4.0764,
+      "learning_rate": 1.7126957955482275e-05,
+      "loss": 2.1079,
       "step": 4522
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00017147568013190436,
-      "loss": 3.9563,
+      "learning_rate": 1.7106347897774115e-05,
+      "loss": 2.2155,
       "step": 4523
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00017126957955482274,
-      "loss": 4.2778,
+      "learning_rate": 1.708573784006595e-05,
+      "loss": 2.3448,
       "step": 4524
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00017106347897774112,
-      "loss": 4.1989,
+      "learning_rate": 1.706512778235779e-05,
+      "loss": 2.8065,
       "step": 4525
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.0001708573784006595,
-      "loss": 4.2573,
+      "learning_rate": 1.704451772464963e-05,
+      "loss": 2.3611,
       "step": 4526
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00017065127782357793,
-      "loss": 4.2192,
+      "learning_rate": 1.7023907666941468e-05,
+      "loss": 2.201,
       "step": 4527
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00017044517724649631,
-      "loss": 4.3957,
+      "learning_rate": 1.7003297609233308e-05,
+      "loss": 2.5528,
       "step": 4528
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.0001702390766694147,
-      "loss": 3.9751,
+      "learning_rate": 1.6982687551525145e-05,
+      "loss": 2.0415,
       "step": 4529
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00017003297609233307,
-      "loss": 4.1727,
+      "learning_rate": 1.6962077493816985e-05,
+      "loss": 2.1405,
       "step": 4530
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00016982687551525145,
-      "loss": 4.1672,
+      "learning_rate": 1.694146743610882e-05,
+      "loss": 2.2039,
       "step": 4531
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00016962077493816983,
-      "loss": 4.1682,
+      "learning_rate": 1.6920857378400658e-05,
+      "loss": 2.0506,
       "step": 4532
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.0001694146743610882,
-      "loss": 3.8833,
+      "learning_rate": 1.69002473206925e-05,
+      "loss": 1.8635,
       "step": 4533
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.0001692085737840066,
-      "loss": 4.1081,
+      "learning_rate": 1.6879637262984338e-05,
+      "loss": 2.0127,
       "step": 4534
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.000169002473206925,
-      "loss": 3.7789,
+      "learning_rate": 1.6859027205276174e-05,
+      "loss": 1.7494,
       "step": 4535
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00016879637262984338,
-      "loss": 4.2123,
+      "learning_rate": 1.6838417147568014e-05,
+      "loss": 2.5564,
       "step": 4536
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00016859027205276176,
-      "loss": 4.4283,
+      "learning_rate": 1.681780708985985e-05,
+      "loss": 2.7096,
       "step": 4537
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00016838417147568014,
-      "loss": 4.5046,
+      "learning_rate": 1.679719703215169e-05,
+      "loss": 3.1452,
       "step": 4538
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.00016817807089859852,
-      "loss": 4.1229,
+      "learning_rate": 1.6776586974443528e-05,
+      "loss": 2.5011,
       "step": 4539
     },
     {
       "epoch": 2.54,
-      "learning_rate": 0.0001679719703215169,
-      "loss": 4.0794,
+      "learning_rate": 1.6755976916735368e-05,
+      "loss": 2.1067,
       "step": 4540
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016776586974443528,
-      "loss": 4.0489,
+      "learning_rate": 1.6735366859027208e-05,
+      "loss": 2.0121,
       "step": 4541
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016755976916735366,
-      "loss": 3.844,
+      "learning_rate": 1.6714756801319044e-05,
+      "loss": 2.2198,
       "step": 4542
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016735366859027206,
-      "loss": 4.0522,
+      "learning_rate": 1.6694146743610884e-05,
+      "loss": 2.1501,
       "step": 4543
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016714756801319044,
-      "loss": 4.0933,
+      "learning_rate": 1.667353668590272e-05,
+      "loss": 2.0194,
       "step": 4544
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016694146743610882,
-      "loss": 3.9495,
+      "learning_rate": 1.665292662819456e-05,
+      "loss": 1.7786,
       "step": 4545
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.0001667353668590272,
-      "loss": 3.7867,
+      "learning_rate": 1.6632316570486397e-05,
+      "loss": 2.3264,
       "step": 4546
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016652926628194558,
-      "loss": 4.242,
+      "learning_rate": 1.6611706512778234e-05,
+      "loss": 2.1897,
       "step": 4547
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016632316570486396,
-      "loss": 4.3095,
+      "learning_rate": 1.6591096455070074e-05,
+      "loss": 2.4573,
       "step": 4548
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016611706512778234,
-      "loss": 3.9004,
+      "learning_rate": 1.6570486397361914e-05,
+      "loss": 2.1512,
       "step": 4549
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016591096455070072,
-      "loss": 4.032,
+      "learning_rate": 1.6549876339653754e-05,
+      "loss": 1.9334,
       "step": 4550
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016570486397361915,
-      "loss": 4.0965,
+      "learning_rate": 1.652926628194559e-05,
+      "loss": 1.9523,
       "step": 4551
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016549876339653753,
-      "loss": 4.0481,
+      "learning_rate": 1.6508656224237427e-05,
+      "loss": 1.9884,
       "step": 4552
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.0001652926628194559,
-      "loss": 3.9588,
+      "learning_rate": 1.6488046166529267e-05,
+      "loss": 1.9701,
       "step": 4553
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.0001650865622423743,
-      "loss": 4.219,
+      "learning_rate": 1.6467436108821104e-05,
+      "loss": 2.4332,
       "step": 4554
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016488046166529267,
-      "loss": 3.8771,
+      "learning_rate": 1.6446826051112944e-05,
+      "loss": 1.6958,
       "step": 4555
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016467436108821105,
-      "loss": 3.8846,
+      "learning_rate": 1.642621599340478e-05,
+      "loss": 1.6358,
       "step": 4556
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.00016446826051112943,
-      "loss": 3.9496,
+      "learning_rate": 1.640560593569662e-05,
+      "loss": 1.8517,
       "step": 4557
     },
     {
       "epoch": 2.55,
-      "learning_rate": 0.0001642621599340478,
-      "loss": 3.8547,
+      "learning_rate": 1.638499587798846e-05,
+      "loss": 1.6737,
       "step": 4558
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016405605935696622,
-      "loss": 3.7413,
+      "learning_rate": 1.6364385820280297e-05,
+      "loss": 1.5227,
       "step": 4559
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.0001638499587798846,
-      "loss": 3.8909,
+      "learning_rate": 1.6343775762572137e-05,
+      "loss": 1.5572,
       "step": 4560
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016364385820280298,
-      "loss": 3.9593,
+      "learning_rate": 1.6323165704863974e-05,
+      "loss": 1.6074,
       "step": 4561
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016343775762572136,
-      "loss": 3.7484,
+      "learning_rate": 1.6302555647155814e-05,
+      "loss": 1.4111,
       "step": 4562
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016323165704863974,
-      "loss": 3.6006,
+      "learning_rate": 1.628194558944765e-05,
+      "loss": 1.1831,
       "step": 4563
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016302555647155812,
-      "loss": 3.73,
+      "learning_rate": 1.6261335531739487e-05,
+      "loss": 1.2658,
       "step": 4564
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.0001628194558944765,
-      "loss": 3.7694,
+      "learning_rate": 1.624072547403133e-05,
+      "loss": 1.3683,
       "step": 4565
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016261335531739487,
-      "loss": 3.6685,
+      "learning_rate": 1.6220115416323167e-05,
+      "loss": 1.1529,
       "step": 4566
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016240725474031328,
-      "loss": 3.2225,
+      "learning_rate": 1.6199505358615007e-05,
+      "loss": 1.3049,
       "step": 4567
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016220115416323166,
-      "loss": 3.7021,
+      "learning_rate": 1.6178895300906843e-05,
+      "loss": 1.5477,
       "step": 4568
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016199505358615004,
-      "loss": 3.9801,
+      "learning_rate": 1.615828524319868e-05,
+      "loss": 2.2551,
       "step": 4569
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016178895300906842,
-      "loss": 4.0441,
+      "learning_rate": 1.613767518549052e-05,
+      "loss": 2.3971,
       "step": 4570
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.0001615828524319868,
-      "loss": 4.1885,
+      "learning_rate": 1.6117065127782357e-05,
+      "loss": 2.4964,
       "step": 4571
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016137675185490518,
-      "loss": 3.9689,
+      "learning_rate": 1.6096455070074197e-05,
+      "loss": 2.2397,
       "step": 4572
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016117065127782356,
-      "loss": 4.1468,
+      "learning_rate": 1.6075845012366037e-05,
+      "loss": 2.1579,
       "step": 4573
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016096455070074197,
-      "loss": 4.0761,
+      "learning_rate": 1.6055234954657873e-05,
+      "loss": 2.1344,
       "step": 4574
     },
     {
       "epoch": 2.56,
-      "learning_rate": 0.00016075845012366037,
-      "loss": 3.9796,
+      "learning_rate": 1.6034624896949713e-05,
+      "loss": 2.2595,
       "step": 4575
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00016055234954657875,
-      "loss": 4.192,
+      "learning_rate": 1.601401483924155e-05,
+      "loss": 2.4596,
       "step": 4576
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00016034624896949713,
-      "loss": 4.1491,
+      "learning_rate": 1.599340478153339e-05,
+      "loss": 2.4691,
       "step": 4577
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.0001601401483924155,
-      "loss": 4.1111,
+      "learning_rate": 1.5972794723825226e-05,
+      "loss": 2.2734,
       "step": 4578
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.0001599340478153339,
-      "loss": 4.1536,
+      "learning_rate": 1.5952184666117066e-05,
+      "loss": 2.2523,
       "step": 4579
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015972794723825227,
-      "loss": 3.8145,
+      "learning_rate": 1.5931574608408903e-05,
+      "loss": 1.9953,
       "step": 4580
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015952184666117065,
-      "loss": 4.2785,
+      "learning_rate": 1.5910964550700743e-05,
+      "loss": 2.83,
       "step": 4581
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015931574608408903,
-      "loss": 3.8792,
+      "learning_rate": 1.5890354492992583e-05,
+      "loss": 1.9884,
       "step": 4582
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015910964550700744,
-      "loss": 4.0433,
+      "learning_rate": 1.586974443528442e-05,
+      "loss": 2.0386,
       "step": 4583
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015890354492992582,
-      "loss": 4.0371,
+      "learning_rate": 1.584913437757626e-05,
+      "loss": 2.203,
       "step": 4584
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.0001586974443528442,
-      "loss": 4.3614,
+      "learning_rate": 1.5828524319868096e-05,
+      "loss": 2.4379,
       "step": 4585
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015849134377576257,
-      "loss": 4.0529,
+      "learning_rate": 1.5807914262159933e-05,
+      "loss": 2.512,
       "step": 4586
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015828524319868095,
-      "loss": 4.5024,
+      "learning_rate": 1.5787304204451773e-05,
+      "loss": 2.7932,
       "step": 4587
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015807914262159933,
-      "loss": 3.8636,
+      "learning_rate": 1.5766694146743613e-05,
+      "loss": 2.2195,
       "step": 4588
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.0001578730420445177,
-      "loss": 4.177,
+      "learning_rate": 1.5746084089035453e-05,
+      "loss": 2.3924,
       "step": 4589
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015766694146743612,
-      "loss": 4.0446,
+      "learning_rate": 1.572547403132729e-05,
+      "loss": 2.0153,
       "step": 4590
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.0001574608408903545,
-      "loss": 4.1794,
+      "learning_rate": 1.5704863973619126e-05,
+      "loss": 2.1608,
       "step": 4591
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015725474031327288,
-      "loss": 4.2114,
+      "learning_rate": 1.5684253915910966e-05,
+      "loss": 2.5149,
       "step": 4592
     },
     {
       "epoch": 2.57,
-      "learning_rate": 0.00015704863973619126,
-      "loss": 3.9537,
+      "learning_rate": 1.5663643858202802e-05,
+      "loss": 1.9645,
       "step": 4593
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015684253915910964,
-      "loss": 4.1271,
+      "learning_rate": 1.5643033800494642e-05,
+      "loss": 2.1042,
       "step": 4594
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015663643858202802,
-      "loss": 4.0257,
+      "learning_rate": 1.562242374278648e-05,
+      "loss": 1.8466,
       "step": 4595
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0001564303380049464,
-      "loss": 3.9439,
+      "learning_rate": 1.560181368507832e-05,
+      "loss": 2.5588,
       "step": 4596
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015622423742786478,
-      "loss": 3.8106,
+      "learning_rate": 1.558120362737016e-05,
+      "loss": 1.7943,
       "step": 4597
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0001560181368507832,
-      "loss": 4.206,
+      "learning_rate": 1.5560593569661996e-05,
+      "loss": 2.0667,
       "step": 4598
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0001558120362737016,
-      "loss": 3.8263,
+      "learning_rate": 1.5539983511953836e-05,
+      "loss": 2.0547,
       "step": 4599
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015560593569661997,
-      "loss": 4.1825,
+      "learning_rate": 1.5519373454245672e-05,
+      "loss": 2.4097,
       "step": 4600
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015539983511953835,
-      "loss": 4.086,
+      "learning_rate": 1.5498763396537512e-05,
+      "loss": 1.9154,
       "step": 4601
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015519373454245673,
-      "loss": 3.6768,
+      "learning_rate": 1.547815333882935e-05,
+      "loss": 1.7396,
       "step": 4602
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0001549876339653751,
-      "loss": 4.1132,
+      "learning_rate": 1.5457543281121185e-05,
+      "loss": 2.0391,
       "step": 4603
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0001547815333882935,
-      "loss": 3.7393,
+      "learning_rate": 1.543693322341303e-05,
+      "loss": 2.1192,
       "step": 4604
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015457543281121187,
-      "loss": 4.2305,
+      "learning_rate": 1.5416323165704865e-05,
+      "loss": 2.234,
       "step": 4605
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015436933223413027,
-      "loss": 4.2687,
+      "learning_rate": 1.5395713107996705e-05,
+      "loss": 2.6558,
       "step": 4606
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015416323165704865,
-      "loss": 4.051,
+      "learning_rate": 1.5375103050288542e-05,
+      "loss": 2.1539,
       "step": 4607
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015395713107996703,
-      "loss": 3.8112,
+      "learning_rate": 1.535449299258038e-05,
+      "loss": 1.6205,
       "step": 4608
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0001537510305028854,
-      "loss": 4.1069,
+      "learning_rate": 1.533388293487222e-05,
+      "loss": 2.0586,
       "step": 4609
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.0001535449299258038,
-      "loss": 4.158,
+      "learning_rate": 1.5313272877164055e-05,
+      "loss": 1.8729,
       "step": 4610
     },
     {
       "epoch": 2.58,
-      "learning_rate": 0.00015333882934872217,
-      "loss": 3.9009,
+      "learning_rate": 1.5292662819455895e-05,
+      "loss": 2.0609,
       "step": 4611
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015313272877164055,
-      "loss": 3.7662,
+      "learning_rate": 1.5272052761747735e-05,
+      "loss": 1.8058,
       "step": 4612
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015292662819455893,
-      "loss": 4.091,
+      "learning_rate": 1.5251442704039573e-05,
+      "loss": 1.5024,
       "step": 4613
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015272052761747734,
-      "loss": 3.328,
+      "learning_rate": 1.5230832646331412e-05,
+      "loss": 1.1933,
       "step": 4614
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015251442704039572,
-      "loss": 3.6052,
+      "learning_rate": 1.5210222588623248e-05,
+      "loss": 1.2381,
       "step": 4615
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.0001523083264633141,
-      "loss": 3.556,
+      "learning_rate": 1.5189612530915087e-05,
+      "loss": 1.3897,
       "step": 4616
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015210222588623248,
-      "loss": 3.6166,
+      "learning_rate": 1.5169002473206925e-05,
+      "loss": 1.3183,
       "step": 4617
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015189612530915086,
-      "loss": 2.9811,
+      "learning_rate": 1.5148392415498763e-05,
+      "loss": 1.3549,
       "step": 4618
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015169002473206924,
-      "loss": 4.8348,
+      "learning_rate": 1.5127782357790602e-05,
+      "loss": 3.4596,
       "step": 4619
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015148392415498762,
-      "loss": 4.3396,
+      "learning_rate": 1.5107172300082442e-05,
+      "loss": 2.6094,
       "step": 4620
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015127782357790602,
-      "loss": 4.0664,
+      "learning_rate": 1.508656224237428e-05,
+      "loss": 2.8527,
       "step": 4621
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015107172300082443,
-      "loss": 4.1203,
+      "learning_rate": 1.5065952184666118e-05,
+      "loss": 2.5059,
       "step": 4622
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.0001508656224237428,
-      "loss": 4.2238,
+      "learning_rate": 1.5045342126957956e-05,
+      "loss": 3.0299,
       "step": 4623
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.0001506595218466612,
-      "loss": 4.2009,
+      "learning_rate": 1.5024732069249795e-05,
+      "loss": 2.6882,
       "step": 4624
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015045342126957957,
-      "loss": 4.2993,
+      "learning_rate": 1.5004122011541633e-05,
+      "loss": 2.3623,
       "step": 4625
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015024732069249795,
-      "loss": 4.2291,
+      "learning_rate": 1.4983511953833471e-05,
+      "loss": 2.6293,
       "step": 4626
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00015004122011541633,
-      "loss": 4.2534,
+      "learning_rate": 1.4962901896125308e-05,
+      "loss": 2.4881,
       "step": 4627
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.0001498351195383347,
-      "loss": 4.2264,
+      "learning_rate": 1.494229183841715e-05,
+      "loss": 2.3836,
       "step": 4628
     },
     {
       "epoch": 2.59,
-      "learning_rate": 0.00014962901896125309,
-      "loss": 4.4302,
+      "learning_rate": 1.4921681780708988e-05,
+      "loss": 2.7185,
       "step": 4629
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.0001494229183841715,
-      "loss": 4.1627,
+      "learning_rate": 1.4901071723000826e-05,
+      "loss": 2.5286,
       "step": 4630
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014921681780708987,
-      "loss": 4.0143,
+      "learning_rate": 1.4880461665292665e-05,
+      "loss": 2.3255,
       "step": 4631
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014901071723000825,
-      "loss": 4.201,
+      "learning_rate": 1.4859851607584501e-05,
+      "loss": 2.3807,
       "step": 4632
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014880461665292663,
-      "loss": 4.0288,
+      "learning_rate": 1.483924154987634e-05,
+      "loss": 2.1842,
       "step": 4633
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.000148598516075845,
-      "loss": 4.0679,
+      "learning_rate": 1.4818631492168178e-05,
+      "loss": 2.2449,
       "step": 4634
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.0001483924154987634,
-      "loss": 3.9816,
+      "learning_rate": 1.4798021434460016e-05,
+      "loss": 2.0377,
       "step": 4635
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014818631492168177,
-      "loss": 4.0635,
+      "learning_rate": 1.4777411376751856e-05,
+      "loss": 2.147,
       "step": 4636
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014798021434460015,
-      "loss": 4.0209,
+      "learning_rate": 1.4756801319043694e-05,
+      "loss": 2.1265,
       "step": 4637
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014777411376751856,
-      "loss": 4.093,
+      "learning_rate": 1.4736191261335533e-05,
+      "loss": 2.2749,
       "step": 4638
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014756801319043694,
-      "loss": 4.1456,
+      "learning_rate": 1.4715581203627371e-05,
+      "loss": 2.5828,
       "step": 4639
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014736191261335532,
-      "loss": 4.0307,
+      "learning_rate": 1.469497114591921e-05,
+      "loss": 2.4312,
       "step": 4640
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.0001471558120362737,
-      "loss": 3.783,
+      "learning_rate": 1.4674361088211048e-05,
+      "loss": 1.8591,
       "step": 4641
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014694971145919208,
-      "loss": 4.1906,
+      "learning_rate": 1.4653751030502886e-05,
+      "loss": 2.2154,
       "step": 4642
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014674361088211045,
-      "loss": 4.3647,
+      "learning_rate": 1.4633140972794724e-05,
+      "loss": 2.4599,
       "step": 4643
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014653751030502886,
-      "loss": 3.9119,
+      "learning_rate": 1.4612530915086564e-05,
+      "loss": 2.0866,
       "step": 4644
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014633140972794724,
-      "loss": 4.0297,
+      "learning_rate": 1.4591920857378402e-05,
+      "loss": 2.1685,
       "step": 4645
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014612530915086565,
-      "loss": 4.3018,
+      "learning_rate": 1.457131079967024e-05,
+      "loss": 2.8488,
       "step": 4646
     },
     {
       "epoch": 2.6,
-      "learning_rate": 0.00014591920857378403,
-      "loss": 4.0806,
+      "learning_rate": 1.4550700741962079e-05,
+      "loss": 2.2057,
       "step": 4647
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0001457131079967024,
-      "loss": 3.9516,
+      "learning_rate": 1.4530090684253916e-05,
+      "loss": 2.4797,
       "step": 4648
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0001455070074196208,
-      "loss": 4.1027,
+      "learning_rate": 1.4509480626545754e-05,
+      "loss": 2.145,
       "step": 4649
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014530090684253917,
-      "loss": 3.7111,
+      "learning_rate": 1.4488870568837592e-05,
+      "loss": 2.1727,
       "step": 4650
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014509480626545755,
-      "loss": 4.0773,
+      "learning_rate": 1.446826051112943e-05,
+      "loss": 2.2786,
       "step": 4651
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014488870568837593,
-      "loss": 3.7453,
+      "learning_rate": 1.4447650453421272e-05,
+      "loss": 1.8147,
       "step": 4652
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0001446826051112943,
-      "loss": 4.2575,
+      "learning_rate": 1.4427040395713109e-05,
+      "loss": 2.2661,
       "step": 4653
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0001444765045342127,
-      "loss": 4.1788,
+      "learning_rate": 1.4406430338004947e-05,
+      "loss": 2.0108,
       "step": 4654
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0001442704039571311,
-      "loss": 3.8356,
+      "learning_rate": 1.4385820280296785e-05,
+      "loss": 1.7158,
       "step": 4655
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014406430338004947,
-      "loss": 4.2063,
+      "learning_rate": 1.4365210222588624e-05,
+      "loss": 2.2087,
       "step": 4656
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014385820280296785,
-      "loss": 3.9853,
+      "learning_rate": 1.4344600164880462e-05,
+      "loss": 1.8459,
       "step": 4657
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014365210222588623,
-      "loss": 3.9686,
+      "learning_rate": 1.43239901071723e-05,
+      "loss": 2.2225,
       "step": 4658
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.0001434460016488046,
-      "loss": 3.9942,
+      "learning_rate": 1.4303380049464139e-05,
+      "loss": 1.8139,
       "step": 4659
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.000143239901071723,
-      "loss": 3.837,
+      "learning_rate": 1.4282769991755979e-05,
+      "loss": 1.7926,
       "step": 4660
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014303380049464137,
-      "loss": 3.8776,
+      "learning_rate": 1.4262159934047817e-05,
+      "loss": 1.6155,
       "step": 4661
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014282769991755978,
-      "loss": 3.7688,
+      "learning_rate": 1.4241549876339655e-05,
+      "loss": 1.8042,
       "step": 4662
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014262159934047815,
-      "loss": 3.4921,
+      "learning_rate": 1.4220939818631493e-05,
+      "loss": 1.3079,
       "step": 4663
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014241549876339653,
-      "loss": 4.0959,
+      "learning_rate": 1.4200329760923332e-05,
+      "loss": 1.9987,
       "step": 4664
     },
     {
       "epoch": 2.61,
-      "learning_rate": 0.00014220939818631491,
-      "loss": 3.5117,
+      "learning_rate": 1.4179719703215168e-05,
+      "loss": 1.3167,
       "step": 4665
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.0001420032976092333,
-      "loss": 3.4592,
+      "learning_rate": 1.4159109645507007e-05,
+      "loss": 1.202,
       "step": 4666
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.0001417971970321517,
-      "loss": 3.7219,
+      "learning_rate": 1.4138499587798845e-05,
+      "loss": 1.2504,
       "step": 4667
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014159109645507008,
-      "loss": 3.4239,
+      "learning_rate": 1.4117889530090687e-05,
+      "loss": 1.5257,
       "step": 4668
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014138499587798846,
-      "loss": 4.2929,
+      "learning_rate": 1.4097279472382525e-05,
+      "loss": 2.739,
       "step": 4669
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014117889530090687,
-      "loss": 4.2338,
+      "learning_rate": 1.4076669414674362e-05,
+      "loss": 2.9412,
       "step": 4670
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014097279472382525,
-      "loss": 4.3697,
+      "learning_rate": 1.40560593569662e-05,
+      "loss": 2.946,
       "step": 4671
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014076669414674363,
-      "loss": 4.1693,
+      "learning_rate": 1.4035449299258038e-05,
+      "loss": 2.5063,
       "step": 4672
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.000140560593569662,
-      "loss": 3.8348,
+      "learning_rate": 1.4014839241549876e-05,
+      "loss": 2.1487,
       "step": 4673
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014035449299258038,
-      "loss": 4.2477,
+      "learning_rate": 1.3994229183841715e-05,
+      "loss": 2.3875,
       "step": 4674
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00014014839241549876,
-      "loss": 4.0977,
+      "learning_rate": 1.3973619126133553e-05,
+      "loss": 2.3105,
       "step": 4675
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00013994229183841714,
-      "loss": 4.1638,
+      "learning_rate": 1.3953009068425393e-05,
+      "loss": 2.3391,
       "step": 4676
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00013973619126133552,
-      "loss": 3.9306,
+      "learning_rate": 1.3932399010717231e-05,
+      "loss": 2.6366,
       "step": 4677
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00013953009068425393,
-      "loss": 4.0746,
+      "learning_rate": 1.391178895300907e-05,
+      "loss": 2.7296,
       "step": 4678
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.0001393239901071723,
-      "loss": 4.264,
+      "learning_rate": 1.3891178895300908e-05,
+      "loss": 2.504,
       "step": 4679
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.0001391178895300907,
-      "loss": 4.2246,
+      "learning_rate": 1.3870568837592746e-05,
+      "loss": 2.3954,
       "step": 4680
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00013891178895300907,
-      "loss": 4.1687,
+      "learning_rate": 1.3849958779884584e-05,
+      "loss": 2.6751,
       "step": 4681
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00013870568837592745,
-      "loss": 3.8738,
+      "learning_rate": 1.3829348722176421e-05,
+      "loss": 2.074,
       "step": 4682
     },
     {
       "epoch": 2.62,
-      "learning_rate": 0.00013849958779884583,
-      "loss": 4.1893,
+      "learning_rate": 1.380873866446826e-05,
+      "loss": 2.3076,
       "step": 4683
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.0001382934872217642,
-      "loss": 3.9788,
+      "learning_rate": 1.3788128606760101e-05,
+      "loss": 2.162,
       "step": 4684
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.0001380873866446826,
-      "loss": 4.1328,
+      "learning_rate": 1.376751854905194e-05,
+      "loss": 2.1683,
       "step": 4685
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.000137881286067601,
-      "loss": 4.0293,
+      "learning_rate": 1.3746908491343776e-05,
+      "loss": 2.076,
       "step": 4686
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013767518549051937,
-      "loss": 4.1305,
+      "learning_rate": 1.3726298433635614e-05,
+      "loss": 2.2365,
       "step": 4687
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013746908491343775,
-      "loss": 3.8114,
+      "learning_rate": 1.3705688375927453e-05,
+      "loss": 2.3023,
       "step": 4688
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013726298433635613,
-      "loss": 4.2089,
+      "learning_rate": 1.3685078318219291e-05,
+      "loss": 3.0006,
       "step": 4689
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.0001370568837592745,
-      "loss": 3.8248,
+      "learning_rate": 1.366446826051113e-05,
+      "loss": 1.975,
       "step": 4690
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013685078318219292,
-      "loss": 4.4089,
+      "learning_rate": 1.3643858202802967e-05,
+      "loss": 2.868,
       "step": 4691
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.0001366446826051113,
-      "loss": 4.1576,
+      "learning_rate": 1.3623248145094807e-05,
+      "loss": 2.2646,
       "step": 4692
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013643858202802968,
-      "loss": 4.0801,
+      "learning_rate": 1.3602638087386646e-05,
+      "loss": 2.2288,
       "step": 4693
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013623248145094808,
-      "loss": 4.061,
+      "learning_rate": 1.3582028029678484e-05,
+      "loss": 2.3675,
       "step": 4694
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013602638087386646,
-      "loss": 3.9781,
+      "learning_rate": 1.3561417971970322e-05,
+      "loss": 2.1079,
       "step": 4695
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013582028029678484,
-      "loss": 4.1386,
+      "learning_rate": 1.354080791426216e-05,
+      "loss": 2.2927,
       "step": 4696
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013561417971970322,
-      "loss": 3.952,
+      "learning_rate": 1.3520197856553999e-05,
+      "loss": 2.3137,
       "step": 4697
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.0001354080791426216,
-      "loss": 4.3411,
+      "learning_rate": 1.3499587798845836e-05,
+      "loss": 2.481,
       "step": 4698
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013520197856553998,
-      "loss": 4.1418,
+      "learning_rate": 1.3478977741137677e-05,
+      "loss": 2.3454,
       "step": 4699
     },
     {
       "epoch": 2.63,
-      "learning_rate": 0.00013499587798845836,
-      "loss": 3.9478,
+      "learning_rate": 1.3458367683429516e-05,
+      "loss": 2.1182,
       "step": 4700
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013478977741137677,
-      "loss": 4.02,
+      "learning_rate": 1.3437757625721354e-05,
+      "loss": 2.1436,
       "step": 4701
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013458367683429515,
-      "loss": 3.8414,
+      "learning_rate": 1.3417147568013192e-05,
+      "loss": 2.0027,
       "step": 4702
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013437757625721353,
-      "loss": 4.0611,
+      "learning_rate": 1.3396537510305029e-05,
+      "loss": 2.4432,
       "step": 4703
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.0001341714756801319,
-      "loss": 4.0525,
+      "learning_rate": 1.3375927452596867e-05,
+      "loss": 2.011,
       "step": 4704
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.0001339653751030503,
-      "loss": 4.1811,
+      "learning_rate": 1.3355317394888705e-05,
+      "loss": 2.132,
       "step": 4705
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013375927452596867,
-      "loss": 4.0968,
+      "learning_rate": 1.3334707337180544e-05,
+      "loss": 1.9601,
       "step": 4706
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013355317394888705,
-      "loss": 4.0181,
+      "learning_rate": 1.3314097279472385e-05,
+      "loss": 1.9877,
       "step": 4707
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013334707337180543,
-      "loss": 4.1776,
+      "learning_rate": 1.3293487221764222e-05,
+      "loss": 2.3213,
       "step": 4708
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013314097279472383,
-      "loss": 3.9946,
+      "learning_rate": 1.327287716405606e-05,
+      "loss": 2.3675,
       "step": 4709
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.0001329348722176422,
-      "loss": 4.0251,
+      "learning_rate": 1.3252267106347899e-05,
+      "loss": 1.7346,
       "step": 4710
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.0001327287716405606,
-      "loss": 3.7346,
+      "learning_rate": 1.3231657048639737e-05,
+      "loss": 1.6118,
       "step": 4711
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013252267106347897,
-      "loss": 3.7563,
+      "learning_rate": 1.3211046990931575e-05,
+      "loss": 1.5287,
       "step": 4712
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013231657048639735,
-      "loss": 3.5387,
+      "learning_rate": 1.3190436933223413e-05,
+      "loss": 1.3988,
       "step": 4713
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013211046990931576,
-      "loss": 3.6172,
+      "learning_rate": 1.3169826875515252e-05,
+      "loss": 1.2769,
       "step": 4714
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013190436933223414,
-      "loss": 3.3348,
+      "learning_rate": 1.3149216817807092e-05,
+      "loss": 1.1809,
       "step": 4715
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013169826875515252,
-      "loss": 3.7602,
+      "learning_rate": 1.312860676009893e-05,
+      "loss": 1.4263,
       "step": 4716
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.00013149216817807092,
-      "loss": 4.1846,
+      "learning_rate": 1.3107996702390768e-05,
+      "loss": 1.6885,
       "step": 4717
     },
     {
       "epoch": 2.64,
-      "learning_rate": 0.0001312860676009893,
-      "loss": 3.5609,
+      "learning_rate": 1.3087386644682607e-05,
+      "loss": 1.4553,
       "step": 4718
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00013107996702390768,
-      "loss": 4.5563,
+      "learning_rate": 1.3066776586974445e-05,
+      "loss": 3.0733,
       "step": 4719
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00013087386644682606,
-      "loss": 4.1744,
+      "learning_rate": 1.3046166529266281e-05,
+      "loss": 2.491,
       "step": 4720
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00013066776586974444,
-      "loss": 4.0958,
+      "learning_rate": 1.302555647155812e-05,
+      "loss": 2.395,
       "step": 4721
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00013046166529266282,
-      "loss": 4.094,
+      "learning_rate": 1.3004946413849958e-05,
+      "loss": 2.3881,
       "step": 4722
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.0001302555647155812,
-      "loss": 4.0284,
+      "learning_rate": 1.29843363561418e-05,
+      "loss": 2.4895,
       "step": 4723
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00013004946413849958,
-      "loss": 4.5333,
+      "learning_rate": 1.2963726298433636e-05,
+      "loss": 2.7021,
       "step": 4724
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.000129843363561418,
-      "loss": 4.3174,
+      "learning_rate": 1.2943116240725475e-05,
+      "loss": 2.5576,
       "step": 4725
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012963726298433637,
-      "loss": 4.1769,
+      "learning_rate": 1.2922506183017313e-05,
+      "loss": 2.5773,
       "step": 4726
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012943116240725475,
-      "loss": 3.9789,
+      "learning_rate": 1.2901896125309151e-05,
+      "loss": 2.2503,
       "step": 4727
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012922506183017313,
-      "loss": 3.9255,
+      "learning_rate": 1.288128606760099e-05,
+      "loss": 2.4327,
       "step": 4728
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.0001290189612530915,
-      "loss": 4.257,
+      "learning_rate": 1.2860676009892828e-05,
+      "loss": 2.4384,
       "step": 4729
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012881286067600989,
-      "loss": 4.3208,
+      "learning_rate": 1.2840065952184666e-05,
+      "loss": 2.6023,
       "step": 4730
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012860676009892826,
-      "loss": 3.8751,
+      "learning_rate": 1.2819455894476506e-05,
+      "loss": 2.1335,
       "step": 4731
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012840065952184664,
-      "loss": 4.2571,
+      "learning_rate": 1.2798845836768344e-05,
+      "loss": 2.4856,
       "step": 4732
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012819455894476505,
-      "loss": 4.1136,
+      "learning_rate": 1.2778235779060183e-05,
+      "loss": 2.5236,
       "step": 4733
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.00012798845836768343,
-      "loss": 4.1469,
+      "learning_rate": 1.2757625721352021e-05,
+      "loss": 2.4175,
       "step": 4734
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.0001277823577906018,
-      "loss": 4.1639,
+      "learning_rate": 1.273701566364386e-05,
+      "loss": 2.2739,
       "step": 4735
     },
     {
       "epoch": 2.65,
-      "learning_rate": 0.0001275762572135202,
-      "loss": 4.3048,
+      "learning_rate": 1.2716405605935696e-05,
+      "loss": 2.3805,
       "step": 4736
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.0001273701566364386,
-      "loss": 3.7976,
+      "learning_rate": 1.2695795548227534e-05,
+      "loss": 2.1027,
       "step": 4737
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012716405605935698,
-      "loss": 4.0234,
+      "learning_rate": 1.2675185490519373e-05,
+      "loss": 2.6438,
       "step": 4738
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012695795548227536,
-      "loss": 4.1319,
+      "learning_rate": 1.2654575432811214e-05,
+      "loss": 2.4123,
       "step": 4739
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012675185490519374,
-      "loss": 3.864,
+      "learning_rate": 1.2633965375103053e-05,
+      "loss": 2.2026,
       "step": 4740
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012654575432811214,
-      "loss": 4.2154,
+      "learning_rate": 1.2613355317394889e-05,
+      "loss": 2.5611,
       "step": 4741
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012633965375103052,
-      "loss": 3.7397,
+      "learning_rate": 1.2592745259686727e-05,
+      "loss": 1.9181,
       "step": 4742
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.0001261335531739489,
-      "loss": 3.9282,
+      "learning_rate": 1.2572135201978566e-05,
+      "loss": 2.034,
       "step": 4743
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012592745259686728,
-      "loss": 4.1275,
+      "learning_rate": 1.2551525144270404e-05,
+      "loss": 2.3807,
       "step": 4744
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012572135201978566,
-      "loss": 3.9722,
+      "learning_rate": 1.2530915086562242e-05,
+      "loss": 2.2381,
       "step": 4745
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012551525144270404,
-      "loss": 4.0371,
+      "learning_rate": 1.251030502885408e-05,
+      "loss": 2.0974,
       "step": 4746
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012530915086562242,
-      "loss": 4.1809,
+      "learning_rate": 1.2489694971145919e-05,
+      "loss": 2.3514,
       "step": 4747
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.0001251030502885408,
-      "loss": 3.9882,
+      "learning_rate": 1.2469084913437757e-05,
+      "loss": 2.0674,
       "step": 4748
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.0001248969497114592,
-      "loss": 4.129,
+      "learning_rate": 1.2448474855729597e-05,
+      "loss": 2.305,
       "step": 4749
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012469084913437759,
-      "loss": 4.5172,
+      "learning_rate": 1.2427864798021435e-05,
+      "loss": 2.855,
       "step": 4750
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012448474855729597,
-      "loss": 4.0818,
+      "learning_rate": 1.2407254740313274e-05,
+      "loss": 2.187,
       "step": 4751
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012427864798021434,
-      "loss": 3.8811,
+      "learning_rate": 1.2386644682605112e-05,
+      "loss": 2.3901,
       "step": 4752
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.00012407254740313272,
-      "loss": 4.1954,
+      "learning_rate": 1.236603462489695e-05,
+      "loss": 2.4073,
       "step": 4753
     },
     {
       "epoch": 2.66,
-      "learning_rate": 0.0001238664468260511,
-      "loss": 3.9921,
+      "learning_rate": 1.2345424567188789e-05,
+      "loss": 2.0367,
       "step": 4754
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0001236603462489695,
-      "loss": 4.1134,
+      "learning_rate": 1.2324814509480627e-05,
+      "loss": 2.4256,
       "step": 4755
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0001234542456718879,
-      "loss": 4.2043,
+      "learning_rate": 1.2304204451772465e-05,
+      "loss": 2.2616,
       "step": 4756
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012324814509480627,
-      "loss": 4.2192,
+      "learning_rate": 1.2283594394064305e-05,
+      "loss": 2.0676,
       "step": 4757
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012304204451772465,
-      "loss": 3.771,
+      "learning_rate": 1.2262984336356142e-05,
+      "loss": 1.8131,
       "step": 4758
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012283594394064303,
-      "loss": 3.8709,
+      "learning_rate": 1.224237427864798e-05,
+      "loss": 2.0727,
       "step": 4759
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0001226298433635614,
-      "loss": 4.2509,
+      "learning_rate": 1.2221764220939818e-05,
+      "loss": 1.9453,
       "step": 4760
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012242374278647982,
-      "loss": 3.8805,
+      "learning_rate": 1.2201154163231658e-05,
+      "loss": 1.6471,
       "step": 4761
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0001222176422093982,
-      "loss": 3.7339,
+      "learning_rate": 1.2180544105523497e-05,
+      "loss": 1.7307,
       "step": 4762
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012201154163231657,
-      "loss": 3.4137,
+      "learning_rate": 1.2159934047815335e-05,
+      "loss": 1.3297,
       "step": 4763
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012180544105523495,
-      "loss": 3.6801,
+      "learning_rate": 1.2139323990107172e-05,
+      "loss": 1.5519,
       "step": 4764
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012159934047815333,
-      "loss": 3.9871,
+      "learning_rate": 1.2118713932399012e-05,
+      "loss": 1.6424,
       "step": 4765
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012139323990107171,
-      "loss": 3.5351,
+      "learning_rate": 1.209810387469085e-05,
+      "loss": 1.2789,
       "step": 4766
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012118713932399012,
-      "loss": 3.8323,
+      "learning_rate": 1.2077493816982688e-05,
+      "loss": 1.5446,
       "step": 4767
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.0001209810387469085,
-      "loss": 3.0118,
+      "learning_rate": 1.2056883759274527e-05,
+      "loss": 1.4289,
       "step": 4768
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012077493816982688,
-      "loss": 4.4507,
+      "learning_rate": 1.2036273701566365e-05,
+      "loss": 3.198,
       "step": 4769
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012056883759274526,
-      "loss": 4.258,
+      "learning_rate": 1.2015663643858203e-05,
+      "loss": 2.7152,
       "step": 4770
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012036273701566365,
-      "loss": 4.1163,
+      "learning_rate": 1.1995053586150041e-05,
+      "loss": 2.4473,
       "step": 4771
     },
     {
       "epoch": 2.67,
-      "learning_rate": 0.00012015663643858203,
-      "loss": 4.1499,
+      "learning_rate": 1.197444352844188e-05,
+      "loss": 2.5579,
       "step": 4772
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011995053586150041,
-      "loss": 3.9667,
+      "learning_rate": 1.195383347073372e-05,
+      "loss": 2.4407,
       "step": 4773
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011974443528441879,
-      "loss": 3.9557,
+      "learning_rate": 1.1933223413025556e-05,
+      "loss": 2.2227,
       "step": 4774
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011953833470733718,
-      "loss": 4.0949,
+      "learning_rate": 1.1912613355317395e-05,
+      "loss": 2.4583,
       "step": 4775
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011933223413025556,
-      "loss": 4.099,
+      "learning_rate": 1.1892003297609233e-05,
+      "loss": 2.4496,
       "step": 4776
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011912613355317394,
-      "loss": 4.1776,
+      "learning_rate": 1.1871393239901073e-05,
+      "loss": 2.4678,
       "step": 4777
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011892003297609234,
-      "loss": 3.852,
+      "learning_rate": 1.1850783182192911e-05,
+      "loss": 2.477,
       "step": 4778
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011871393239901073,
-      "loss": 3.9028,
+      "learning_rate": 1.183017312448475e-05,
+      "loss": 2.2158,
       "step": 4779
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011850783182192911,
-      "loss": 4.0701,
+      "learning_rate": 1.1809563066776586e-05,
+      "loss": 2.5779,
       "step": 4780
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011830173124484749,
-      "loss": 3.9176,
+      "learning_rate": 1.1788953009068426e-05,
+      "loss": 2.2758,
       "step": 4781
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011809563066776587,
-      "loss": 3.9677,
+      "learning_rate": 1.1768342951360264e-05,
+      "loss": 2.2607,
       "step": 4782
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011788953009068426,
-      "loss": 3.6118,
+      "learning_rate": 1.1747732893652103e-05,
+      "loss": 2.0466,
       "step": 4783
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011768342951360264,
-      "loss": 3.8207,
+      "learning_rate": 1.1727122835943941e-05,
+      "loss": 2.2564,
       "step": 4784
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011747732893652102,
-      "loss": 4.1745,
+      "learning_rate": 1.170651277823578e-05,
+      "loss": 2.7875,
       "step": 4785
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.0001172712283594394,
-      "loss": 3.9395,
+      "learning_rate": 1.1685902720527618e-05,
+      "loss": 2.4901,
       "step": 4786
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011706512778235779,
-      "loss": 4.3215,
+      "learning_rate": 1.1665292662819456e-05,
+      "loss": 2.9346,
       "step": 4787
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011685902720527617,
-      "loss": 4.0716,
+      "learning_rate": 1.1644682605111294e-05,
+      "loss": 2.6442,
       "step": 4788
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011665292662819455,
-      "loss": 3.9891,
+      "learning_rate": 1.1624072547403134e-05,
+      "loss": 2.1686,
       "step": 4789
     },
     {
       "epoch": 2.68,
-      "learning_rate": 0.00011644682605111295,
-      "loss": 4.063,
+      "learning_rate": 1.1603462489694972e-05,
+      "loss": 2.4002,
       "step": 4790
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011624072547403134,
-      "loss": 3.9047,
+      "learning_rate": 1.1582852431986809e-05,
+      "loss": 2.3902,
       "step": 4791
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011603462489694972,
-      "loss": 4.2279,
+      "learning_rate": 1.1562242374278647e-05,
+      "loss": 2.7494,
       "step": 4792
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.0001158285243198681,
-      "loss": 3.9869,
+      "learning_rate": 1.1541632316570487e-05,
+      "loss": 2.3909,
       "step": 4793
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011562242374278648,
-      "loss": 3.9489,
+      "learning_rate": 1.1521022258862326e-05,
+      "loss": 2.3294,
       "step": 4794
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011541632316570487,
-      "loss": 4.068,
+      "learning_rate": 1.1500412201154164e-05,
+      "loss": 2.4136,
       "step": 4795
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011521022258862325,
-      "loss": 3.9465,
+      "learning_rate": 1.1479802143446002e-05,
+      "loss": 2.4178,
       "step": 4796
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011500412201154163,
-      "loss": 4.1374,
+      "learning_rate": 1.145919208573784e-05,
+      "loss": 2.3618,
       "step": 4797
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011479802143446001,
-      "loss": 4.2393,
+      "learning_rate": 1.1438582028029679e-05,
+      "loss": 2.5859,
       "step": 4798
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.0001145919208573784,
-      "loss": 4.0904,
+      "learning_rate": 1.1417971970321517e-05,
+      "loss": 2.6764,
       "step": 4799
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011438582028029678,
-      "loss": 4.013,
+      "learning_rate": 1.1397361912613355e-05,
+      "loss": 2.3548,
       "step": 4800
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011417971970321516,
-      "loss": 3.8992,
+      "learning_rate": 1.1376751854905195e-05,
+      "loss": 1.956,
       "step": 4801
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011397361912613355,
-      "loss": 4.3932,
+      "learning_rate": 1.1356141797197032e-05,
+      "loss": 2.656,
       "step": 4802
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011376751854905195,
-      "loss": 3.8581,
+      "learning_rate": 1.133553173948887e-05,
+      "loss": 2.0666,
       "step": 4803
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011356141797197033,
-      "loss": 3.8381,
+      "learning_rate": 1.131492168178071e-05,
+      "loss": 1.9845,
       "step": 4804
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.0001133553173948887,
-      "loss": 4.3234,
+      "learning_rate": 1.1294311624072549e-05,
+      "loss": 2.6211,
       "step": 4805
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.0001131492168178071,
-      "loss": 3.7634,
+      "learning_rate": 1.1273701566364387e-05,
+      "loss": 1.951,
       "step": 4806
     },
     {
       "epoch": 2.69,
-      "learning_rate": 0.00011294311624072548,
-      "loss": 3.8043,
+      "learning_rate": 1.1253091508656225e-05,
+      "loss": 2.0269,
       "step": 4807
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011273701566364386,
-      "loss": 3.6171,
+      "learning_rate": 1.1232481450948064e-05,
+      "loss": 1.8953,
       "step": 4808
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011253091508656224,
-      "loss": 3.9244,
+      "learning_rate": 1.1211871393239902e-05,
+      "loss": 2.0414,
       "step": 4809
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011232481450948063,
-      "loss": 4.1457,
+      "learning_rate": 1.119126133553174e-05,
+      "loss": 2.5391,
       "step": 4810
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011211871393239901,
-      "loss": 3.9543,
+      "learning_rate": 1.1170651277823578e-05,
+      "loss": 1.985,
       "step": 4811
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011191261335531739,
-      "loss": 4.0763,
+      "learning_rate": 1.1150041220115417e-05,
+      "loss": 1.9643,
       "step": 4812
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011170651277823578,
-      "loss": 3.8289,
+      "learning_rate": 1.1129431162407255e-05,
+      "loss": 1.7974,
       "step": 4813
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011150041220115418,
-      "loss": 3.7572,
+      "learning_rate": 1.1108821104699093e-05,
+      "loss": 1.8384,
       "step": 4814
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011129431162407256,
-      "loss": 3.3532,
+      "learning_rate": 1.1088211046990932e-05,
+      "loss": 1.5902,
       "step": 4815
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011108821104699094,
-      "loss": 3.3987,
+      "learning_rate": 1.1067600989282772e-05,
+      "loss": 1.2767,
       "step": 4816
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011088211046990932,
-      "loss": 4.1033,
+      "learning_rate": 1.104699093157461e-05,
+      "loss": 1.7878,
       "step": 4817
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011067600989282771,
-      "loss": 3.4816,
+      "learning_rate": 1.1026380873866446e-05,
+      "loss": 1.6119,
       "step": 4818
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011046990931574609,
-      "loss": 4.1807,
+      "learning_rate": 1.1005770816158285e-05,
+      "loss": 2.6589,
       "step": 4819
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011026380873866447,
-      "loss": 4.3756,
+      "learning_rate": 1.0985160758450125e-05,
+      "loss": 3.0049,
       "step": 4820
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00011005770816158285,
-      "loss": 4.0864,
+      "learning_rate": 1.0964550700741963e-05,
+      "loss": 3.0027,
       "step": 4821
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00010985160758450124,
-      "loss": 4.0796,
+      "learning_rate": 1.0943940643033801e-05,
+      "loss": 2.9592,
       "step": 4822
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.00010964550700741962,
-      "loss": 3.9667,
+      "learning_rate": 1.092333058532564e-05,
+      "loss": 2.5144,
       "step": 4823
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.000109439406430338,
-      "loss": 4.1096,
+      "learning_rate": 1.0902720527617478e-05,
+      "loss": 2.7415,
       "step": 4824
     },
     {
       "epoch": 2.7,
-      "learning_rate": 0.0001092333058532564,
-      "loss": 4.1128,
+      "learning_rate": 1.0882110469909316e-05,
+      "loss": 2.786,
       "step": 4825
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010902720527617479,
-      "loss": 3.9939,
+      "learning_rate": 1.0861500412201155e-05,
+      "loss": 2.5206,
       "step": 4826
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010882110469909317,
-      "loss": 4.4454,
+      "learning_rate": 1.0840890354492993e-05,
+      "loss": 2.9546,
       "step": 4827
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010861500412201155,
-      "loss": 3.9747,
+      "learning_rate": 1.0820280296784833e-05,
+      "loss": 2.4711,
       "step": 4828
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010840890354492993,
-      "loss": 4.0162,
+      "learning_rate": 1.079967023907667e-05,
+      "loss": 2.4421,
       "step": 4829
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010820280296784832,
-      "loss": 4.1327,
+      "learning_rate": 1.0779060181368508e-05,
+      "loss": 2.6037,
       "step": 4830
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.0001079967023907667,
-      "loss": 3.8298,
+      "learning_rate": 1.0758450123660346e-05,
+      "loss": 2.2897,
       "step": 4831
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010779060181368508,
-      "loss": 3.9623,
+      "learning_rate": 1.0737840065952186e-05,
+      "loss": 2.464,
       "step": 4832
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010758450123660346,
-      "loss": 4.23,
+      "learning_rate": 1.0717230008244024e-05,
+      "loss": 2.8247,
       "step": 4833
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010737840065952185,
-      "loss": 4.1793,
+      "learning_rate": 1.0696619950535863e-05,
+      "loss": 3.0895,
       "step": 4834
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010717230008244023,
-      "loss": 4.0039,
+      "learning_rate": 1.06760098928277e-05,
+      "loss": 2.5345,
       "step": 4835
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010696619950535862,
-      "loss": 3.9986,
+      "learning_rate": 1.065539983511954e-05,
+      "loss": 2.4541,
       "step": 4836
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.000106760098928277,
-      "loss": 4.1194,
+      "learning_rate": 1.0634789777411378e-05,
+      "loss": 2.5034,
       "step": 4837
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.0001065539983511954,
-      "loss": 3.829,
+      "learning_rate": 1.0614179719703216e-05,
+      "loss": 2.4639,
       "step": 4838
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010634789777411378,
-      "loss": 4.0765,
+      "learning_rate": 1.0593569661995054e-05,
+      "loss": 2.3903,
       "step": 4839
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010614179719703215,
-      "loss": 4.1087,
+      "learning_rate": 1.0572959604286892e-05,
+      "loss": 2.5232,
       "step": 4840
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010593569661995053,
-      "loss": 4.2709,
+      "learning_rate": 1.055234954657873e-05,
+      "loss": 2.5986,
       "step": 4841
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010572959604286893,
-      "loss": 4.093,
+      "learning_rate": 1.0531739488870569e-05,
+      "loss": 2.4615,
       "step": 4842
     },
     {
       "epoch": 2.71,
-      "learning_rate": 0.00010552349546578731,
-      "loss": 3.8756,
+      "learning_rate": 1.0511129431162407e-05,
+      "loss": 2.3113,
       "step": 4843
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010531739488870569,
-      "loss": 3.8849,
+      "learning_rate": 1.0490519373454247e-05,
+      "loss": 2.3574,
       "step": 4844
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010511129431162407,
-      "loss": 4.0173,
+      "learning_rate": 1.0469909315746086e-05,
+      "loss": 2.2582,
       "step": 4845
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010490519373454246,
-      "loss": 4.33,
+      "learning_rate": 1.0449299258037922e-05,
+      "loss": 2.7333,
       "step": 4846
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010469909315746084,
-      "loss": 3.6238,
+      "learning_rate": 1.042868920032976e-05,
+      "loss": 2.0345,
       "step": 4847
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010449299258037923,
-      "loss": 3.8127,
+      "learning_rate": 1.04080791426216e-05,
+      "loss": 2.3721,
       "step": 4848
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010428689200329761,
-      "loss": 4.2908,
+      "learning_rate": 1.0387469084913439e-05,
+      "loss": 2.8272,
       "step": 4849
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.000104080791426216,
-      "loss": 4.1978,
+      "learning_rate": 1.0366859027205277e-05,
+      "loss": 2.6508,
       "step": 4850
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010387469084913438,
-      "loss": 3.7541,
+      "learning_rate": 1.0346248969497115e-05,
+      "loss": 2.1347,
       "step": 4851
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010366859027205276,
-      "loss": 4.2765,
+      "learning_rate": 1.0325638911788954e-05,
+      "loss": 2.929,
       "step": 4852
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010346248969497114,
-      "loss": 3.8571,
+      "learning_rate": 1.0305028854080792e-05,
+      "loss": 2.2052,
       "step": 4853
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010325638911788954,
-      "loss": 4.3368,
+      "learning_rate": 1.028441879637263e-05,
+      "loss": 2.8242,
       "step": 4854
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010305028854080792,
-      "loss": 4.0531,
+      "learning_rate": 1.0263808738664469e-05,
+      "loss": 2.3845,
       "step": 4855
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.0001028441879637263,
-      "loss": 3.9592,
+      "learning_rate": 1.0243198680956307e-05,
+      "loss": 2.1336,
       "step": 4856
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010263808738664468,
-      "loss": 3.9429,
+      "learning_rate": 1.0222588623248145e-05,
+      "loss": 2.2178,
       "step": 4857
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010243198680956307,
-      "loss": 3.8624,
+      "learning_rate": 1.0201978565539983e-05,
+      "loss": 2.0492,
       "step": 4858
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010222588623248145,
-      "loss": 3.9816,
+      "learning_rate": 1.0181368507831822e-05,
+      "loss": 2.0526,
       "step": 4859
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010201978565539984,
-      "loss": 4.476,
+      "learning_rate": 1.0160758450123662e-05,
+      "loss": 2.5147,
       "step": 4860
     },
     {
       "epoch": 2.72,
-      "learning_rate": 0.00010181368507831822,
-      "loss": 3.5891,
+      "learning_rate": 1.01401483924155e-05,
+      "loss": 1.626,
       "step": 4861
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00010160758450123661,
-      "loss": 4.1855,
+      "learning_rate": 1.0119538334707337e-05,
+      "loss": 2.3596,
       "step": 4862
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.000101401483924155,
-      "loss": 3.6596,
+      "learning_rate": 1.0098928276999175e-05,
+      "loss": 1.6348,
       "step": 4863
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00010119538334707337,
-      "loss": 4.0405,
+      "learning_rate": 1.0078318219291015e-05,
+      "loss": 1.7866,
       "step": 4864
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00010098928276999175,
-      "loss": 3.3664,
+      "learning_rate": 1.0057708161582853e-05,
+      "loss": 1.2508,
       "step": 4865
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00010078318219291015,
-      "loss": 3.8735,
+      "learning_rate": 1.0037098103874692e-05,
+      "loss": 1.6359,
       "step": 4866
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00010057708161582853,
-      "loss": 3.7382,
+      "learning_rate": 1.001648804616653e-05,
+      "loss": 1.6171,
       "step": 4867
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.0001003709810387469,
-      "loss": 3.3152,
+      "learning_rate": 9.995877988458368e-06,
+      "loss": 1.5817,
       "step": 4868
     },
     {
       "epoch": 2.73,
-      "learning_rate": 0.00010016488046166528,
-      "loss": 4.3013,
+      "learning_rate": 9.975267930750206e-06,
+      "loss": 2.921,
       "step": 4869
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.995877988458368e-05,
-      "loss": 4.1549,
+      "learning_rate": 9.954657873042045e-06,
+      "loss": 2.6249,
       "step": 4870
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.975267930750207e-05,
-      "loss": 4.1639,
+      "learning_rate": 9.934047815333883e-06,
+      "loss": 2.7352,
       "step": 4871
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.954657873042045e-05,
-      "loss": 4.2283,
+      "learning_rate": 9.913437757625723e-06,
+      "loss": 2.6267,
       "step": 4872
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.934047815333883e-05,
-      "loss": 4.1777,
+      "learning_rate": 9.89282769991756e-06,
+      "loss": 3.1403,
       "step": 4873
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.913437757625722e-05,
-      "loss": 3.9231,
+      "learning_rate": 9.872217642209398e-06,
+      "loss": 2.4889,
       "step": 4874
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.89282769991756e-05,
-      "loss": 4.2395,
+      "learning_rate": 9.851607584501236e-06,
+      "loss": 2.862,
       "step": 4875
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.872217642209398e-05,
-      "loss": 4.016,
+      "learning_rate": 9.830997526793076e-06,
+      "loss": 2.418,
       "step": 4876
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.851607584501236e-05,
-      "loss": 4.1604,
+      "learning_rate": 9.810387469084915e-06,
+      "loss": 2.7866,
       "step": 4877
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.830997526793076e-05,
-      "loss": 3.9011,
+      "learning_rate": 9.789777411376753e-06,
+      "loss": 2.491,
       "step": 4878
     },
     {
       "epoch": 2.73,
-      "learning_rate": 9.810387469084913e-05,
-      "loss": 4.1842,
+      "learning_rate": 9.76916735366859e-06,
+      "loss": 2.9754,
       "step": 4879
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.789777411376751e-05,
-      "loss": 4.2167,
+      "learning_rate": 9.74855729596043e-06,
+      "loss": 2.5448,
       "step": 4880
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.76916735366859e-05,
-      "loss": 3.9214,
+      "learning_rate": 9.727947238252268e-06,
+      "loss": 2.143,
       "step": 4881
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.748557295960429e-05,
-      "loss": 4.1699,
+      "learning_rate": 9.707337180544106e-06,
+      "loss": 2.8961,
       "step": 4882
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.727947238252268e-05,
-      "loss": 4.0379,
+      "learning_rate": 9.686727122835944e-06,
+      "loss": 2.8331,
       "step": 4883
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.707337180544106e-05,
-      "loss": 3.9369,
+      "learning_rate": 9.666117065127783e-06,
+      "loss": 2.3673,
       "step": 4884
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.686727122835944e-05,
-      "loss": 4.3384,
+      "learning_rate": 9.645507007419621e-06,
+      "loss": 2.6917,
       "step": 4885
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.666117065127783e-05,
-      "loss": 3.9574,
+      "learning_rate": 9.62489694971146e-06,
+      "loss": 2.7653,
       "step": 4886
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.645507007419621e-05,
-      "loss": 4.1505,
+      "learning_rate": 9.604286892003297e-06,
+      "loss": 2.4972,
       "step": 4887
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.624896949711459e-05,
-      "loss": 3.9677,
+      "learning_rate": 9.583676834295137e-06,
+      "loss": 3.0382,
       "step": 4888
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.604286892003297e-05,
-      "loss": 4.4408,
+      "learning_rate": 9.563066776586976e-06,
+      "loss": 2.9654,
       "step": 4889
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.583676834295136e-05,
-      "loss": 4.0143,
+      "learning_rate": 9.542456718878812e-06,
+      "loss": 2.5181,
       "step": 4890
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.563066776586974e-05,
-      "loss": 4.0088,
+      "learning_rate": 9.52184666117065e-06,
+      "loss": 2.5408,
       "step": 4891
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.542456718878812e-05,
-      "loss": 4.0876,
+      "learning_rate": 9.50123660346249e-06,
+      "loss": 2.6677,
       "step": 4892
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.52184666117065e-05,
-      "loss": 4.0073,
+      "learning_rate": 9.480626545754329e-06,
+      "loss": 2.3904,
       "step": 4893
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.50123660346249e-05,
-      "loss": 3.9534,
+      "learning_rate": 9.460016488046167e-06,
+      "loss": 2.8317,
       "step": 4894
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.480626545754329e-05,
-      "loss": 4.0666,
+      "learning_rate": 9.439406430338006e-06,
+      "loss": 2.5779,
       "step": 4895
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.460016488046167e-05,
-      "loss": 3.8993,
+      "learning_rate": 9.418796372629844e-06,
+      "loss": 2.6286,
       "step": 4896
     },
     {
       "epoch": 2.74,
-      "learning_rate": 9.439406430338005e-05,
-      "loss": 4.0755,
+      "learning_rate": 9.398186314921682e-06,
+      "loss": 2.7634,
       "step": 4897
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.418796372629844e-05,
-      "loss": 4.0357,
+      "learning_rate": 9.37757625721352e-06,
+      "loss": 2.5058,
       "step": 4898
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.398186314921682e-05,
-      "loss": 3.9521,
+      "learning_rate": 9.356966199505359e-06,
+      "loss": 2.4266,
       "step": 4899
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.37757625721352e-05,
-      "loss": 4.2231,
+      "learning_rate": 9.336356141797197e-06,
+      "loss": 2.815,
       "step": 4900
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.356966199505358e-05,
-      "loss": 3.9994,
+      "learning_rate": 9.315746084089035e-06,
+      "loss": 2.2848,
       "step": 4901
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.336356141797197e-05,
-      "loss": 4.2229,
+      "learning_rate": 9.295136026380874e-06,
+      "loss": 2.7043,
       "step": 4902
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.315746084089035e-05,
-      "loss": 4.002,
+      "learning_rate": 9.274525968672712e-06,
+      "loss": 2.4903,
       "step": 4903
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.295136026380873e-05,
-      "loss": 3.7344,
+      "learning_rate": 9.253915910964552e-06,
+      "loss": 2.207,
       "step": 4904
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.274525968672711e-05,
-      "loss": 3.9874,
+      "learning_rate": 9.23330585325639e-06,
+      "loss": 2.2419,
       "step": 4905
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.253915910964552e-05,
-      "loss": 3.7257,
+      "learning_rate": 9.212695795548227e-06,
+      "loss": 2.1222,
       "step": 4906
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.23330585325639e-05,
-      "loss": 3.8268,
+      "learning_rate": 9.192085737840065e-06,
+      "loss": 2.1965,
       "step": 4907
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.212695795548228e-05,
-      "loss": 4.0252,
+      "learning_rate": 9.171475680131905e-06,
+      "loss": 2.2933,
       "step": 4908
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.192085737840066e-05,
-      "loss": 3.6188,
+      "learning_rate": 9.150865622423743e-06,
+      "loss": 1.9238,
       "step": 4909
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.171475680131905e-05,
-      "loss": 4.1624,
+      "learning_rate": 9.130255564715582e-06,
+      "loss": 2.6091,
       "step": 4910
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.150865622423743e-05,
-      "loss": 4.0217,
+      "learning_rate": 9.10964550700742e-06,
+      "loss": 2.3484,
       "step": 4911
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.130255564715581e-05,
-      "loss": 3.7265,
+      "learning_rate": 9.089035449299258e-06,
+      "loss": 1.8057,
       "step": 4912
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.109645507007419e-05,
-      "loss": 3.6108,
+      "learning_rate": 9.068425391591097e-06,
+      "loss": 1.6471,
       "step": 4913
     },
     {
       "epoch": 2.75,
-      "learning_rate": 9.089035449299258e-05,
-      "loss": 3.9056,
+      "learning_rate": 9.047815333882935e-06,
+      "loss": 1.8093,
       "step": 4914
     },
     {
       "epoch": 2.76,
-      "learning_rate": 9.068425391591096e-05,
-      "loss": 3.5867,
+      "learning_rate": 9.027205276174775e-06,
+      "loss": 1.6392,
       "step": 4915
     },
     {
       "epoch": 2.76,
-      "learning_rate": 9.047815333882934e-05,
-      "loss": 3.5735,
+      "learning_rate": 9.006595218466613e-06,
+      "loss": 1.7231,
       "step": 4916
     },
     {
       "epoch": 2.76,
-      "learning_rate": 9.027205276174774e-05,
-      "loss": 3.8849,
+      "learning_rate": 8.98598516075845e-06,
+      "loss": 1.8259,
       "step": 4917
     },
     {
       "epoch": 2.76,
-      "learning_rate": 9.006595218466613e-05,
-      "loss": 3.794,
+      "learning_rate": 8.965375103050288e-06,
+      "loss": 1.853,
       "step": 4918
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.985985160758451e-05,
-      "loss": 4.0421,
+      "learning_rate": 8.944765045342128e-06,
+      "loss": 2.5986,
       "step": 4919
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.965375103050289e-05,
-      "loss": 4.0963,
+      "learning_rate": 8.924154987633966e-06,
+      "loss": 2.6431,
       "step": 4920
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.944765045342128e-05,
-      "loss": 4.1222,
+      "learning_rate": 8.903544929925805e-06,
+      "loss": 2.8023,
       "step": 4921
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.924154987633966e-05,
-      "loss": 4.3553,
+      "learning_rate": 8.882934872217643e-06,
+      "loss": 3.0283,
       "step": 4922
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.903544929925804e-05,
-      "loss": 3.9195,
+      "learning_rate": 8.862324814509481e-06,
+      "loss": 2.4856,
       "step": 4923
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.882934872217642e-05,
-      "loss": 4.3621,
+      "learning_rate": 8.84171475680132e-06,
+      "loss": 3.0852,
       "step": 4924
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.862324814509481e-05,
-      "loss": 4.0662,
+      "learning_rate": 8.821104699093158e-06,
+      "loss": 2.7449,
       "step": 4925
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.841714756801319e-05,
-      "loss": 4.0381,
+      "learning_rate": 8.800494641384996e-06,
+      "loss": 2.9246,
       "step": 4926
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.821104699093157e-05,
-      "loss": 4.1145,
+      "learning_rate": 8.779884583676836e-06,
+      "loss": 2.7738,
       "step": 4927
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.800494641384995e-05,
-      "loss": 4.3163,
+      "learning_rate": 8.759274525968673e-06,
+      "loss": 2.897,
       "step": 4928
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.779884583676834e-05,
-      "loss": 3.9003,
+      "learning_rate": 8.738664468260511e-06,
+      "loss": 3.1541,
       "step": 4929
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.759274525968674e-05,
-      "loss": 3.9313,
+      "learning_rate": 8.71805441055235e-06,
+      "loss": 2.7278,
       "step": 4930
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.738664468260512e-05,
-      "loss": 4.1562,
+      "learning_rate": 8.69744435284419e-06,
+      "loss": 2.6411,
       "step": 4931
     },
     {
       "epoch": 2.76,
-      "learning_rate": 8.71805441055235e-05,
-      "loss": 4.5369,
+      "learning_rate": 8.676834295136028e-06,
+      "loss": 3.3862,
       "step": 4932
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.697444352844189e-05,
-      "loss": 4.2737,
+      "learning_rate": 8.656224237427866e-06,
+      "loss": 2.6279,
       "step": 4933
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.676834295136027e-05,
-      "loss": 4.0935,
+      "learning_rate": 8.635614179719703e-06,
+      "loss": 2.834,
       "step": 4934
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.656224237427865e-05,
-      "loss": 3.9734,
+      "learning_rate": 8.615004122011543e-06,
+      "loss": 3.0022,
       "step": 4935
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.635614179719703e-05,
-      "loss": 4.4055,
+      "learning_rate": 8.59439406430338e-06,
+      "loss": 2.9237,
       "step": 4936
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.615004122011542e-05,
-      "loss": 4.1519,
+      "learning_rate": 8.573784006595219e-06,
+      "loss": 2.7552,
       "step": 4937
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.59439406430338e-05,
-      "loss": 4.1157,
+      "learning_rate": 8.553173948887057e-06,
+      "loss": 2.663,
       "step": 4938
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.573784006595218e-05,
-      "loss": 4.1236,
+      "learning_rate": 8.532563891178896e-06,
+      "loss": 2.6413,
       "step": 4939
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.553173948887056e-05,
-      "loss": 4.0669,
+      "learning_rate": 8.511953833470734e-06,
+      "loss": 2.8393,
       "step": 4940
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.532563891178897e-05,
-      "loss": 3.9359,
+      "learning_rate": 8.491343775762572e-06,
+      "loss": 2.3304,
       "step": 4941
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.511953833470735e-05,
-      "loss": 3.9318,
+      "learning_rate": 8.47073371805441e-06,
+      "loss": 2.5338,
       "step": 4942
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.491343775762573e-05,
-      "loss": 3.985,
+      "learning_rate": 8.45012366034625e-06,
+      "loss": 2.5519,
       "step": 4943
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.47073371805441e-05,
-      "loss": 4.2771,
+      "learning_rate": 8.429513602638087e-06,
+      "loss": 2.8694,
       "step": 4944
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.45012366034625e-05,
-      "loss": 3.9242,
+      "learning_rate": 8.408903544929926e-06,
+      "loss": 2.316,
       "step": 4945
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.429513602638088e-05,
-      "loss": 4.0654,
+      "learning_rate": 8.388293487221764e-06,
+      "loss": 2.4713,
       "step": 4946
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.408903544929926e-05,
-      "loss": 3.9158,
+      "learning_rate": 8.367683429513604e-06,
+      "loss": 2.5208,
       "step": 4947
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.388293487221764e-05,
-      "loss": 4.3161,
+      "learning_rate": 8.347073371805442e-06,
+      "loss": 3.2156,
       "step": 4948
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.367683429513603e-05,
-      "loss": 4.1337,
+      "learning_rate": 8.32646331409728e-06,
+      "loss": 2.5688,
       "step": 4949
     },
     {
       "epoch": 2.77,
-      "learning_rate": 8.347073371805441e-05,
-      "loss": 3.8717,
+      "learning_rate": 8.305853256389117e-06,
+      "loss": 2.3208,
       "step": 4950
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.326463314097279e-05,
-      "loss": 4.1009,
+      "learning_rate": 8.285243198680957e-06,
+      "loss": 2.475,
       "step": 4951
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.305853256389117e-05,
-      "loss": 4.0677,
+      "learning_rate": 8.264633140972795e-06,
+      "loss": 2.7859,
       "step": 4952
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.285243198680958e-05,
-      "loss": 4.2589,
+      "learning_rate": 8.244023083264634e-06,
+      "loss": 2.6469,
       "step": 4953
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.264633140972796e-05,
-      "loss": 3.8468,
+      "learning_rate": 8.223413025556472e-06,
+      "loss": 2.2307,
       "step": 4954
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.244023083264634e-05,
-      "loss": 3.9947,
+      "learning_rate": 8.20280296784831e-06,
+      "loss": 2.8248,
       "step": 4955
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.223413025556472e-05,
-      "loss": 3.7464,
+      "learning_rate": 8.182192910140148e-06,
+      "loss": 2.0915,
       "step": 4956
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.202802967848311e-05,
-      "loss": 4.03,
+      "learning_rate": 8.161582852431987e-06,
+      "loss": 2.2368,
       "step": 4957
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.182192910140149e-05,
-      "loss": 4.2251,
+      "learning_rate": 8.140972794723825e-06,
+      "loss": 2.7482,
       "step": 4958
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.161582852431987e-05,
-      "loss": 4.0451,
+      "learning_rate": 8.120362737015665e-06,
+      "loss": 2.2849,
       "step": 4959
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.140972794723825e-05,
-      "loss": 4.0621,
+      "learning_rate": 8.099752679307503e-06,
+      "loss": 2.1408,
       "step": 4960
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.120362737015664e-05,
-      "loss": 3.6358,
+      "learning_rate": 8.07914262159934e-06,
+      "loss": 1.9468,
       "step": 4961
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.099752679307502e-05,
-      "loss": 3.6623,
+      "learning_rate": 8.058532563891178e-06,
+      "loss": 1.8918,
       "step": 4962
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.07914262159934e-05,
-      "loss": 3.7436,
+      "learning_rate": 8.037922506183018e-06,
+      "loss": 1.7268,
       "step": 4963
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.058532563891178e-05,
-      "loss": 4.141,
+      "learning_rate": 8.017312448474857e-06,
+      "loss": 2.2602,
       "step": 4964
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.037922506183019e-05,
-      "loss": 3.5503,
+      "learning_rate": 7.996702390766695e-06,
+      "loss": 1.6797,
       "step": 4965
     },
     {
       "epoch": 2.78,
-      "learning_rate": 8.017312448474857e-05,
-      "loss": 3.8188,
+      "learning_rate": 7.976092333058533e-06,
+      "loss": 1.6778,
       "step": 4966
     },
     {
       "epoch": 2.78,
-      "learning_rate": 7.996702390766695e-05,
-      "loss": 3.7271,
+      "learning_rate": 7.955482275350371e-06,
+      "loss": 1.8023,
       "step": 4967
     },
     {
       "epoch": 2.78,
-      "learning_rate": 7.976092333058532e-05,
-      "loss": 3.376,
+      "learning_rate": 7.93487221764221e-06,
+      "loss": 1.9218,
       "step": 4968
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.955482275350372e-05,
-      "loss": 4.1115,
+      "learning_rate": 7.914262159934048e-06,
+      "loss": 2.9654,
       "step": 4969
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.93487221764221e-05,
-      "loss": 4.1032,
+      "learning_rate": 7.893652102225886e-06,
+      "loss": 2.8864,
       "step": 4970
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.914262159934048e-05,
-      "loss": 4.1524,
+      "learning_rate": 7.873042044517726e-06,
+      "loss": 2.7395,
       "step": 4971
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.893652102225886e-05,
-      "loss": 3.9931,
+      "learning_rate": 7.852431986809563e-06,
+      "loss": 2.9185,
       "step": 4972
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.873042044517725e-05,
-      "loss": 4.0175,
+      "learning_rate": 7.831821929101401e-06,
+      "loss": 2.6264,
       "step": 4973
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.852431986809563e-05,
-      "loss": 4.0084,
+      "learning_rate": 7.81121187139324e-06,
+      "loss": 2.8592,
       "step": 4974
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.831821929101401e-05,
-      "loss": 3.8754,
+      "learning_rate": 7.79060181368508e-06,
+      "loss": 2.769,
       "step": 4975
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.811211871393239e-05,
-      "loss": 4.2908,
+      "learning_rate": 7.769991755976918e-06,
+      "loss": 3.0036,
       "step": 4976
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.79060181368508e-05,
-      "loss": 3.9824,
+      "learning_rate": 7.749381698268756e-06,
+      "loss": 2.7407,
       "step": 4977
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.769991755976917e-05,
-      "loss": 4.1766,
+      "learning_rate": 7.728771640560593e-06,
+      "loss": 2.8225,
       "step": 4978
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.749381698268755e-05,
-      "loss": 4.1298,
+      "learning_rate": 7.708161582852433e-06,
+      "loss": 2.8475,
       "step": 4979
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.728771640560593e-05,
-      "loss": 3.9726,
+      "learning_rate": 7.687551525144271e-06,
+      "loss": 2.5315,
       "step": 4980
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.708161582852433e-05,
-      "loss": 4.0816,
+      "learning_rate": 7.66694146743611e-06,
+      "loss": 2.6507,
       "step": 4981
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.68755152514427e-05,
-      "loss": 3.7708,
+      "learning_rate": 7.646331409727948e-06,
+      "loss": 2.4826,
       "step": 4982
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.666941467436109e-05,
-      "loss": 4.2322,
+      "learning_rate": 7.625721352019787e-06,
+      "loss": 3.0714,
       "step": 4983
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.646331409727947e-05,
-      "loss": 3.9224,
+      "learning_rate": 7.605111294311624e-06,
+      "loss": 2.5255,
       "step": 4984
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.625721352019786e-05,
-      "loss": 3.9643,
+      "learning_rate": 7.5845012366034625e-06,
+      "loss": 2.6783,
       "step": 4985
     },
     {
       "epoch": 2.79,
-      "learning_rate": 7.605111294311624e-05,
-      "loss": 3.905,
+      "learning_rate": 7.563891178895301e-06,
+      "loss": 2.667,
       "step": 4986
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.584501236603462e-05,
-      "loss": 3.803,
+      "learning_rate": 7.54328112118714e-06,
+      "loss": 2.5679,
       "step": 4987
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.563891178895301e-05,
-      "loss": 4.1213,
+      "learning_rate": 7.522671063478978e-06,
+      "loss": 2.7657,
       "step": 4988
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.54328112118714e-05,
-      "loss": 4.0797,
+      "learning_rate": 7.5020610057708165e-06,
+      "loss": 2.6141,
       "step": 4989
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.522671063478978e-05,
-      "loss": 4.1437,
+      "learning_rate": 7.481450948062654e-06,
+      "loss": 2.7376,
       "step": 4990
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.502061005770816e-05,
-      "loss": 3.9164,
+      "learning_rate": 7.460840890354494e-06,
+      "loss": 2.7716,
       "step": 4991
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.481450948062654e-05,
-      "loss": 3.8516,
+      "learning_rate": 7.440230832646332e-06,
+      "loss": 2.4922,
       "step": 4992
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.460840890354494e-05,
-      "loss": 4.1073,
+      "learning_rate": 7.41962077493817e-06,
+      "loss": 2.8476,
       "step": 4993
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.440230832646332e-05,
-      "loss": 4.1057,
+      "learning_rate": 7.399010717230008e-06,
+      "loss": 2.7031,
       "step": 4994
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.41962077493817e-05,
-      "loss": 4.1811,
+      "learning_rate": 7.378400659521847e-06,
+      "loss": 2.8157,
       "step": 4995
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.399010717230008e-05,
-      "loss": 3.6598,
+      "learning_rate": 7.3577906018136855e-06,
+      "loss": 2.4595,
       "step": 4996
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.378400659521847e-05,
-      "loss": 4.0507,
+      "learning_rate": 7.337180544105524e-06,
+      "loss": 2.769,
       "step": 4997
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.357790601813685e-05,
-      "loss": 4.1363,
+      "learning_rate": 7.316570486397362e-06,
+      "loss": 2.6814,
       "step": 4998
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.337180544105523e-05,
-      "loss": 4.2833,
+      "learning_rate": 7.295960428689201e-06,
+      "loss": 2.7096,
       "step": 4999
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.316570486397362e-05,
-      "loss": 3.8714,
+      "learning_rate": 7.2753503709810395e-06,
+      "loss": 2.658,
       "step": 5000
     },
     {
       "epoch": 2.8,
-      "eval_loss": 5.234638214111328,
-      "eval_runtime": 1248.2731,
-      "eval_samples_per_second": 2.117,
-      "eval_steps_per_second": 0.265,
-      "eval_wer": 1.9970646568821897,
+      "eval_loss": 6.554072380065918,
+      "eval_runtime": 1253.527,
+      "eval_samples_per_second": 2.108,
+      "eval_steps_per_second": 0.264,
+      "eval_wer": 1.9877429591431972,
       "step": 5000
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.295960428689201e-05,
-      "loss": 3.925,
+      "learning_rate": 7.254740313272877e-06,
+      "loss": 2.521,
       "step": 5001
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.27535037098104e-05,
-      "loss": 4.1954,
+      "learning_rate": 7.234130255564715e-06,
+      "loss": 2.6417,
       "step": 5002
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.254740313272877e-05,
-      "loss": 3.8644,
+      "learning_rate": 7.213520197856554e-06,
+      "loss": 2.464,
       "step": 5003
     },
     {
       "epoch": 2.8,
-      "learning_rate": 7.234130255564715e-05,
-      "loss": 4.1649,
+      "learning_rate": 7.192910140148393e-06,
+      "loss": 2.6886,
       "step": 5004
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.213520197856555e-05,
-      "loss": 3.9817,
+      "learning_rate": 7.172300082440231e-06,
+      "loss": 2.5723,
       "step": 5005
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.192910140148393e-05,
-      "loss": 4.116,
+      "learning_rate": 7.151690024732069e-06,
+      "loss": 2.7634,
       "step": 5006
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.17230008244023e-05,
-      "loss": 3.9249,
+      "learning_rate": 7.1310799670239084e-06,
+      "loss": 2.2908,
       "step": 5007
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.151690024732068e-05,
-      "loss": 4.2712,
+      "learning_rate": 7.110469909315747e-06,
+      "loss": 2.8367,
       "step": 5008
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.131079967023908e-05,
-      "loss": 3.7553,
+      "learning_rate": 7.089859851607584e-06,
+      "loss": 2.2804,
       "step": 5009
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.110469909315746e-05,
-      "loss": 3.9739,
+      "learning_rate": 7.0692497938994225e-06,
+      "loss": 2.4933,
       "step": 5010
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.089859851607585e-05,
-      "loss": 4.2576,
+      "learning_rate": 7.0486397361912625e-06,
+      "loss": 2.6174,
       "step": 5011
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.069249793899423e-05,
-      "loss": 3.9414,
+      "learning_rate": 7.0280296784831e-06,
+      "loss": 2.1334,
       "step": 5012
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.048639736191262e-05,
-      "loss": 3.9823,
+      "learning_rate": 7.007419620774938e-06,
+      "loss": 2.4437,
       "step": 5013
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.0280296784831e-05,
-      "loss": 3.7228,
+      "learning_rate": 6.9868095630667765e-06,
+      "loss": 1.635,
       "step": 5014
     },
     {
       "epoch": 2.81,
-      "learning_rate": 7.007419620774938e-05,
-      "loss": 3.5145,
+      "learning_rate": 6.966199505358616e-06,
+      "loss": 1.7575,
       "step": 5015
     },
     {
       "epoch": 2.81,
-      "learning_rate": 6.986809563066776e-05,
-      "loss": 3.7612,
+      "learning_rate": 6.945589447650454e-06,
+      "loss": 1.6098,
       "step": 5016
     },
     {
       "epoch": 2.81,
-      "learning_rate": 6.966199505358615e-05,
-      "loss": 3.6675,
+      "learning_rate": 6.924979389942292e-06,
+      "loss": 1.7285,
       "step": 5017
     },
     {
       "epoch": 2.81,
-      "learning_rate": 6.945589447650453e-05,
-      "loss": 3.4239,
+      "learning_rate": 6.90436933223413e-06,
+      "loss": 1.6823,
       "step": 5018
     },
     {
       "epoch": 2.81,
-      "learning_rate": 6.924979389942291e-05,
-      "loss": 3.9699,
+      "learning_rate": 6.88375927452597e-06,
+      "loss": 3.0864,
       "step": 5019
     },
     {
       "epoch": 2.81,
-      "learning_rate": 6.90436933223413e-05,
-      "loss": 4.1043,
+      "learning_rate": 6.863149216817807e-06,
+      "loss": 3.0969,
       "step": 5020
     },
     {
       "epoch": 2.81,
-      "learning_rate": 6.883759274525969e-05,
-      "loss": 4.2392,
+      "learning_rate": 6.8425391591096454e-06,
+      "loss": 3.0424,
       "step": 5021
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.863149216817807e-05,
-      "loss": 4.3094,
+      "learning_rate": 6.821929101401484e-06,
+      "loss": 3.1769,
       "step": 5022
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.842539159109646e-05,
-      "loss": 4.3282,
+      "learning_rate": 6.801319043693323e-06,
+      "loss": 3.246,
       "step": 5023
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.821929101401484e-05,
-      "loss": 3.9143,
+      "learning_rate": 6.780708985985161e-06,
+      "loss": 2.6372,
       "step": 5024
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.801319043693323e-05,
-      "loss": 3.969,
+      "learning_rate": 6.7600989282769995e-06,
+      "loss": 2.6404,
       "step": 5025
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.780708985985161e-05,
-      "loss": 4.253,
+      "learning_rate": 6.739488870568839e-06,
+      "loss": 3.0533,
       "step": 5026
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.760098928276999e-05,
-      "loss": 3.8793,
+      "learning_rate": 6.718878812860677e-06,
+      "loss": 2.7088,
       "step": 5027
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.739488870568838e-05,
-      "loss": 4.1997,
+      "learning_rate": 6.698268755152514e-06,
+      "loss": 2.8977,
       "step": 5028
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.718878812860676e-05,
-      "loss": 4.1319,
+      "learning_rate": 6.677658697444353e-06,
+      "loss": 2.9093,
       "step": 5029
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.698268755152514e-05,
-      "loss": 3.9122,
+      "learning_rate": 6.657048639736193e-06,
+      "loss": 2.7787,
       "step": 5030
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.677658697444352e-05,
-      "loss": 4.3918,
+      "learning_rate": 6.63643858202803e-06,
+      "loss": 3.3988,
       "step": 5031
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.657048639736192e-05,
-      "loss": 3.8107,
+      "learning_rate": 6.615828524319868e-06,
+      "loss": 2.5641,
       "step": 5032
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.63643858202803e-05,
-      "loss": 4.026,
+      "learning_rate": 6.595218466611707e-06,
+      "loss": 2.9786,
       "step": 5033
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.615828524319868e-05,
-      "loss": 3.7601,
+      "learning_rate": 6.574608408903546e-06,
+      "loss": 2.513,
       "step": 5034
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.595218466611707e-05,
-      "loss": 3.9656,
+      "learning_rate": 6.553998351195384e-06,
+      "loss": 2.8531,
       "step": 5035
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.574608408903546e-05,
-      "loss": 3.8697,
+      "learning_rate": 6.5333882934872224e-06,
+      "loss": 2.9833,
       "step": 5036
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.553998351195384e-05,
-      "loss": 3.8761,
+      "learning_rate": 6.51277823577906e-06,
+      "loss": 2.7162,
       "step": 5037
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.533388293487222e-05,
-      "loss": 4.1502,
+      "learning_rate": 6.4921681780709e-06,
+      "loss": 3.0478,
       "step": 5038
     },
     {
       "epoch": 2.82,
-      "learning_rate": 6.51277823577906e-05,
-      "loss": 4.0038,
+      "learning_rate": 6.471558120362737e-06,
+      "loss": 2.773,
       "step": 5039
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.4921681780709e-05,
-      "loss": 4.3661,
+      "learning_rate": 6.450948062654576e-06,
+      "loss": 3.0097,
       "step": 5040
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.471558120362737e-05,
-      "loss": 4.0039,
+      "learning_rate": 6.430338004946414e-06,
+      "loss": 2.9396,
       "step": 5041
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.450948062654575e-05,
-      "loss": 3.933,
+      "learning_rate": 6.409727947238253e-06,
+      "loss": 2.8084,
       "step": 5042
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.430338004946413e-05,
-      "loss": 4.3495,
+      "learning_rate": 6.389117889530091e-06,
+      "loss": 3.2174,
       "step": 5043
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.409727947238253e-05,
-      "loss": 4.3116,
+      "learning_rate": 6.36850783182193e-06,
+      "loss": 3.0681,
       "step": 5044
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.38911788953009e-05,
-      "loss": 3.8859,
+      "learning_rate": 6.347897774113767e-06,
+      "loss": 2.4343,
       "step": 5045
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.36850783182193e-05,
-      "loss": 4.1695,
+      "learning_rate": 6.327287716405607e-06,
+      "loss": 2.7448,
       "step": 5046
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.347897774113768e-05,
-      "loss": 3.977,
+      "learning_rate": 6.3066776586974446e-06,
+      "loss": 2.6659,
       "step": 5047
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.327287716405607e-05,
-      "loss": 3.8761,
+      "learning_rate": 6.286067600989283e-06,
+      "loss": 2.4004,
       "step": 5048
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.306677658697445e-05,
-      "loss": 3.8942,
+      "learning_rate": 6.265457543281121e-06,
+      "loss": 2.6102,
       "step": 5049
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.286067600989283e-05,
-      "loss": 4.2719,
+      "learning_rate": 6.2448474855729595e-06,
+      "loss": 2.9613,
       "step": 5050
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.265457543281121e-05,
-      "loss": 3.9412,
+      "learning_rate": 6.224237427864799e-06,
+      "loss": 2.6269,
       "step": 5051
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.24484748557296e-05,
-      "loss": 3.9942,
+      "learning_rate": 6.203627370156637e-06,
+      "loss": 2.6094,
       "step": 5052
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.224237427864798e-05,
-      "loss": 4.0148,
+      "learning_rate": 6.183017312448475e-06,
+      "loss": 2.7355,
       "step": 5053
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.203627370156636e-05,
-      "loss": 3.6956,
+      "learning_rate": 6.1624072547403135e-06,
+      "loss": 2.2396,
       "step": 5054
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.183017312448476e-05,
-      "loss": 3.9805,
+      "learning_rate": 6.141797197032153e-06,
+      "loss": 2.6823,
       "step": 5055
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.162407254740313e-05,
-      "loss": 3.8139,
+      "learning_rate": 6.12118713932399e-06,
+      "loss": 2.4007,
       "step": 5056
     },
     {
       "epoch": 2.83,
-      "learning_rate": 6.141797197032151e-05,
-      "loss": 4.5294,
+      "learning_rate": 6.100577081615829e-06,
+      "loss": 3.1116,
       "step": 5057
     },
     {
       "epoch": 2.84,
-      "learning_rate": 6.121187139323991e-05,
-      "loss": 3.7592,
+      "learning_rate": 6.0799670239076675e-06,
+      "loss": 2.3735,
       "step": 5058
     },
     {
       "epoch": 2.84,
-      "learning_rate": 6.100577081615829e-05,
-      "loss": 4.2819,
+      "learning_rate": 6.059356966199506e-06,
+      "loss": 2.7301,
       "step": 5059
     },
     {
       "epoch": 2.84,
-      "learning_rate": 6.079967023907667e-05,
-      "loss": 4.0447,
+      "learning_rate": 6.038746908491344e-06,
+      "loss": 2.5734,
       "step": 5060
     },
     {
       "epoch": 2.84,
-      "learning_rate": 6.059356966199506e-05,
-      "loss": 3.8971,
+      "learning_rate": 6.018136850783182e-06,
+      "loss": 2.2567,
       "step": 5061
     },
     {
       "epoch": 2.84,
-      "learning_rate": 6.038746908491344e-05,
-      "loss": 3.9026,
+      "learning_rate": 5.997526793075021e-06,
+      "loss": 2.3449,
       "step": 5062
     },
     {
       "epoch": 2.84,
-      "learning_rate": 6.0181368507831826e-05,
-      "loss": 3.7493,
+      "learning_rate": 5.97691673536686e-06,
+      "loss": 2.282,
       "step": 5063
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.9975267930750205e-05,
-      "loss": 3.8572,
+      "learning_rate": 5.956306677658697e-06,
+      "loss": 2.3835,
       "step": 5064
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.976916735366859e-05,
-      "loss": 3.5972,
+      "learning_rate": 5.9356966199505365e-06,
+      "loss": 1.6963,
       "step": 5065
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.956306677658697e-05,
-      "loss": 3.4988,
+      "learning_rate": 5.915086562242375e-06,
+      "loss": 1.76,
       "step": 5066
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.9356966199505365e-05,
-      "loss": 3.2196,
+      "learning_rate": 5.894476504534213e-06,
+      "loss": 1.6243,
       "step": 5067
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.9150865622423744e-05,
-      "loss": 3.5891,
+      "learning_rate": 5.873866446826051e-06,
+      "loss": 1.7335,
       "step": 5068
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.894476504534213e-05,
-      "loss": 4.0832,
+      "learning_rate": 5.85325638911789e-06,
+      "loss": 3.1063,
       "step": 5069
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.873866446826051e-05,
-      "loss": 4.3453,
+      "learning_rate": 5.832646331409728e-06,
+      "loss": 3.3056,
       "step": 5070
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.8532563891178896e-05,
-      "loss": 4.2198,
+      "learning_rate": 5.812036273701567e-06,
+      "loss": 3.33,
       "step": 5071
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.8326463314097276e-05,
-      "loss": 4.0503,
+      "learning_rate": 5.7914262159934045e-06,
+      "loss": 2.8813,
       "step": 5072
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.812036273701567e-05,
-      "loss": 4.0209,
+      "learning_rate": 5.770816158285244e-06,
+      "loss": 2.9737,
       "step": 5073
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.791426215993405e-05,
-      "loss": 4.1641,
+      "learning_rate": 5.750206100577082e-06,
+      "loss": 2.9732,
       "step": 5074
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.7708161582852435e-05,
-      "loss": 3.8484,
+      "learning_rate": 5.72959604286892e-06,
+      "loss": 3.0141,
       "step": 5075
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.7502061005770815e-05,
-      "loss": 4.0826,
+      "learning_rate": 5.7089859851607586e-06,
+      "loss": 2.9984,
       "step": 5076
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.72959604286892e-05,
-      "loss": 4.3472,
+      "learning_rate": 5.688375927452598e-06,
+      "loss": 3.27,
       "step": 5077
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.708985985160758e-05,
-      "loss": 4.1495,
+      "learning_rate": 5.667765869744435e-06,
+      "loss": 2.9603,
       "step": 5078
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.6883759274525974e-05,
-      "loss": 4.0718,
+      "learning_rate": 5.647155812036274e-06,
+      "loss": 2.9237,
       "step": 5079
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.667765869744435e-05,
-      "loss": 3.9429,
+      "learning_rate": 5.626545754328113e-06,
+      "loss": 2.8017,
       "step": 5080
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.647155812036274e-05,
-      "loss": 4.1381,
+      "learning_rate": 5.605935696619951e-06,
+      "loss": 3.1472,
       "step": 5081
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.626545754328112e-05,
-      "loss": 3.5712,
+      "learning_rate": 5.585325638911789e-06,
+      "loss": 2.4794,
       "step": 5082
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.6059356966199506e-05,
-      "loss": 4.3062,
+      "learning_rate": 5.5647155812036275e-06,
+      "loss": 3.3236,
       "step": 5083
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.585325638911789e-05,
-      "loss": 4.0753,
+      "learning_rate": 5.544105523495466e-06,
+      "loss": 3.1498,
       "step": 5084
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.564715581203628e-05,
-      "loss": 3.9187,
+      "learning_rate": 5.523495465787305e-06,
+      "loss": 2.8378,
       "step": 5085
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.544105523495466e-05,
-      "loss": 3.8732,
+      "learning_rate": 5.502885408079142e-06,
+      "loss": 2.6053,
       "step": 5086
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.5234954657873044e-05,
-      "loss": 3.9193,
+      "learning_rate": 5.4822753503709815e-06,
+      "loss": 2.6109,
       "step": 5087
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.5028854080791424e-05,
-      "loss": 4.1241,
+      "learning_rate": 5.46166529266282e-06,
+      "loss": 3.1169,
       "step": 5088
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.482275350370981e-05,
-      "loss": 3.8449,
+      "learning_rate": 5.441055234954658e-06,
+      "loss": 2.7629,
       "step": 5089
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.46166529266282e-05,
-      "loss": 3.8768,
+      "learning_rate": 5.4204451772464964e-06,
+      "loss": 2.7383,
       "step": 5090
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.441055234954658e-05,
-      "loss": 3.767,
+      "learning_rate": 5.399835119538335e-06,
+      "loss": 2.6762,
       "step": 5091
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.420445177246496e-05,
-      "loss": 4.0612,
+      "learning_rate": 5.379225061830173e-06,
+      "loss": 2.898,
       "step": 5092
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.399835119538335e-05,
-      "loss": 4.0057,
+      "learning_rate": 5.358615004122012e-06,
+      "loss": 2.8954,
       "step": 5093
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.379225061830173e-05,
-      "loss": 4.0735,
+      "learning_rate": 5.33800494641385e-06,
+      "loss": 2.8884,
       "step": 5094
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.3586150041220115e-05,
-      "loss": 3.7822,
+      "learning_rate": 5.317394888705689e-06,
+      "loss": 2.6273,
       "step": 5095
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.33800494641385e-05,
-      "loss": 3.906,
+      "learning_rate": 5.296784830997527e-06,
+      "loss": 2.814,
       "step": 5096
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.317394888705689e-05,
-      "loss": 3.8806,
+      "learning_rate": 5.276174773289365e-06,
+      "loss": 2.7076,
       "step": 5097
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.296784830997527e-05,
-      "loss": 3.9278,
+      "learning_rate": 5.255564715581204e-06,
+      "loss": 2.6034,
       "step": 5098
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.2761747732893654e-05,
-      "loss": 3.937,
+      "learning_rate": 5.234954657873043e-06,
+      "loss": 2.6131,
       "step": 5099
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.255564715581203e-05,
-      "loss": 3.8321,
+      "learning_rate": 5.21434460016488e-06,
+      "loss": 2.6736,
       "step": 5100
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.234954657873042e-05,
-      "loss": 3.9065,
+      "learning_rate": 5.193734542456719e-06,
+      "loss": 2.6867,
       "step": 5101
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.2143446001648806e-05,
-      "loss": 3.8293,
+      "learning_rate": 5.173124484748558e-06,
+      "loss": 2.6614,
       "step": 5102
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.193734542456719e-05,
-      "loss": 4.2167,
+      "learning_rate": 5.152514427040396e-06,
+      "loss": 2.7635,
       "step": 5103
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.173124484748557e-05,
-      "loss": 3.7697,
+      "learning_rate": 5.131904369332234e-06,
+      "loss": 2.5807,
       "step": 5104
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.152514427040396e-05,
-      "loss": 3.6977,
+      "learning_rate": 5.111294311624073e-06,
+      "loss": 2.326,
       "step": 5105
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.131904369332234e-05,
-      "loss": 3.8337,
+      "learning_rate": 5.090684253915911e-06,
+      "loss": 2.4377,
       "step": 5106
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.1112943116240724e-05,
-      "loss": 3.8683,
+      "learning_rate": 5.07007419620775e-06,
+      "loss": 2.4645,
       "step": 5107
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.090684253915911e-05,
-      "loss": 3.9502,
+      "learning_rate": 5.0494641384995875e-06,
+      "loss": 2.6588,
       "step": 5108
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.07007419620775e-05,
-      "loss": 3.849,
+      "learning_rate": 5.028854080791427e-06,
+      "loss": 2.5451,
       "step": 5109
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.0494641384995876e-05,
-      "loss": 3.7362,
+      "learning_rate": 5.008244023083265e-06,
+      "loss": 2.4675,
       "step": 5110
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.028854080791426e-05,
-      "loss": 3.6482,
+      "learning_rate": 4.987633965375103e-06,
+      "loss": 2.0445,
       "step": 5111
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.008244023083264e-05,
-      "loss": 3.7275,
+      "learning_rate": 4.9670239076669415e-06,
+      "loss": 1.9924,
       "step": 5112
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.9876339653751036e-05,
-      "loss": 3.5425,
+      "learning_rate": 4.94641384995878e-06,
+      "loss": 2.0357,
       "step": 5113
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.9670239076669415e-05,
-      "loss": 3.5925,
+      "learning_rate": 4.925803792250618e-06,
+      "loss": 1.9034,
       "step": 5114
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.94641384995878e-05,
-      "loss": 4.0113,
+      "learning_rate": 4.905193734542457e-06,
+      "loss": 2.267,
       "step": 5115
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.925803792250618e-05,
-      "loss": 3.9936,
+      "learning_rate": 4.884583676834295e-06,
+      "loss": 1.7643,
       "step": 5116
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.905193734542457e-05,
-      "loss": 3.6822,
+      "learning_rate": 4.863973619126134e-06,
+      "loss": 1.8202,
       "step": 5117
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.884583676834295e-05,
-      "loss": 3.1375,
+      "learning_rate": 4.843363561417972e-06,
+      "loss": 1.8264,
       "step": 5118
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.863973619126134e-05,
-      "loss": 4.5436,
+      "learning_rate": 4.8227535037098104e-06,
+      "loss": 3.5489,
       "step": 5119
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.843363561417972e-05,
-      "loss": 4.263,
+      "learning_rate": 4.802143446001649e-06,
+      "loss": 3.348,
       "step": 5120
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.8227535037098106e-05,
-      "loss": 3.9375,
+      "learning_rate": 4.781533388293488e-06,
+      "loss": 3.0055,
       "step": 5121
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.8021434460016486e-05,
-      "loss": 4.0512,
+      "learning_rate": 4.760923330585325e-06,
+      "loss": 3.0438,
       "step": 5122
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.781533388293487e-05,
-      "loss": 4.1574,
+      "learning_rate": 4.7403132728771645e-06,
+      "loss": 3.1617,
       "step": 5123
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.760923330585325e-05,
-      "loss": 3.9939,
+      "learning_rate": 4.719703215169003e-06,
+      "loss": 3.0364,
       "step": 5124
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.7403132728771645e-05,
-      "loss": 4.2656,
+      "learning_rate": 4.699093157460841e-06,
+      "loss": 3.217,
       "step": 5125
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.7197032151690024e-05,
-      "loss": 4.2514,
+      "learning_rate": 4.678483099752679e-06,
+      "loss": 3.1693,
       "step": 5126
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.699093157460841e-05,
-      "loss": 4.0657,
+      "learning_rate": 4.657873042044518e-06,
+      "loss": 3.0886,
       "step": 5127
     },
     {
       "epoch": 2.87,
-      "learning_rate": 4.678483099752679e-05,
-      "loss": 4.2185,
+      "learning_rate": 4.637262984336356e-06,
+      "loss": 3.2217,
       "step": 5128
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.657873042044518e-05,
-      "loss": 4.0487,
+      "learning_rate": 4.616652926628195e-06,
+      "loss": 3.065,
       "step": 5129
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.6372629843363556e-05,
-      "loss": 3.9993,
+      "learning_rate": 4.5960428689200326e-06,
+      "loss": 2.8974,
       "step": 5130
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.616652926628195e-05,
-      "loss": 4.2062,
+      "learning_rate": 4.575432811211872e-06,
+      "loss": 3.3696,
       "step": 5131
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.596042868920033e-05,
-      "loss": 3.8972,
+      "learning_rate": 4.55482275350371e-06,
+      "loss": 2.8114,
       "step": 5132
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.5754328112118715e-05,
-      "loss": 4.0557,
+      "learning_rate": 4.534212695795548e-06,
+      "loss": 2.9336,
       "step": 5133
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.5548227535037095e-05,
-      "loss": 3.8048,
+      "learning_rate": 4.5136026380873874e-06,
+      "loss": 2.8956,
       "step": 5134
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.534212695795548e-05,
-      "loss": 4.1197,
+      "learning_rate": 4.492992580379225e-06,
+      "loss": 2.9862,
       "step": 5135
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.513602638087387e-05,
-      "loss": 3.9957,
+      "learning_rate": 4.472382522671064e-06,
+      "loss": 3.2831,
       "step": 5136
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.4929925803792254e-05,
-      "loss": 3.9668,
+      "learning_rate": 4.451772464962902e-06,
+      "loss": 2.8944,
       "step": 5137
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.472382522671064e-05,
-      "loss": 3.8691,
+      "learning_rate": 4.431162407254741e-06,
+      "loss": 2.6725,
       "step": 5138
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.451772464962902e-05,
-      "loss": 3.9509,
+      "learning_rate": 4.410552349546579e-06,
+      "loss": 3.0144,
       "step": 5139
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.4311624072547406e-05,
-      "loss": 3.8592,
+      "learning_rate": 4.389942291838418e-06,
+      "loss": 2.7725,
       "step": 5140
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.4105523495465786e-05,
-      "loss": 4.1966,
+      "learning_rate": 4.3693322341302555e-06,
+      "loss": 2.99,
       "step": 5141
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.389942291838417e-05,
-      "loss": 4.3523,
+      "learning_rate": 4.348722176422095e-06,
+      "loss": 3.3938,
       "step": 5142
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.369332234130256e-05,
-      "loss": 4.2822,
+      "learning_rate": 4.328112118713933e-06,
+      "loss": 2.996,
       "step": 5143
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.3487221764220945e-05,
-      "loss": 4.1066,
+      "learning_rate": 4.307502061005771e-06,
+      "loss": 3.0724,
       "step": 5144
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.3281121187139325e-05,
-      "loss": 3.9248,
+      "learning_rate": 4.2868920032976096e-06,
+      "loss": 2.8293,
       "step": 5145
     },
     {
       "epoch": 2.88,
-      "learning_rate": 4.307502061005771e-05,
-      "loss": 3.7842,
+      "learning_rate": 4.266281945589448e-06,
+      "loss": 2.5969,
       "step": 5146
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.286892003297609e-05,
-      "loss": 4.0734,
+      "learning_rate": 4.245671887881286e-06,
+      "loss": 2.832,
       "step": 5147
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.2662819455894484e-05,
-      "loss": 3.8602,
+      "learning_rate": 4.225061830173125e-06,
+      "loss": 2.7046,
       "step": 5148
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.245671887881286e-05,
-      "loss": 3.8822,
+      "learning_rate": 4.204451772464963e-06,
+      "loss": 2.7692,
       "step": 5149
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.225061830173125e-05,
-      "loss": 4.1478,
+      "learning_rate": 4.183841714756802e-06,
+      "loss": 2.8769,
       "step": 5150
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.204451772464963e-05,
-      "loss": 3.8257,
+      "learning_rate": 4.16323165704864e-06,
+      "loss": 2.5867,
       "step": 5151
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.1838417147568016e-05,
-      "loss": 3.903,
+      "learning_rate": 4.1426215993404785e-06,
+      "loss": 2.7033,
       "step": 5152
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.1632316570486395e-05,
-      "loss": 4.215,
+      "learning_rate": 4.122011541632317e-06,
+      "loss": 3.1186,
       "step": 5153
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.142621599340479e-05,
-      "loss": 4.1005,
+      "learning_rate": 4.101401483924155e-06,
+      "loss": 2.8603,
       "step": 5154
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.122011541632317e-05,
-      "loss": 3.8093,
+      "learning_rate": 4.080791426215993e-06,
+      "loss": 2.597,
       "step": 5155
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.1014014839241554e-05,
-      "loss": 3.552,
+      "learning_rate": 4.0601813685078325e-06,
+      "loss": 2.2935,
       "step": 5156
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.0807914262159934e-05,
-      "loss": 3.8922,
+      "learning_rate": 4.03957131079967e-06,
+      "loss": 2.4955,
       "step": 5157
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.060181368507832e-05,
-      "loss": 4.5108,
+      "learning_rate": 4.018961253091509e-06,
+      "loss": 3.152,
       "step": 5158
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.03957131079967e-05,
-      "loss": 4.0738,
+      "learning_rate": 3.9983511953833474e-06,
+      "loss": 2.8606,
       "step": 5159
     },
     {
       "epoch": 2.89,
-      "learning_rate": 4.018961253091509e-05,
-      "loss": 4.0982,
+      "learning_rate": 3.977741137675186e-06,
+      "loss": 2.6385,
       "step": 5160
     },
     {
       "epoch": 2.89,
-      "learning_rate": 3.998351195383347e-05,
-      "loss": 3.9686,
+      "learning_rate": 3.957131079967024e-06,
+      "loss": 2.5382,
       "step": 5161
     },
     {
       "epoch": 2.89,
-      "learning_rate": 3.977741137675186e-05,
-      "loss": 4.1081,
+      "learning_rate": 3.936521022258863e-06,
+      "loss": 2.4438,
       "step": 5162
     },
     {
       "epoch": 2.89,
-      "learning_rate": 3.957131079967024e-05,
-      "loss": 3.6158,
+      "learning_rate": 3.915910964550701e-06,
+      "loss": 1.9206,
       "step": 5163
     },
     {
       "epoch": 2.89,
-      "learning_rate": 3.9365210222588625e-05,
-      "loss": 3.846,
+      "learning_rate": 3.89530090684254e-06,
+      "loss": 2.2609,
       "step": 5164
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.9159109645507004e-05,
-      "loss": 3.8718,
+      "learning_rate": 3.874690849134378e-06,
+      "loss": 2.1876,
       "step": 5165
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.89530090684254e-05,
-      "loss": 3.6072,
+      "learning_rate": 3.854080791426216e-06,
+      "loss": 1.8856,
       "step": 5166
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.874690849134378e-05,
-      "loss": 3.2825,
+      "learning_rate": 3.833470733718055e-06,
+      "loss": 1.7425,
       "step": 5167
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.8540807914262164e-05,
-      "loss": 3.6551,
+      "learning_rate": 3.8128606760098934e-06,
+      "loss": 1.9527,
       "step": 5168
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.833470733718054e-05,
-      "loss": 4.431,
+      "learning_rate": 3.7922506183017312e-06,
+      "loss": 3.5306,
       "step": 5169
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.812860676009893e-05,
-      "loss": 3.8475,
+      "learning_rate": 3.77164056059357e-06,
+      "loss": 3.0068,
       "step": 5170
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.792250618301731e-05,
-      "loss": 3.9162,
+      "learning_rate": 3.7510305028854083e-06,
+      "loss": 3.1075,
       "step": 5171
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.77164056059357e-05,
-      "loss": 4.2181,
+      "learning_rate": 3.730420445177247e-06,
+      "loss": 3.3154,
       "step": 5172
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.751030502885408e-05,
-      "loss": 3.7959,
+      "learning_rate": 3.709810387469085e-06,
+      "loss": 2.8456,
       "step": 5173
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.730420445177247e-05,
-      "loss": 4.1886,
+      "learning_rate": 3.6892003297609236e-06,
+      "loss": 3.1885,
       "step": 5174
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.709810387469085e-05,
-      "loss": 3.8347,
+      "learning_rate": 3.668590272052762e-06,
+      "loss": 2.8839,
       "step": 5175
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.6892003297609234e-05,
-      "loss": 4.2774,
+      "learning_rate": 3.6479802143446006e-06,
+      "loss": 3.373,
       "step": 5176
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.6685902720527614e-05,
-      "loss": 4.1383,
+      "learning_rate": 3.6273701566364385e-06,
+      "loss": 3.4457,
       "step": 5177
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.647980214344601e-05,
-      "loss": 4.1758,
+      "learning_rate": 3.606760098928277e-06,
+      "loss": 3.3475,
       "step": 5178
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.6273701566364386e-05,
-      "loss": 3.8702,
+      "learning_rate": 3.5861500412201155e-06,
+      "loss": 3.0519,
       "step": 5179
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.606760098928277e-05,
-      "loss": 4.0338,
+      "learning_rate": 3.5655399835119542e-06,
+      "loss": 2.9679,
       "step": 5180
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.586150041220115e-05,
-      "loss": 3.9308,
+      "learning_rate": 3.544929925803792e-06,
+      "loss": 2.9905,
       "step": 5181
     },
     {
       "epoch": 2.9,
-      "learning_rate": 3.565539983511954e-05,
-      "loss": 4.1627,
+      "learning_rate": 3.5243198680956312e-06,
+      "loss": 2.9698,
       "step": 5182
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.5449299258037925e-05,
-      "loss": 3.9854,
+      "learning_rate": 3.503709810387469e-06,
+      "loss": 3.0342,
       "step": 5183
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.524319868095631e-05,
-      "loss": 3.7288,
+      "learning_rate": 3.483099752679308e-06,
+      "loss": 2.745,
       "step": 5184
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.503709810387469e-05,
-      "loss": 4.0555,
+      "learning_rate": 3.462489694971146e-06,
+      "loss": 3.1327,
       "step": 5185
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.483099752679308e-05,
-      "loss": 4.0818,
+      "learning_rate": 3.441879637262985e-06,
+      "loss": 3.0156,
       "step": 5186
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.462489694971146e-05,
-      "loss": 4.2081,
+      "learning_rate": 3.4212695795548227e-06,
+      "loss": 3.14,
       "step": 5187
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.441879637262984e-05,
-      "loss": 3.7993,
+      "learning_rate": 3.4006595218466614e-06,
+      "loss": 2.9217,
       "step": 5188
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.421269579554823e-05,
-      "loss": 3.9008,
+      "learning_rate": 3.3800494641384997e-06,
+      "loss": 3.0457,
       "step": 5189
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.4006595218466616e-05,
-      "loss": 3.8967,
+      "learning_rate": 3.3594394064303385e-06,
+      "loss": 2.9877,
       "step": 5190
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.3800494641384996e-05,
-      "loss": 3.7799,
+      "learning_rate": 3.3388293487221763e-06,
+      "loss": 2.7438,
       "step": 5191
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.359439406430338e-05,
-      "loss": 3.8599,
+      "learning_rate": 3.318219291014015e-06,
+      "loss": 2.9854,
       "step": 5192
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.338829348722176e-05,
-      "loss": 3.9188,
+      "learning_rate": 3.2976092333058533e-06,
+      "loss": 2.8281,
       "step": 5193
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.318219291014015e-05,
-      "loss": 4.1941,
+      "learning_rate": 3.276999175597692e-06,
+      "loss": 3.1121,
       "step": 5194
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.2976092333058534e-05,
-      "loss": 4.1234,
+      "learning_rate": 3.25638911788953e-06,
+      "loss": 3.2604,
       "step": 5195
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.276999175597692e-05,
-      "loss": 3.903,
+      "learning_rate": 3.2357790601813687e-06,
+      "loss": 2.8613,
       "step": 5196
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.25638911788953e-05,
-      "loss": 4.0267,
+      "learning_rate": 3.215169002473207e-06,
+      "loss": 3.0842,
       "step": 5197
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.235779060181369e-05,
-      "loss": 3.8702,
+      "learning_rate": 3.1945589447650457e-06,
+      "loss": 2.9538,
       "step": 5198
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.2151690024732066e-05,
-      "loss": 4.0343,
+      "learning_rate": 3.1739488870568836e-06,
+      "loss": 2.9861,
       "step": 5199
     },
     {
       "epoch": 2.91,
-      "learning_rate": 3.194558944765045e-05,
-      "loss": 4.1882,
+      "learning_rate": 3.1533388293487223e-06,
+      "loss": 3.0756,
       "step": 5200
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.173948887056884e-05,
-      "loss": 4.3037,
+      "learning_rate": 3.1327287716405606e-06,
+      "loss": 3.2077,
       "step": 5201
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.1533388293487225e-05,
-      "loss": 3.8102,
+      "learning_rate": 3.1121187139323993e-06,
+      "loss": 2.8276,
       "step": 5202
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.1327287716405605e-05,
-      "loss": 3.8718,
+      "learning_rate": 3.0915086562242376e-06,
+      "loss": 2.823,
       "step": 5203
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.112118713932399e-05,
-      "loss": 3.7969,
+      "learning_rate": 3.0708985985160763e-06,
+      "loss": 2.6899,
       "step": 5204
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.091508656224238e-05,
-      "loss": 3.9047,
+      "learning_rate": 3.0502885408079146e-06,
+      "loss": 2.861,
       "step": 5205
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.070898598516076e-05,
-      "loss": 3.7546,
+      "learning_rate": 3.029678483099753e-06,
+      "loss": 2.6917,
       "step": 5206
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.0502885408079144e-05,
-      "loss": 3.7169,
+      "learning_rate": 3.009068425391591e-06,
+      "loss": 2.6855,
       "step": 5207
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.029678483099753e-05,
-      "loss": 4.0869,
+      "learning_rate": 2.98845836768343e-06,
+      "loss": 2.9842,
       "step": 5208
     },
     {
       "epoch": 2.92,
-      "learning_rate": 3.0090684253915913e-05,
-      "loss": 4.241,
+      "learning_rate": 2.9678483099752682e-06,
+      "loss": 2.9879,
       "step": 5209
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.9884583676834296e-05,
-      "loss": 3.8232,
+      "learning_rate": 2.9472382522671065e-06,
+      "loss": 2.7565,
       "step": 5210
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.9678483099752682e-05,
-      "loss": 3.5642,
+      "learning_rate": 2.926628194558945e-06,
+      "loss": 2.2791,
       "step": 5211
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.9472382522671065e-05,
-      "loss": 3.7639,
+      "learning_rate": 2.9060181368507835e-06,
+      "loss": 2.5758,
       "step": 5212
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.9266281945589448e-05,
-      "loss": 3.3243,
+      "learning_rate": 2.885408079142622e-06,
+      "loss": 2.0448,
       "step": 5213
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.9060181368507835e-05,
-      "loss": 3.7364,
+      "learning_rate": 2.86479802143446e-06,
+      "loss": 2.2905,
       "step": 5214
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.8854080791426218e-05,
-      "loss": 4.0159,
+      "learning_rate": 2.844187963726299e-06,
+      "loss": 2.4114,
       "step": 5215
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.86479802143446e-05,
-      "loss": 3.3586,
+      "learning_rate": 2.823577906018137e-06,
+      "loss": 1.7599,
       "step": 5216
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.8441879637262987e-05,
-      "loss": 3.6416,
+      "learning_rate": 2.8029678483099755e-06,
+      "loss": 1.8952,
       "step": 5217
     },
     {
       "epoch": 2.92,
-      "learning_rate": 2.823577906018137e-05,
-      "loss": 3.1922,
+      "learning_rate": 2.7823577906018138e-06,
+      "loss": 1.8052,
       "step": 5218
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.8029678483099753e-05,
-      "loss": 4.3309,
+      "learning_rate": 2.7617477328936525e-06,
+      "loss": 3.6316,
       "step": 5219
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.782357790601814e-05,
-      "loss": 4.2992,
+      "learning_rate": 2.7411376751854908e-06,
+      "loss": 3.4847,
       "step": 5220
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.7617477328936522e-05,
-      "loss": 4.0482,
+      "learning_rate": 2.720527617477329e-06,
+      "loss": 3.2803,
       "step": 5221
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.7411376751854905e-05,
-      "loss": 3.8889,
+      "learning_rate": 2.6999175597691674e-06,
+      "loss": 3.1195,
       "step": 5222
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.720527617477329e-05,
-      "loss": 4.3519,
+      "learning_rate": 2.679307502061006e-06,
+      "loss": 3.7024,
       "step": 5223
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.6999175597691674e-05,
-      "loss": 4.2582,
+      "learning_rate": 2.6586974443528444e-06,
+      "loss": 3.3643,
       "step": 5224
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.6793075020610057e-05,
-      "loss": 4.2566,
+      "learning_rate": 2.6380873866446827e-06,
+      "loss": 3.3714,
       "step": 5225
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.6586974443528444e-05,
-      "loss": 4.0225,
+      "learning_rate": 2.6174773289365214e-06,
+      "loss": 3.2946,
       "step": 5226
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.6380873866446827e-05,
-      "loss": 4.2688,
+      "learning_rate": 2.5968672712283597e-06,
+      "loss": 3.4441,
       "step": 5227
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.617477328936521e-05,
-      "loss": 3.8302,
+      "learning_rate": 2.576257213520198e-06,
+      "loss": 3.073,
       "step": 5228
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.5968672712283596e-05,
-      "loss": 3.8832,
+      "learning_rate": 2.5556471558120363e-06,
+      "loss": 3.0482,
       "step": 5229
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.576257213520198e-05,
-      "loss": 4.0177,
+      "learning_rate": 2.535037098103875e-06,
+      "loss": 3.1307,
       "step": 5230
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.5556471558120362e-05,
-      "loss": 3.7836,
+      "learning_rate": 2.5144270403957133e-06,
+      "loss": 2.9329,
       "step": 5231
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.535037098103875e-05,
-      "loss": 4.173,
+      "learning_rate": 2.4938169826875516e-06,
+      "loss": 3.2326,
       "step": 5232
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.514427040395713e-05,
-      "loss": 3.8221,
+      "learning_rate": 2.47320692497939e-06,
+      "loss": 3.0146,
       "step": 5233
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.4938169826875518e-05,
-      "loss": 3.8122,
+      "learning_rate": 2.4525968672712286e-06,
+      "loss": 2.9402,
       "step": 5234
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.47320692497939e-05,
-      "loss": 4.0053,
+      "learning_rate": 2.431986809563067e-06,
+      "loss": 3.1025,
       "step": 5235
     },
     {
       "epoch": 2.93,
-      "learning_rate": 2.4525968672712284e-05,
-      "loss": 4.2828,
+      "learning_rate": 2.4113767518549052e-06,
+      "loss": 3.3869,
       "step": 5236
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.431986809563067e-05,
-      "loss": 3.7595,
+      "learning_rate": 2.390766694146744e-06,
+      "loss": 3.0018,
       "step": 5237
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.4113767518549053e-05,
-      "loss": 4.2288,
+      "learning_rate": 2.3701566364385822e-06,
+      "loss": 3.3171,
       "step": 5238
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.3907666941467436e-05,
-      "loss": 3.8431,
+      "learning_rate": 2.3495465787304205e-06,
+      "loss": 2.9157,
       "step": 5239
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.3701566364385822e-05,
-      "loss": 3.8778,
+      "learning_rate": 2.328936521022259e-06,
+      "loss": 3.0481,
       "step": 5240
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.3495465787304205e-05,
-      "loss": 4.0112,
+      "learning_rate": 2.3083264633140976e-06,
+      "loss": 3.2009,
       "step": 5241
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.328936521022259e-05,
-      "loss": 3.7304,
+      "learning_rate": 2.287716405605936e-06,
+      "loss": 2.8119,
       "step": 5242
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.3083264633140975e-05,
-      "loss": 3.9794,
+      "learning_rate": 2.267106347897774e-06,
+      "loss": 3.1136,
       "step": 5243
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.2877164056059358e-05,
-      "loss": 3.9738,
+      "learning_rate": 2.2464962901896124e-06,
+      "loss": 3.0552,
       "step": 5244
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.267106347897774e-05,
-      "loss": 3.7783,
+      "learning_rate": 2.225886232481451e-06,
+      "loss": 3.0085,
       "step": 5245
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.2464962901896127e-05,
-      "loss": 3.8678,
+      "learning_rate": 2.2052761747732895e-06,
+      "loss": 3.0095,
       "step": 5246
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.225886232481451e-05,
-      "loss": 4.1737,
+      "learning_rate": 2.1846661170651278e-06,
+      "loss": 3.3293,
       "step": 5247
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.2052761747732893e-05,
-      "loss": 3.8529,
+      "learning_rate": 2.1640560593569665e-06,
+      "loss": 2.838,
       "step": 5248
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.184666117065128e-05,
-      "loss": 4.0275,
+      "learning_rate": 2.1434460016488048e-06,
+      "loss": 3.1049,
       "step": 5249
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.1640560593569662e-05,
-      "loss": 3.9318,
+      "learning_rate": 2.122835943940643e-06,
+      "loss": 3.1776,
       "step": 5250
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.1434460016488045e-05,
-      "loss": 3.7196,
+      "learning_rate": 2.1022258862324814e-06,
+      "loss": 2.6988,
       "step": 5251
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.122835943940643e-05,
-      "loss": 3.9434,
+      "learning_rate": 2.08161582852432e-06,
+      "loss": 3.1222,
       "step": 5252
     },
     {
       "epoch": 2.94,
-      "learning_rate": 2.1022258862324815e-05,
-      "loss": 4.0266,
+      "learning_rate": 2.0610057708161584e-06,
+      "loss": 3.0,
       "step": 5253
     },
     {
       "epoch": 2.95,
-      "learning_rate": 2.0816158285243198e-05,
-      "loss": 3.8123,
+      "learning_rate": 2.0403957131079967e-06,
+      "loss": 2.7389,
       "step": 5254
     },
     {
       "epoch": 2.95,
-      "learning_rate": 2.0610057708161584e-05,
-      "loss": 3.7585,
+      "learning_rate": 2.019785655399835e-06,
+      "loss": 2.7525,
       "step": 5255
     },
     {
       "epoch": 2.95,
-      "learning_rate": 2.0403957131079967e-05,
-      "loss": 4.0173,
+      "learning_rate": 1.9991755976916737e-06,
+      "loss": 2.8647,
       "step": 5256
     },
     {
       "epoch": 2.95,
-      "learning_rate": 2.019785655399835e-05,
-      "loss": 3.8176,
+      "learning_rate": 1.978565539983512e-06,
+      "loss": 2.6262,
       "step": 5257
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.9991755976916736e-05,
-      "loss": 3.799,
+      "learning_rate": 1.9579554822753503e-06,
+      "loss": 2.7546,
       "step": 5258
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.978565539983512e-05,
-      "loss": 3.9159,
+      "learning_rate": 1.937345424567189e-06,
+      "loss": 3.0131,
       "step": 5259
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.9579554822753502e-05,
-      "loss": 4.1107,
+      "learning_rate": 1.9167353668590273e-06,
+      "loss": 3.0204,
       "step": 5260
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.937345424567189e-05,
-      "loss": 3.306,
+      "learning_rate": 1.8961253091508656e-06,
+      "loss": 2.1039,
       "step": 5261
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.916735366859027e-05,
-      "loss": 3.9473,
+      "learning_rate": 1.8755152514427041e-06,
+      "loss": 2.6302,
       "step": 5262
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.8961253091508655e-05,
-      "loss": 3.8644,
+      "learning_rate": 1.8549051937345424e-06,
+      "loss": 2.3502,
       "step": 5263
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.875515251442704e-05,
-      "loss": 3.3905,
+      "learning_rate": 1.834295136026381e-06,
+      "loss": 2.0444,
       "step": 5264
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.8549051937345424e-05,
-      "loss": 3.1935,
+      "learning_rate": 1.8136850783182192e-06,
+      "loss": 1.818,
       "step": 5265
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.8342951360263807e-05,
-      "loss": 3.4585,
+      "learning_rate": 1.7930750206100577e-06,
+      "loss": 1.9051,
       "step": 5266
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.8136850783182193e-05,
-      "loss": 3.234,
+      "learning_rate": 1.772464962901896e-06,
+      "loss": 1.788,
       "step": 5267
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.7930750206100576e-05,
-      "loss": 3.3882,
+      "learning_rate": 1.7518549051937346e-06,
+      "loss": 1.7212,
       "step": 5268
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.7724649629018963e-05,
-      "loss": 4.3939,
+      "learning_rate": 1.731244847485573e-06,
+      "loss": 3.7102,
       "step": 5269
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.7518549051937346e-05,
-      "loss": 4.0379,
+      "learning_rate": 1.7106347897774114e-06,
+      "loss": 3.4155,
       "step": 5270
     },
     {
       "epoch": 2.95,
-      "learning_rate": 1.731244847485573e-05,
-      "loss": 4.0738,
+      "learning_rate": 1.6900247320692499e-06,
+      "loss": 3.4086,
       "step": 5271
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.7106347897774115e-05,
-      "loss": 4.175,
+      "learning_rate": 1.6694146743610882e-06,
+      "loss": 3.3386,
       "step": 5272
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.6900247320692498e-05,
-      "loss": 3.9153,
+      "learning_rate": 1.6488046166529267e-06,
+      "loss": 3.1525,
       "step": 5273
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.669414674361088e-05,
-      "loss": 4.3006,
+      "learning_rate": 1.628194558944765e-06,
+      "loss": 3.6711,
       "step": 5274
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.6488046166529267e-05,
-      "loss": 3.983,
+      "learning_rate": 1.6075845012366035e-06,
+      "loss": 3.3028,
       "step": 5275
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.628194558944765e-05,
-      "loss": 4.0034,
+      "learning_rate": 1.5869744435284418e-06,
+      "loss": 3.3338,
       "step": 5276
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.6075845012366033e-05,
-      "loss": 4.418,
+      "learning_rate": 1.5663643858202803e-06,
+      "loss": 3.7264,
       "step": 5277
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.586974443528442e-05,
-      "loss": 4.0683,
+      "learning_rate": 1.5457543281121188e-06,
+      "loss": 3.4774,
       "step": 5278
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.5663643858202802e-05,
-      "loss": 3.8721,
+      "learning_rate": 1.5251442704039573e-06,
+      "loss": 3.1375,
       "step": 5279
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.545754328112119e-05,
-      "loss": 3.7206,
+      "learning_rate": 1.5045342126957956e-06,
+      "loss": 3.0396,
       "step": 5280
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.5251442704039572e-05,
-      "loss": 3.8008,
+      "learning_rate": 1.4839241549876341e-06,
+      "loss": 3.0385,
       "step": 5281
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.5045342126957956e-05,
-      "loss": 4.0717,
+      "learning_rate": 1.4633140972794724e-06,
+      "loss": 3.2645,
       "step": 5282
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.4839241549876341e-05,
-      "loss": 4.1223,
+      "learning_rate": 1.442704039571311e-06,
+      "loss": 3.3026,
       "step": 5283
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.4633140972794724e-05,
-      "loss": 4.2332,
+      "learning_rate": 1.4220939818631494e-06,
+      "loss": 3.4283,
       "step": 5284
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.4427040395713109e-05,
-      "loss": 3.8893,
+      "learning_rate": 1.4014839241549877e-06,
+      "loss": 3.2519,
       "step": 5285
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.4220939818631493e-05,
-      "loss": 4.0638,
+      "learning_rate": 1.3808738664468262e-06,
+      "loss": 3.3053,
       "step": 5286
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.4014839241549876e-05,
-      "loss": 4.0834,
+      "learning_rate": 1.3602638087386645e-06,
+      "loss": 3.3407,
       "step": 5287
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.3808738664468261e-05,
-      "loss": 3.8107,
+      "learning_rate": 1.339653751030503e-06,
+      "loss": 3.06,
       "step": 5288
     },
     {
       "epoch": 2.96,
-      "learning_rate": 1.3602638087386646e-05,
-      "loss": 4.0629,
+      "learning_rate": 1.3190436933223413e-06,
+      "loss": 3.324,
       "step": 5289
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.3396537510305029e-05,
-      "loss": 3.9057,
+      "learning_rate": 1.2984336356141798e-06,
+      "loss": 3.0158,
       "step": 5290
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.3190436933223413e-05,
-      "loss": 3.8925,
+      "learning_rate": 1.2778235779060181e-06,
+      "loss": 3.1668,
       "step": 5291
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.2984336356141798e-05,
-      "loss": 3.965,
+      "learning_rate": 1.2572135201978567e-06,
+      "loss": 3.1565,
       "step": 5292
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.2778235779060181e-05,
-      "loss": 4.0511,
+      "learning_rate": 1.236603462489695e-06,
+      "loss": 3.3154,
       "step": 5293
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.2572135201978566e-05,
-      "loss": 4.1415,
+      "learning_rate": 1.2159934047815335e-06,
+      "loss": 3.41,
       "step": 5294
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.236603462489695e-05,
-      "loss": 4.3386,
+      "learning_rate": 1.195383347073372e-06,
+      "loss": 3.531,
       "step": 5295
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.2159934047815335e-05,
-      "loss": 3.8571,
+      "learning_rate": 1.1747732893652103e-06,
+      "loss": 3.0163,
       "step": 5296
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.1953833470733718e-05,
-      "loss": 3.6936,
+      "learning_rate": 1.1541632316570488e-06,
+      "loss": 2.9016,
       "step": 5297
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.1747732893652103e-05,
-      "loss": 4.0286,
+      "learning_rate": 1.133553173948887e-06,
+      "loss": 3.2808,
       "step": 5298
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.1541632316570487e-05,
-      "loss": 4.0293,
+      "learning_rate": 1.1129431162407256e-06,
+      "loss": 3.1558,
       "step": 5299
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.133553173948887e-05,
-      "loss": 4.0922,
+      "learning_rate": 1.0923330585325639e-06,
+      "loss": 3.1435,
       "step": 5300
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.1129431162407255e-05,
-      "loss": 4.3619,
+      "learning_rate": 1.0717230008244024e-06,
+      "loss": 3.353,
       "step": 5301
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.092333058532564e-05,
-      "loss": 3.8714,
+      "learning_rate": 1.0511129431162407e-06,
+      "loss": 2.8817,
       "step": 5302
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.0717230008244023e-05,
-      "loss": 4.1102,
+      "learning_rate": 1.0305028854080792e-06,
+      "loss": 3.1479,
       "step": 5303
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.0511129431162407e-05,
-      "loss": 4.0078,
+      "learning_rate": 1.0098928276999175e-06,
+      "loss": 3.1745,
       "step": 5304
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.0305028854080792e-05,
-      "loss": 3.5789,
+      "learning_rate": 9.89282769991756e-07,
+      "loss": 2.8111,
       "step": 5305
     },
     {
       "epoch": 2.97,
-      "learning_rate": 1.0098928276999175e-05,
-      "loss": 3.5775,
+      "learning_rate": 9.686727122835945e-07,
+      "loss": 2.6645,
       "step": 5306
     },
     {
       "epoch": 2.97,
-      "learning_rate": 9.89282769991756e-06,
-      "loss": 4.02,
+      "learning_rate": 9.480626545754328e-07,
+      "loss": 3.1673,
       "step": 5307
     },
     {
       "epoch": 2.98,
-      "learning_rate": 9.686727122835944e-06,
-      "loss": 3.9004,
+      "learning_rate": 9.274525968672712e-07,
+      "loss": 3.0789,
       "step": 5308
     },
     {
       "epoch": 2.98,
-      "learning_rate": 9.480626545754327e-06,
-      "loss": 3.8523,
+      "learning_rate": 9.068425391591096e-07,
+      "loss": 2.8944,
       "step": 5309
     },
     {
       "epoch": 2.98,
-      "learning_rate": 9.274525968672712e-06,
-      "loss": 4.0244,
+      "learning_rate": 8.86232481450948e-07,
+      "loss": 2.8598,
       "step": 5310
     },
     {
       "epoch": 2.98,
-      "learning_rate": 9.068425391591097e-06,
-      "loss": 3.9133,
+      "learning_rate": 8.656224237427865e-07,
+      "loss": 2.8603,
       "step": 5311
     },
     {
       "epoch": 2.98,
-      "learning_rate": 8.862324814509481e-06,
-      "loss": 3.8616,
+      "learning_rate": 8.450123660346249e-07,
+      "loss": 2.8169,
       "step": 5312
     },
     {
       "epoch": 2.98,
-      "learning_rate": 8.656224237427864e-06,
-      "loss": 3.7944,
+      "learning_rate": 8.244023083264633e-07,
+      "loss": 2.3045,
       "step": 5313
     },
     {
       "epoch": 2.98,
-      "learning_rate": 8.450123660346249e-06,
-      "loss": 3.51,
+      "learning_rate": 8.037922506183017e-07,
+      "loss": 2.3683,
       "step": 5314
     },
     {
       "epoch": 2.98,
-      "learning_rate": 8.244023083264634e-06,
-      "loss": 3.4487,
+      "learning_rate": 7.831821929101401e-07,
+      "loss": 2.1184,
       "step": 5315
     },
     {
       "epoch": 2.98,
-      "learning_rate": 8.037922506183017e-06,
-      "loss": 3.4854,
+      "learning_rate": 7.625721352019787e-07,
+      "loss": 2.053,
       "step": 5316
     },
     {
       "epoch": 2.98,
-      "learning_rate": 7.831821929101401e-06,
-      "loss": 3.4285,
+      "learning_rate": 7.419620774938171e-07,
+      "loss": 1.8888,
       "step": 5317
     },
     {
       "epoch": 2.98,
-      "learning_rate": 7.625721352019786e-06,
-      "loss": 3.1365,
+      "learning_rate": 7.213520197856555e-07,
+      "loss": 1.8517,
       "step": 5318
     },
     {
       "epoch": 2.98,
-      "learning_rate": 7.4196207749381706e-06,
-      "loss": 4.1348,
+      "learning_rate": 7.007419620774939e-07,
+      "loss": 3.6159,
       "step": 5319
     },
     {
       "epoch": 2.98,
-      "learning_rate": 7.213520197856554e-06,
-      "loss": 3.9531,
+      "learning_rate": 6.801319043693323e-07,
+      "loss": 3.3571,
       "step": 5320
     },
     {
       "epoch": 2.98,
-      "learning_rate": 7.007419620774938e-06,
-      "loss": 3.8508,
+      "learning_rate": 6.595218466611707e-07,
+      "loss": 3.2404,
       "step": 5321
     },
     {
       "epoch": 2.98,
-      "learning_rate": 6.801319043693323e-06,
-      "loss": 4.1207,
+      "learning_rate": 6.389117889530091e-07,
+      "loss": 3.4741,
       "step": 5322
     },
     {
       "epoch": 2.98,
-      "learning_rate": 6.595218466611707e-06,
-      "loss": 4.0903,
+      "learning_rate": 6.183017312448475e-07,
+      "loss": 3.5234,
       "step": 5323
     },
     {
       "epoch": 2.98,
-      "learning_rate": 6.3891178895300905e-06,
-      "loss": 4.1029,
+      "learning_rate": 5.97691673536686e-07,
+      "loss": 3.4236,
       "step": 5324
     },
     {
       "epoch": 2.98,
-      "learning_rate": 6.183017312448475e-06,
-      "loss": 4.0303,
+      "learning_rate": 5.770816158285244e-07,
+      "loss": 3.5056,
       "step": 5325
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.976916735366859e-06,
-      "loss": 3.9227,
+      "learning_rate": 5.564715581203628e-07,
+      "loss": 3.2966,
       "step": 5326
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.770816158285244e-06,
-      "loss": 3.9048,
+      "learning_rate": 5.358615004122012e-07,
+      "loss": 3.2723,
       "step": 5327
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.5647155812036275e-06,
-      "loss": 4.0601,
+      "learning_rate": 5.152514427040396e-07,
+      "loss": 3.3741,
       "step": 5328
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.358615004122011e-06,
-      "loss": 4.1536,
+      "learning_rate": 4.94641384995878e-07,
+      "loss": 3.4641,
       "step": 5329
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.152514427040396e-06,
-      "loss": 4.0253,
+      "learning_rate": 4.740313272877164e-07,
+      "loss": 3.3985,
       "step": 5330
     },
     {
       "epoch": 2.99,
-      "learning_rate": 4.94641384995878e-06,
-      "loss": 4.1337,
+      "learning_rate": 4.534212695795548e-07,
+      "loss": 3.5321,
       "step": 5331
     },
     {
       "epoch": 2.99,
-      "learning_rate": 4.740313272877164e-06,
-      "loss": 3.8758,
+      "learning_rate": 4.3281121187139327e-07,
+      "loss": 3.1327,
       "step": 5332
     },
     {
       "epoch": 2.99,
-      "learning_rate": 4.534212695795548e-06,
-      "loss": 4.0683,
+      "learning_rate": 4.1220115416323167e-07,
+      "loss": 3.434,
       "step": 5333
     },
     {
       "epoch": 2.99,
-      "learning_rate": 4.328112118713932e-06,
-      "loss": 3.7324,
+      "learning_rate": 3.9159109645507007e-07,
+      "loss": 3.0494,
       "step": 5334
     },
     {
       "epoch": 2.99,
-      "learning_rate": 4.122011541632317e-06,
-      "loss": 3.9437,
+      "learning_rate": 3.7098103874690853e-07,
+      "loss": 3.3216,
       "step": 5335
     },
     {
       "epoch": 2.99,
-      "learning_rate": 3.915910964550701e-06,
-      "loss": 3.7429,
+      "learning_rate": 3.5037098103874693e-07,
+      "loss": 3.135,
       "step": 5336
     },
     {
       "epoch": 2.99,
-      "learning_rate": 3.7098103874690853e-06,
-      "loss": 4.181,
+      "learning_rate": 3.2976092333058533e-07,
+      "loss": 3.4183,
       "step": 5337
     },
     {
       "epoch": 2.99,
-      "learning_rate": 3.503709810387469e-06,
-      "loss": 3.9541,
+      "learning_rate": 3.0915086562242374e-07,
+      "loss": 3.2047,
       "step": 5338
     },
     {
       "epoch": 2.99,
-      "learning_rate": 3.2976092333058533e-06,
-      "loss": 4.0066,
+      "learning_rate": 2.885408079142622e-07,
+      "loss": 3.233,
       "step": 5339
     },
     {
       "epoch": 2.99,
-      "learning_rate": 3.0915086562242376e-06,
-      "loss": 3.8814,
+      "learning_rate": 2.679307502061006e-07,
+      "loss": 3.1799,
       "step": 5340
     },
     {
       "epoch": 2.99,
-      "learning_rate": 2.885408079142622e-06,
-      "loss": 3.797,
+      "learning_rate": 2.47320692497939e-07,
+      "loss": 3.05,
       "step": 5341
     },
     {
       "epoch": 2.99,
-      "learning_rate": 2.6793075020610057e-06,
-      "loss": 3.5633,
+      "learning_rate": 2.267106347897774e-07,
+      "loss": 2.7407,
       "step": 5342
     },
     {
       "epoch": 2.99,
-      "learning_rate": 2.47320692497939e-06,
-      "loss": 3.6679,
+      "learning_rate": 2.0610057708161583e-07,
+      "loss": 2.8923,
       "step": 5343
     },
     {
       "epoch": 3.0,
-      "learning_rate": 2.267106347897774e-06,
-      "loss": 3.9039,
+      "learning_rate": 1.8549051937345426e-07,
+      "loss": 3.1625,
       "step": 5344
     },
     {
       "epoch": 3.0,
-      "learning_rate": 2.0610057708161584e-06,
-      "loss": 3.8726,
+      "learning_rate": 1.6488046166529267e-07,
+      "loss": 3.0095,
       "step": 5345
     },
     {
       "epoch": 3.0,
-      "learning_rate": 1.8549051937345426e-06,
-      "loss": 3.9625,
+      "learning_rate": 1.442704039571311e-07,
+      "loss": 2.9184,
       "step": 5346
     },
     {
       "epoch": 3.0,
-      "learning_rate": 1.6488046166529267e-06,
-      "loss": 3.9666,
+      "learning_rate": 1.236603462489695e-07,
+      "loss": 2.9979,
       "step": 5347
     },
     {
       "epoch": 3.0,
-      "learning_rate": 1.442704039571311e-06,
-      "loss": 3.9754,
+      "learning_rate": 1.0305028854080792e-07,
+      "loss": 2.9077,
       "step": 5348
     },
     {
       "epoch": 3.0,
-      "learning_rate": 1.236603462489695e-06,
-      "loss": 3.648,
+      "learning_rate": 8.244023083264633e-08,
+      "loss": 2.4611,
       "step": 5349
     },
     {
       "epoch": 3.0,
-      "learning_rate": 1.0305028854080792e-06,
-      "loss": 3.5492,
+      "learning_rate": 6.183017312448475e-08,
+      "loss": 2.1063,
       "step": 5350
     },
     {
       "epoch": 3.0,
-      "learning_rate": 8.244023083264633e-07,
-      "loss": 3.1506,
+      "learning_rate": 4.122011541632317e-08,
+      "loss": 1.8737,
       "step": 5351
     },
     {
       "epoch": 3.0,
-      "learning_rate": 6.183017312448475e-07,
-      "loss": 3.1039,
+      "learning_rate": 2.0610057708161583e-08,
+      "loss": 1.9856,
       "step": 5352
     },
     {
       "epoch": 3.0,
       "step": 5352,
       "total_flos": 0.0,
-      "train_loss": 4.915239981070762,
-      "train_runtime": 30235.4571,
+      "train_loss": 2.6358130432342377,
+      "train_runtime": 30229.1632,
       "train_samples_per_second": 2.832,
       "train_steps_per_second": 0.177
     }