diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -9,26790 +9,26790 @@
   "log_history": [
     {
       "epoch": 0.0,
-      "learning_rate": 1.0000000000000001e-07,
-      "loss": 10.524,
+      "learning_rate": 3e-07,
+      "loss": 10.5843,
       "step": 1
     },
     {
       "epoch": 0.0,
-      "learning_rate": 2.0000000000000002e-07,
-      "loss": 10.5614,
+      "learning_rate": 6e-07,
+      "loss": 10.5849,
       "step": 2
     },
     {
       "epoch": 0.0,
-      "learning_rate": 3.0000000000000004e-07,
-      "loss": 10.5227,
+      "learning_rate": 9e-07,
+      "loss": 10.5561,
       "step": 3
     },
     {
       "epoch": 0.0,
-      "learning_rate": 4.0000000000000003e-07,
-      "loss": 10.526,
+      "learning_rate": 1.2e-06,
+      "loss": 10.5004,
       "step": 4
     },
     {
       "epoch": 0.01,
-      "learning_rate": 5.000000000000001e-07,
-      "loss": 10.5023,
+      "learning_rate": 1.4999999999999998e-06,
+      "loss": 10.4464,
       "step": 5
     },
     {
       "epoch": 0.01,
-      "learning_rate": 6.000000000000001e-07,
-      "loss": 10.4477,
+      "learning_rate": 1.8e-06,
+      "loss": 10.3393,
       "step": 6
     },
     {
       "epoch": 0.01,
-      "learning_rate": 7.000000000000001e-07,
-      "loss": 10.4006,
+      "learning_rate": 2.1e-06,
+      "loss": 10.3008,
       "step": 7
     },
     {
       "epoch": 0.01,
-      "learning_rate": 8.000000000000001e-07,
-      "loss": 10.3794,
+      "learning_rate": 2.4e-06,
+      "loss": 10.186,
       "step": 8
     },
     {
       "epoch": 0.01,
-      "learning_rate": 9e-07,
-      "loss": 10.3246,
+      "learning_rate": 2.6999999999999996e-06,
+      "loss": 10.1141,
       "step": 9
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.0000000000000002e-06,
-      "loss": 10.2437,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 9.991,
       "step": 10
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.1e-06,
-      "loss": 10.2119,
+      "learning_rate": 3.2999999999999993e-06,
+      "loss": 9.9681,
       "step": 11
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.2000000000000002e-06,
-      "loss": 10.2233,
+      "learning_rate": 3.6e-06,
+      "loss": 9.9675,
       "step": 12
     },
     {
       "epoch": 0.01,
-      "learning_rate": 1.3e-06,
-      "loss": 10.079,
+      "learning_rate": 3.899999999999999e-06,
+      "loss": 9.8234,
       "step": 13
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.4000000000000001e-06,
-      "loss": 10.0996,
+      "learning_rate": 4.2e-06,
+      "loss": 9.8416,
       "step": 14
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.5e-06,
-      "loss": 10.0111,
+      "learning_rate": 4.499999999999999e-06,
+      "loss": 9.727,
       "step": 15
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.6000000000000001e-06,
-      "loss": 9.9863,
+      "learning_rate": 4.8e-06,
+      "loss": 9.7406,
       "step": 16
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.7000000000000002e-06,
-      "loss": 9.9272,
+      "learning_rate": 5.1e-06,
+      "loss": 9.6556,
       "step": 17
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.8e-06,
-      "loss": 9.9023,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 9.6216,
       "step": 18
     },
     {
       "epoch": 0.02,
-      "learning_rate": 1.9e-06,
-      "loss": 9.8811,
+      "learning_rate": 5.7e-06,
+      "loss": 9.6108,
       "step": 19
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.0000000000000003e-06,
-      "loss": 9.8505,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 9.5681,
       "step": 20
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.1000000000000002e-06,
-      "loss": 9.8002,
+      "learning_rate": 6.3e-06,
+      "loss": 9.5502,
       "step": 21
     },
     {
       "epoch": 0.02,
-      "learning_rate": 2.2e-06,
-      "loss": 9.7338,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 9.4429,
       "step": 22
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.3e-06,
-      "loss": 9.7399,
+      "learning_rate": 6.899999999999999e-06,
+      "loss": 9.4237,
       "step": 23
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.4000000000000003e-06,
-      "loss": 9.7504,
+      "learning_rate": 7.2e-06,
+      "loss": 9.4884,
       "step": 24
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.5e-06,
-      "loss": 9.6684,
+      "learning_rate": 7.499999999999999e-06,
+      "loss": 9.3977,
       "step": 25
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.6e-06,
-      "loss": 9.7359,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 9.5144,
       "step": 26
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.7e-06,
-      "loss": 9.6496,
+      "learning_rate": 8.099999999999999e-06,
+      "loss": 9.3693,
       "step": 27
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.8000000000000003e-06,
-      "loss": 9.5758,
+      "learning_rate": 8.4e-06,
+      "loss": 9.2639,
       "step": 28
     },
     {
       "epoch": 0.03,
-      "learning_rate": 2.9e-06,
-      "loss": 9.5493,
+      "learning_rate": 8.7e-06,
+      "loss": 9.2363,
       "step": 29
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3e-06,
-      "loss": 9.5444,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 9.2209,
       "step": 30
     },
     {
       "epoch": 0.03,
-      "learning_rate": 3.1e-06,
-      "loss": 9.4931,
+      "learning_rate": 9.299999999999999e-06,
+      "loss": 9.2024,
       "step": 31
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.2000000000000003e-06,
-      "loss": 9.4048,
+      "learning_rate": 9.6e-06,
+      "loss": 9.1079,
       "step": 32
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.3e-06,
-      "loss": 9.4779,
+      "learning_rate": 9.9e-06,
+      "loss": 9.1735,
       "step": 33
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.4000000000000005e-06,
-      "loss": 9.4324,
+      "learning_rate": 1.02e-05,
+      "loss": 9.1361,
       "step": 34
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.5000000000000004e-06,
-      "loss": 9.3813,
+      "learning_rate": 1.05e-05,
+      "loss": 9.0699,
       "step": 35
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.6e-06,
-      "loss": 9.4286,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 9.1254,
       "step": 36
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.7e-06,
-      "loss": 9.3367,
+      "learning_rate": 1.1099999999999999e-05,
+      "loss": 9.0663,
       "step": 37
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.8e-06,
-      "loss": 9.3209,
+      "learning_rate": 1.14e-05,
+      "loss": 9.0342,
       "step": 38
     },
     {
       "epoch": 0.04,
-      "learning_rate": 3.9e-06,
-      "loss": 9.3266,
+      "learning_rate": 1.17e-05,
+      "loss": 9.0232,
       "step": 39
     },
     {
       "epoch": 0.04,
-      "learning_rate": 4.000000000000001e-06,
-      "loss": 9.2631,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 8.9028,
       "step": 40
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.1000000000000006e-06,
-      "loss": 9.1771,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 8.8606,
       "step": 41
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.2000000000000004e-06,
-      "loss": 9.2365,
+      "learning_rate": 1.26e-05,
+      "loss": 8.895,
       "step": 42
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.2999999999999995e-06,
-      "loss": 9.2191,
+      "learning_rate": 1.2899999999999998e-05,
+      "loss": 8.9092,
       "step": 43
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.4e-06,
-      "loss": 9.2209,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 8.9411,
       "step": 44
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.5e-06,
-      "loss": 9.1893,
+      "learning_rate": 1.3499999999999998e-05,
+      "loss": 8.918,
       "step": 45
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.6e-06,
-      "loss": 9.1305,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 8.8122,
       "step": 46
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.7e-06,
-      "loss": 9.1218,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 8.8811,
       "step": 47
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.800000000000001e-06,
-      "loss": 8.9528,
+      "learning_rate": 1.44e-05,
+      "loss": 8.6231,
       "step": 48
     },
     {
       "epoch": 0.05,
-      "learning_rate": 4.9000000000000005e-06,
-      "loss": 8.8499,
+      "learning_rate": 1.47e-05,
+      "loss": 8.5175,
       "step": 49
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5e-06,
-      "loss": 8.7024,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 8.2819,
       "step": 50
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.1e-06,
-      "loss": 9.3015,
+      "learning_rate": 1.53e-05,
+      "loss": 8.9997,
       "step": 51
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.2e-06,
-      "loss": 9.2259,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 8.87,
       "step": 52
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.3e-06,
-      "loss": 9.2597,
+      "learning_rate": 1.5899999999999997e-05,
+      "loss": 8.891,
       "step": 53
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.4e-06,
-      "loss": 9.2894,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 8.9198,
       "step": 54
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.500000000000001e-06,
-      "loss": 9.1694,
+      "learning_rate": 1.6499999999999998e-05,
+      "loss": 8.7707,
       "step": 55
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.600000000000001e-06,
-      "loss": 9.1126,
+      "learning_rate": 1.68e-05,
+      "loss": 8.7757,
       "step": 56
     },
     {
       "epoch": 0.06,
-      "learning_rate": 5.7000000000000005e-06,
-      "loss": 9.1066,
+      "learning_rate": 1.71e-05,
+      "loss": 8.7539,
       "step": 57
     },
     {
       "epoch": 0.07,
-      "learning_rate": 5.8e-06,
-      "loss": 9.1642,
+      "learning_rate": 1.74e-05,
+      "loss": 8.8285,
       "step": 58
     },
     {
       "epoch": 0.07,
-      "learning_rate": 5.9e-06,
-      "loss": 9.021,
+      "learning_rate": 1.7699999999999997e-05,
+      "loss": 8.6511,
       "step": 59
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6e-06,
-      "loss": 9.1068,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 8.6956,
       "step": 60
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.1e-06,
-      "loss": 9.0925,
+      "learning_rate": 1.8299999999999998e-05,
+      "loss": 8.6802,
       "step": 61
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.2e-06,
-      "loss": 9.0216,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 8.5848,
       "step": 62
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.300000000000001e-06,
-      "loss": 9.018,
+      "learning_rate": 1.89e-05,
+      "loss": 8.6135,
       "step": 63
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.4000000000000006e-06,
-      "loss": 9.0933,
+      "learning_rate": 1.92e-05,
+      "loss": 8.6287,
       "step": 64
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.5000000000000004e-06,
-      "loss": 8.9771,
+      "learning_rate": 1.95e-05,
+      "loss": 8.5552,
       "step": 65
     },
     {
       "epoch": 0.07,
-      "learning_rate": 6.6e-06,
-      "loss": 9.1396,
+      "learning_rate": 1.98e-05,
+      "loss": 8.6656,
       "step": 66
     },
     {
       "epoch": 0.08,
-      "learning_rate": 6.700000000000001e-06,
-      "loss": 9.057,
+      "learning_rate": 2.01e-05,
+      "loss": 8.5801,
       "step": 67
     },
     {
       "epoch": 0.08,
-      "learning_rate": 6.800000000000001e-06,
-      "loss": 9.0214,
+      "learning_rate": 2.04e-05,
+      "loss": 8.4858,
       "step": 68
     },
     {
       "epoch": 0.08,
-      "learning_rate": 6.900000000000001e-06,
-      "loss": 8.9321,
+      "learning_rate": 2.07e-05,
+      "loss": 8.4365,
       "step": 69
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.000000000000001e-06,
-      "loss": 8.9314,
+      "learning_rate": 2.1e-05,
+      "loss": 8.4603,
       "step": 70
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.1e-06,
-      "loss": 8.9654,
+      "learning_rate": 2.1299999999999996e-05,
+      "loss": 8.5072,
       "step": 71
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.2e-06,
-      "loss": 8.9496,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 8.4809,
       "step": 72
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.2999999999999996e-06,
-      "loss": 8.8592,
+      "learning_rate": 2.1899999999999997e-05,
+      "loss": 8.3213,
       "step": 73
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.4e-06,
-      "loss": 8.9446,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 8.4078,
       "step": 74
     },
     {
       "epoch": 0.08,
-      "learning_rate": 7.5e-06,
-      "loss": 8.788,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 8.2348,
       "step": 75
     },
     {
       "epoch": 0.09,
-      "learning_rate": 7.6e-06,
-      "loss": 8.9051,
+      "learning_rate": 2.28e-05,
+      "loss": 8.3265,
       "step": 76
     },
     {
       "epoch": 0.09,
-      "learning_rate": 7.7e-06,
-      "loss": 8.8099,
+      "learning_rate": 2.31e-05,
+      "loss": 8.2043,
       "step": 77
     },
     {
       "epoch": 0.09,
-      "learning_rate": 7.8e-06,
-      "loss": 8.7785,
+      "learning_rate": 2.34e-05,
+      "loss": 8.1556,
       "step": 78
     },
     {
       "epoch": 0.09,
-      "learning_rate": 7.9e-06,
-      "loss": 8.7538,
+      "learning_rate": 2.3699999999999997e-05,
+      "loss": 8.1806,
       "step": 79
     },
     {
       "epoch": 0.09,
-      "learning_rate": 8.000000000000001e-06,
-      "loss": 8.7966,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 8.1571,
       "step": 80
     },
     {
       "epoch": 0.09,
-      "learning_rate": 8.1e-06,
-      "loss": 8.7043,
+      "learning_rate": 2.4299999999999998e-05,
+      "loss": 8.0522,
       "step": 81
     },
     {
       "epoch": 0.09,
-      "learning_rate": 8.200000000000001e-06,
-      "loss": 8.6813,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 8.0318,
       "step": 82
     },
     {
       "epoch": 0.09,
-      "learning_rate": 8.3e-06,
-      "loss": 8.7635,
+      "learning_rate": 2.49e-05,
+      "loss": 8.1342,
       "step": 83
     },
     {
       "epoch": 0.09,
-      "learning_rate": 8.400000000000001e-06,
-      "loss": 8.6311,
+      "learning_rate": 2.52e-05,
+      "loss": 8.0017,
       "step": 84
     },
     {
       "epoch": 0.1,
-      "learning_rate": 8.500000000000002e-06,
-      "loss": 8.7172,
+      "learning_rate": 2.55e-05,
+      "loss": 8.048,
       "step": 85
     },
     {
       "epoch": 0.1,
-      "learning_rate": 8.599999999999999e-06,
-      "loss": 8.6846,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 7.9748,
       "step": 86
     },
     {
       "epoch": 0.1,
-      "learning_rate": 8.7e-06,
-      "loss": 8.6557,
+      "learning_rate": 2.6099999999999997e-05,
+      "loss": 7.9719,
       "step": 87
     },
     {
       "epoch": 0.1,
-      "learning_rate": 8.8e-06,
-      "loss": 8.6205,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 7.8933,
       "step": 88
     },
     {
       "epoch": 0.1,
-      "learning_rate": 8.9e-06,
-      "loss": 8.6242,
+      "learning_rate": 2.6699999999999995e-05,
+      "loss": 7.9207,
       "step": 89
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9e-06,
-      "loss": 8.6649,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 7.9082,
       "step": 90
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.100000000000001e-06,
-      "loss": 8.7687,
+      "learning_rate": 2.7299999999999996e-05,
+      "loss": 7.9977,
       "step": 91
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.2e-06,
-      "loss": 8.6335,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 7.8796,
       "step": 92
     },
     {
       "epoch": 0.1,
-      "learning_rate": 9.3e-06,
-      "loss": 8.4674,
+      "learning_rate": 2.7899999999999997e-05,
+      "loss": 7.6762,
       "step": 93
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.4e-06,
-      "loss": 8.5098,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 7.7885,
       "step": 94
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.5e-06,
-      "loss": 8.6162,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 7.8686,
       "step": 95
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.600000000000001e-06,
-      "loss": 8.556,
+      "learning_rate": 2.88e-05,
+      "loss": 7.766,
       "step": 96
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.7e-06,
-      "loss": 8.4114,
+      "learning_rate": 2.91e-05,
+      "loss": 7.6128,
       "step": 97
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.800000000000001e-06,
-      "loss": 8.4872,
+      "learning_rate": 2.94e-05,
+      "loss": 7.6648,
       "step": 98
     },
     {
       "epoch": 0.11,
-      "learning_rate": 9.900000000000002e-06,
-      "loss": 8.234,
+      "learning_rate": 2.97e-05,
+      "loss": 7.3737,
       "step": 99
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1e-05,
-      "loss": 7.9769,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 7.0701,
       "step": 100
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.0100000000000002e-05,
-      "loss": 8.6942,
+      "learning_rate": 3.0299999999999998e-05,
+      "loss": 8.1863,
       "step": 101
     },
     {
       "epoch": 0.11,
-      "learning_rate": 1.02e-05,
-      "loss": 8.742,
+      "learning_rate": 3.06e-05,
+      "loss": 8.1449,
       "step": 102
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.03e-05,
-      "loss": 8.6504,
+      "learning_rate": 3.09e-05,
+      "loss": 7.9056,
       "step": 103
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.04e-05,
-      "loss": 8.5534,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 7.7374,
       "step": 104
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.05e-05,
-      "loss": 8.4726,
+      "learning_rate": 3.149999999999999e-05,
+      "loss": 7.6069,
       "step": 105
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.06e-05,
-      "loss": 8.513,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 7.6667,
       "step": 106
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.0700000000000001e-05,
-      "loss": 8.4577,
+      "learning_rate": 3.2099999999999994e-05,
+      "loss": 7.6818,
       "step": 107
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.08e-05,
-      "loss": 8.5058,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 7.6812,
       "step": 108
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.09e-05,
-      "loss": 8.3679,
+      "learning_rate": 3.2699999999999995e-05,
+      "loss": 7.491,
       "step": 109
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.1000000000000001e-05,
-      "loss": 8.4654,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 7.5533,
       "step": 110
     },
     {
       "epoch": 0.12,
-      "learning_rate": 1.11e-05,
-      "loss": 8.4735,
+      "learning_rate": 3.3299999999999996e-05,
+      "loss": 7.5346,
       "step": 111
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.1200000000000001e-05,
-      "loss": 8.4169,
+      "learning_rate": 3.36e-05,
+      "loss": 7.5446,
       "step": 112
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.13e-05,
-      "loss": 8.3146,
+      "learning_rate": 3.39e-05,
+      "loss": 7.4654,
       "step": 113
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.1400000000000001e-05,
-      "loss": 8.3611,
+      "learning_rate": 3.42e-05,
+      "loss": 7.4675,
       "step": 114
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.1500000000000002e-05,
-      "loss": 8.3565,
+      "learning_rate": 3.45e-05,
+      "loss": 7.4809,
       "step": 115
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.16e-05,
-      "loss": 8.2942,
+      "learning_rate": 3.48e-05,
+      "loss": 7.3496,
       "step": 116
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.1700000000000001e-05,
-      "loss": 8.3655,
+      "learning_rate": 3.51e-05,
+      "loss": 7.4598,
       "step": 117
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.18e-05,
-      "loss": 8.345,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 7.3909,
       "step": 118
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.19e-05,
-      "loss": 8.3265,
+      "learning_rate": 3.5699999999999994e-05,
+      "loss": 7.3825,
       "step": 119
     },
     {
       "epoch": 0.13,
-      "learning_rate": 1.2e-05,
-      "loss": 8.355,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 7.4126,
       "step": 120
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.2100000000000001e-05,
-      "loss": 8.2575,
+      "learning_rate": 3.6299999999999995e-05,
+      "loss": 7.2976,
       "step": 121
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.22e-05,
-      "loss": 8.3287,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 7.4234,
       "step": 122
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.23e-05,
-      "loss": 8.1648,
+      "learning_rate": 3.6899999999999996e-05,
+      "loss": 7.1765,
       "step": 123
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.24e-05,
-      "loss": 8.2166,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 7.3166,
       "step": 124
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.25e-05,
-      "loss": 8.3121,
+      "learning_rate": 3.75e-05,
+      "loss": 7.3641,
       "step": 125
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.2600000000000001e-05,
-      "loss": 8.303,
+      "learning_rate": 3.78e-05,
+      "loss": 7.3974,
       "step": 126
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.27e-05,
-      "loss": 8.3508,
+      "learning_rate": 3.81e-05,
+      "loss": 7.4304,
       "step": 127
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.2800000000000001e-05,
-      "loss": 8.3003,
+      "learning_rate": 3.84e-05,
+      "loss": 7.3458,
       "step": 128
     },
     {
       "epoch": 0.14,
-      "learning_rate": 1.29e-05,
-      "loss": 8.1763,
+      "learning_rate": 3.87e-05,
+      "loss": 7.1876,
       "step": 129
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3000000000000001e-05,
-      "loss": 8.1459,
+      "learning_rate": 3.9e-05,
+      "loss": 7.2804,
       "step": 130
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3100000000000002e-05,
-      "loss": 8.1524,
+      "learning_rate": 3.93e-05,
+      "loss": 7.2696,
       "step": 131
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.32e-05,
-      "loss": 8.1598,
+      "learning_rate": 3.96e-05,
+      "loss": 7.281,
       "step": 132
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3300000000000001e-05,
-      "loss": 8.1088,
+      "learning_rate": 3.99e-05,
+      "loss": 7.2097,
       "step": 133
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3400000000000002e-05,
-      "loss": 8.1396,
+      "learning_rate": 4.02e-05,
+      "loss": 7.2607,
       "step": 134
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3500000000000001e-05,
-      "loss": 8.0239,
+      "learning_rate": 4.05e-05,
+      "loss": 7.0624,
       "step": 135
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3600000000000002e-05,
-      "loss": 8.0443,
+      "learning_rate": 4.08e-05,
+      "loss": 7.1155,
       "step": 136
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3700000000000001e-05,
-      "loss": 8.1425,
+      "learning_rate": 4.11e-05,
+      "loss": 7.2613,
       "step": 137
     },
     {
       "epoch": 0.15,
-      "learning_rate": 1.3800000000000002e-05,
-      "loss": 7.9823,
+      "learning_rate": 4.14e-05,
+      "loss": 7.1187,
       "step": 138
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.3900000000000002e-05,
-      "loss": 8.0094,
+      "learning_rate": 4.17e-05,
+      "loss": 7.0948,
       "step": 139
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.4000000000000001e-05,
-      "loss": 7.9532,
+      "learning_rate": 4.2e-05,
+      "loss": 7.0814,
       "step": 140
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.4099999999999999e-05,
-      "loss": 8.1178,
+      "learning_rate": 4.229999999999999e-05,
+      "loss": 7.2709,
       "step": 141
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.42e-05,
-      "loss": 7.9582,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 7.184,
       "step": 142
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.43e-05,
-      "loss": 7.944,
+      "learning_rate": 4.289999999999999e-05,
+      "loss": 7.1044,
       "step": 143
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.44e-05,
-      "loss": 7.847,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 6.9312,
       "step": 144
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.45e-05,
-      "loss": 7.7702,
+      "learning_rate": 4.3499999999999993e-05,
+      "loss": 6.985,
       "step": 145
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.4599999999999999e-05,
-      "loss": 7.7257,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 6.8889,
       "step": 146
     },
     {
       "epoch": 0.16,
-      "learning_rate": 1.47e-05,
-      "loss": 7.8284,
+      "learning_rate": 4.4099999999999995e-05,
+      "loss": 7.1607,
       "step": 147
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.48e-05,
-      "loss": 7.4674,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 6.6531,
       "step": 148
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.49e-05,
-      "loss": 7.595,
+      "learning_rate": 4.4699999999999996e-05,
+      "loss": 6.9546,
       "step": 149
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.5e-05,
-      "loss": 7.2391,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 6.4883,
       "step": 150
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.51e-05,
-      "loss": 8.4578,
+      "learning_rate": 4.5299999999999997e-05,
+      "loss": 7.1638,
       "step": 151
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.52e-05,
-      "loss": 8.3299,
+      "learning_rate": 4.56e-05,
+      "loss": 7.11,
       "step": 152
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.53e-05,
-      "loss": 8.1074,
+      "learning_rate": 4.59e-05,
+      "loss": 7.1339,
       "step": 153
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.54e-05,
-      "loss": 7.8631,
+      "learning_rate": 4.62e-05,
+      "loss": 6.9843,
       "step": 154
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.55e-05,
-      "loss": 7.8258,
+      "learning_rate": 4.65e-05,
+      "loss": 7.0078,
       "step": 155
     },
     {
       "epoch": 0.17,
-      "learning_rate": 1.56e-05,
-      "loss": 7.8242,
+      "learning_rate": 4.68e-05,
+      "loss": 7.0631,
       "step": 156
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.5700000000000002e-05,
-      "loss": 7.977,
+      "learning_rate": 4.709999999999999e-05,
+      "loss": 7.2238,
       "step": 157
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.58e-05,
-      "loss": 7.9007,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 7.1583,
       "step": 158
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.59e-05,
-      "loss": 7.8585,
+      "learning_rate": 4.7699999999999994e-05,
+      "loss": 7.149,
       "step": 159
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.6000000000000003e-05,
-      "loss": 7.8228,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 7.1311,
       "step": 160
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.6100000000000002e-05,
-      "loss": 7.7343,
+      "learning_rate": 4.8299999999999995e-05,
+      "loss": 6.9915,
       "step": 161
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.62e-05,
-      "loss": 7.7316,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 6.9884,
       "step": 162
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.63e-05,
-      "loss": 7.7656,
+      "learning_rate": 4.8899999999999996e-05,
+      "loss": 7.0466,
       "step": 163
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.6400000000000002e-05,
-      "loss": 7.735,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 7.0318,
       "step": 164
     },
     {
       "epoch": 0.18,
-      "learning_rate": 1.65e-05,
-      "loss": 7.7168,
+      "learning_rate": 4.95e-05,
+      "loss": 7.0411,
       "step": 165
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.66e-05,
-      "loss": 7.7344,
+      "learning_rate": 4.98e-05,
+      "loss": 7.0154,
       "step": 166
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.6700000000000003e-05,
-      "loss": 7.6596,
+      "learning_rate": 5.01e-05,
+      "loss": 6.9601,
       "step": 167
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.6800000000000002e-05,
-      "loss": 7.7084,
+      "learning_rate": 5.04e-05,
+      "loss": 7.0747,
       "step": 168
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.69e-05,
-      "loss": 7.6877,
+      "learning_rate": 5.07e-05,
+      "loss": 7.1022,
       "step": 169
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.7000000000000003e-05,
-      "loss": 7.6282,
+      "learning_rate": 5.1e-05,
+      "loss": 6.9956,
       "step": 170
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.7100000000000002e-05,
-      "loss": 7.7096,
+      "learning_rate": 5.13e-05,
+      "loss": 7.158,
       "step": 171
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.7199999999999998e-05,
-      "loss": 7.7142,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 7.1497,
       "step": 172
     },
     {
       "epoch": 0.19,
-      "learning_rate": 1.73e-05,
-      "loss": 7.6729,
+      "learning_rate": 5.1899999999999994e-05,
+      "loss": 7.1428,
       "step": 173
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.74e-05,
-      "loss": 7.6169,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 7.0342,
       "step": 174
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.75e-05,
-      "loss": 7.5037,
+      "learning_rate": 5.2499999999999995e-05,
+      "loss": 6.917,
       "step": 175
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.76e-05,
-      "loss": 7.5583,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 7.0527,
       "step": 176
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.77e-05,
-      "loss": 7.5254,
+      "learning_rate": 5.309999999999999e-05,
+      "loss": 7.0097,
       "step": 177
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.78e-05,
-      "loss": 7.5086,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 6.9545,
       "step": 178
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.79e-05,
-      "loss": 7.5812,
+      "learning_rate": 5.369999999999999e-05,
+      "loss": 7.092,
       "step": 179
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.8e-05,
-      "loss": 7.5655,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 7.1091,
       "step": 180
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.81e-05,
-      "loss": 7.5851,
+      "learning_rate": 5.429999999999999e-05,
+      "loss": 7.0887,
       "step": 181
     },
     {
       "epoch": 0.2,
-      "learning_rate": 1.8200000000000002e-05,
-      "loss": 7.4313,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 6.9406,
       "step": 182
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.83e-05,
-      "loss": 7.4639,
+      "learning_rate": 5.489999999999999e-05,
+      "loss": 7.0012,
       "step": 183
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.84e-05,
-      "loss": 7.5653,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 7.1445,
       "step": 184
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.85e-05,
-      "loss": 7.4413,
+      "learning_rate": 5.5499999999999994e-05,
+      "loss": 7.0467,
       "step": 185
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.86e-05,
-      "loss": 7.5958,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 7.1512,
       "step": 186
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.87e-05,
-      "loss": 7.4953,
+      "learning_rate": 5.6099999999999995e-05,
+      "loss": 7.0861,
       "step": 187
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.88e-05,
-      "loss": 7.3635,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 6.8862,
       "step": 188
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.8900000000000002e-05,
-      "loss": 7.3922,
+      "learning_rate": 5.6699999999999996e-05,
+      "loss": 6.9668,
       "step": 189
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.9e-05,
-      "loss": 7.3989,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 6.9771,
       "step": 190
     },
     {
       "epoch": 0.21,
-      "learning_rate": 1.91e-05,
-      "loss": 7.3746,
+      "learning_rate": 5.73e-05,
+      "loss": 6.9926,
       "step": 191
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.9200000000000003e-05,
-      "loss": 7.5187,
+      "learning_rate": 5.76e-05,
+      "loss": 7.1739,
       "step": 192
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.93e-05,
-      "loss": 7.4679,
+      "learning_rate": 5.79e-05,
+      "loss": 7.1662,
       "step": 193
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.94e-05,
-      "loss": 7.3724,
+      "learning_rate": 5.82e-05,
+      "loss": 6.9745,
       "step": 194
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.9500000000000003e-05,
-      "loss": 7.125,
+      "learning_rate": 5.85e-05,
+      "loss": 6.7375,
       "step": 195
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.9600000000000002e-05,
-      "loss": 7.4214,
+      "learning_rate": 5.88e-05,
+      "loss": 7.1047,
       "step": 196
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.97e-05,
-      "loss": 7.2072,
+      "learning_rate": 5.91e-05,
+      "loss": 6.872,
       "step": 197
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.9800000000000004e-05,
-      "loss": 7.1467,
+      "learning_rate": 5.94e-05,
+      "loss": 6.7412,
       "step": 198
     },
     {
       "epoch": 0.22,
-      "learning_rate": 1.9900000000000003e-05,
-      "loss": 7.0155,
+      "learning_rate": 5.97e-05,
+      "loss": 6.7171,
       "step": 199
     },
     {
       "epoch": 0.22,
-      "learning_rate": 2e-05,
-      "loss": 6.807,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 6.3975,
       "step": 200
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.01e-05,
-      "loss": 7.4155,
+      "learning_rate": 6.0299999999999995e-05,
+      "loss": 7.1875,
       "step": 201
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.0200000000000003e-05,
-      "loss": 7.4642,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 7.2464,
       "step": 202
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.0300000000000002e-05,
-      "loss": 7.3614,
+      "learning_rate": 6.0899999999999996e-05,
+      "loss": 7.1186,
       "step": 203
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.04e-05,
-      "loss": 7.3381,
+      "learning_rate": 6.12e-05,
+      "loss": 7.1043,
       "step": 204
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.05e-05,
-      "loss": 7.3074,
+      "learning_rate": 6.149999999999999e-05,
+      "loss": 7.0696,
       "step": 205
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.06e-05,
-      "loss": 7.3032,
+      "learning_rate": 6.18e-05,
+      "loss": 7.0071,
       "step": 206
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.07e-05,
-      "loss": 7.3341,
+      "learning_rate": 6.209999999999999e-05,
+      "loss": 7.0313,
       "step": 207
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.08e-05,
-      "loss": 7.2348,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 6.9866,
       "step": 208
     },
     {
       "epoch": 0.23,
-      "learning_rate": 2.09e-05,
-      "loss": 7.2553,
+      "learning_rate": 6.269999999999999e-05,
+      "loss": 6.9826,
       "step": 209
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.1e-05,
-      "loss": 7.2414,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 6.9566,
       "step": 210
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.11e-05,
-      "loss": 7.1534,
+      "learning_rate": 6.33e-05,
+      "loss": 6.8547,
       "step": 211
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.12e-05,
-      "loss": 7.1922,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 6.9183,
       "step": 212
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.13e-05,
-      "loss": 7.2692,
+      "learning_rate": 6.39e-05,
+      "loss": 6.9949,
       "step": 213
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.1400000000000002e-05,
-      "loss": 7.2595,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 7.0497,
       "step": 214
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.15e-05,
-      "loss": 7.2171,
+      "learning_rate": 6.45e-05,
+      "loss": 6.9648,
       "step": 215
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.16e-05,
-      "loss": 7.2786,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 7.083,
       "step": 216
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.1700000000000002e-05,
-      "loss": 7.1871,
+      "learning_rate": 6.51e-05,
+      "loss": 6.9544,
       "step": 217
     },
     {
       "epoch": 0.24,
-      "learning_rate": 2.18e-05,
-      "loss": 7.1653,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 6.9771,
       "step": 218
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.19e-05,
-      "loss": 7.1119,
+      "learning_rate": 6.57e-05,
+      "loss": 6.9191,
       "step": 219
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.2000000000000003e-05,
-      "loss": 7.3016,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 7.1231,
       "step": 220
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.2100000000000002e-05,
-      "loss": 7.2058,
+      "learning_rate": 6.63e-05,
+      "loss": 7.0082,
       "step": 221
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.22e-05,
-      "loss": 7.1048,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 6.919,
       "step": 222
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.23e-05,
-      "loss": 7.2102,
+      "learning_rate": 6.69e-05,
+      "loss": 7.0184,
       "step": 223
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.2400000000000002e-05,
-      "loss": 7.16,
+      "learning_rate": 6.72e-05,
+      "loss": 6.9846,
       "step": 224
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.25e-05,
-      "loss": 7.1367,
+      "learning_rate": 6.75e-05,
+      "loss": 6.9799,
       "step": 225
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.26e-05,
-      "loss": 7.2791,
+      "learning_rate": 6.78e-05,
+      "loss": 7.1756,
       "step": 226
     },
     {
       "epoch": 0.25,
-      "learning_rate": 2.2700000000000003e-05,
-      "loss": 7.1357,
+      "learning_rate": 6.81e-05,
+      "loss": 6.9899,
       "step": 227
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.2800000000000002e-05,
-      "loss": 7.1371,
+      "learning_rate": 6.84e-05,
+      "loss": 6.9766,
       "step": 228
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.29e-05,
-      "loss": 7.2252,
+      "learning_rate": 6.87e-05,
+      "loss": 7.0981,
       "step": 229
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.3000000000000003e-05,
-      "loss": 7.2224,
+      "learning_rate": 6.9e-05,
+      "loss": 7.0846,
       "step": 230
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.3100000000000002e-05,
-      "loss": 7.0719,
+      "learning_rate": 6.93e-05,
+      "loss": 6.9405,
       "step": 231
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.32e-05,
-      "loss": 7.246,
+      "learning_rate": 6.96e-05,
+      "loss": 7.1693,
       "step": 232
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.3300000000000004e-05,
-      "loss": 7.1342,
+      "learning_rate": 6.989999999999999e-05,
+      "loss": 7.007,
       "step": 233
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.3400000000000003e-05,
-      "loss": 7.1486,
+      "learning_rate": 7.02e-05,
+      "loss": 7.0718,
       "step": 234
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.35e-05,
-      "loss": 7.2206,
+      "learning_rate": 7.049999999999999e-05,
+      "loss": 7.1093,
       "step": 235
     },
     {
       "epoch": 0.26,
-      "learning_rate": 2.36e-05,
-      "loss": 7.0119,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 6.8923,
       "step": 236
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.37e-05,
-      "loss": 7.0571,
+      "learning_rate": 7.11e-05,
+      "loss": 6.9165,
       "step": 237
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.38e-05,
-      "loss": 7.2578,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 7.1478,
       "step": 238
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.39e-05,
-      "loss": 7.0406,
+      "learning_rate": 7.17e-05,
+      "loss": 6.9071,
       "step": 239
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.4e-05,
-      "loss": 7.1021,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 6.9748,
       "step": 240
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.41e-05,
-      "loss": 7.0848,
+      "learning_rate": 7.23e-05,
+      "loss": 6.9684,
       "step": 241
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.4200000000000002e-05,
-      "loss": 7.1223,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 7.0586,
       "step": 242
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.43e-05,
-      "loss": 7.0448,
+      "learning_rate": 7.29e-05,
+      "loss": 6.993,
       "step": 243
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.44e-05,
-      "loss": 6.9167,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 6.866,
       "step": 244
     },
     {
       "epoch": 0.27,
-      "learning_rate": 2.45e-05,
-      "loss": 7.0238,
+      "learning_rate": 7.35e-05,
+      "loss": 6.9938,
       "step": 245
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.46e-05,
-      "loss": 6.9453,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 6.9191,
       "step": 246
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.47e-05,
-      "loss": 7.0885,
+      "learning_rate": 7.41e-05,
+      "loss": 7.0152,
       "step": 247
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.48e-05,
-      "loss": 7.0874,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 7.0077,
       "step": 248
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.4900000000000002e-05,
-      "loss": 6.8521,
+      "learning_rate": 7.47e-05,
+      "loss": 6.7886,
       "step": 249
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.5e-05,
-      "loss": 6.6483,
+      "learning_rate": 7.5e-05,
+      "loss": 6.5678,
       "step": 250
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.51e-05,
-      "loss": 7.4374,
+      "learning_rate": 7.529999999999999e-05,
+      "loss": 7.5614,
       "step": 251
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.5200000000000003e-05,
-      "loss": 7.3796,
+      "learning_rate": 7.56e-05,
+      "loss": 7.3851,
       "step": 252
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.5300000000000002e-05,
-      "loss": 7.2914,
+      "learning_rate": 7.589999999999999e-05,
+      "loss": 7.1678,
       "step": 253
     },
     {
       "epoch": 0.28,
-      "learning_rate": 2.54e-05,
-      "loss": 7.3444,
+      "learning_rate": 7.62e-05,
+      "loss": 7.1867,
       "step": 254
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.5500000000000003e-05,
-      "loss": 7.0875,
+      "learning_rate": 7.649999999999999e-05,
+      "loss": 6.9445,
       "step": 255
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.5600000000000002e-05,
-      "loss": 7.2058,
+      "learning_rate": 7.68e-05,
+      "loss": 7.1171,
       "step": 256
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.57e-05,
-      "loss": 7.1327,
+      "learning_rate": 7.709999999999999e-05,
+      "loss": 7.0816,
       "step": 257
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.58e-05,
-      "loss": 7.1093,
+      "learning_rate": 7.74e-05,
+      "loss": 7.0621,
       "step": 258
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.5900000000000003e-05,
-      "loss": 7.0382,
+      "learning_rate": 7.769999999999999e-05,
+      "loss": 6.9891,
       "step": 259
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.6000000000000002e-05,
-      "loss": 7.1595,
+      "learning_rate": 7.8e-05,
+      "loss": 7.1133,
       "step": 260
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.61e-05,
-      "loss": 7.0831,
+      "learning_rate": 7.829999999999999e-05,
+      "loss": 7.0167,
       "step": 261
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.6200000000000003e-05,
-      "loss": 6.9731,
+      "learning_rate": 7.86e-05,
+      "loss": 6.9291,
       "step": 262
     },
     {
       "epoch": 0.29,
-      "learning_rate": 2.6300000000000002e-05,
-      "loss": 7.0958,
+      "learning_rate": 7.89e-05,
+      "loss": 7.0518,
       "step": 263
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.64e-05,
-      "loss": 7.097,
+      "learning_rate": 7.92e-05,
+      "loss": 7.0504,
       "step": 264
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.6500000000000004e-05,
-      "loss": 7.027,
+      "learning_rate": 7.95e-05,
+      "loss": 6.9742,
       "step": 265
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.6600000000000003e-05,
-      "loss": 7.1614,
+      "learning_rate": 7.98e-05,
+      "loss": 7.1124,
       "step": 266
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.6700000000000002e-05,
-      "loss": 7.1645,
+      "learning_rate": 8.01e-05,
+      "loss": 7.1312,
       "step": 267
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.6800000000000004e-05,
-      "loss": 7.137,
+      "learning_rate": 8.04e-05,
+      "loss": 7.0961,
       "step": 268
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.6900000000000003e-05,
-      "loss": 7.0047,
+      "learning_rate": 8.07e-05,
+      "loss": 6.9588,
       "step": 269
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.7000000000000002e-05,
-      "loss": 7.0709,
+      "learning_rate": 8.1e-05,
+      "loss": 7.046,
       "step": 270
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.7100000000000005e-05,
-      "loss": 6.9883,
+      "learning_rate": 8.13e-05,
+      "loss": 6.938,
       "step": 271
     },
     {
       "epoch": 0.3,
-      "learning_rate": 2.7200000000000004e-05,
-      "loss": 6.9964,
+      "learning_rate": 8.16e-05,
+      "loss": 6.9457,
       "step": 272
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.7300000000000003e-05,
-      "loss": 6.9836,
+      "learning_rate": 8.19e-05,
+      "loss": 6.9595,
       "step": 273
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.7400000000000002e-05,
-      "loss": 6.893,
+      "learning_rate": 8.22e-05,
+      "loss": 6.8417,
       "step": 274
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.7500000000000004e-05,
-      "loss": 7.1359,
+      "learning_rate": 8.25e-05,
+      "loss": 7.1162,
       "step": 275
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.7600000000000003e-05,
-      "loss": 6.9874,
+      "learning_rate": 8.28e-05,
+      "loss": 6.933,
       "step": 276
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.7700000000000002e-05,
-      "loss": 7.0387,
+      "learning_rate": 8.31e-05,
+      "loss": 7.0248,
       "step": 277
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.7800000000000005e-05,
-      "loss": 6.9855,
+      "learning_rate": 8.34e-05,
+      "loss": 6.9385,
       "step": 278
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.7900000000000004e-05,
-      "loss": 7.0644,
+      "learning_rate": 8.37e-05,
+      "loss": 7.0425,
       "step": 279
     },
     {
       "epoch": 0.31,
-      "learning_rate": 2.8000000000000003e-05,
-      "loss": 7.1115,
+      "learning_rate": 8.4e-05,
+      "loss": 7.0918,
       "step": 280
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.8100000000000005e-05,
-      "loss": 6.8919,
+      "learning_rate": 8.43e-05,
+      "loss": 6.8816,
       "step": 281
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.8199999999999998e-05,
-      "loss": 7.0904,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 7.0563,
       "step": 282
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.83e-05,
-      "loss": 6.8346,
+      "learning_rate": 8.489999999999999e-05,
+      "loss": 6.8169,
       "step": 283
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.84e-05,
-      "loss": 7.0948,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 7.0907,
       "step": 284
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.8499999999999998e-05,
-      "loss": 7.0456,
+      "learning_rate": 8.549999999999999e-05,
+      "loss": 7.0363,
       "step": 285
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.86e-05,
-      "loss": 6.8849,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 6.8566,
       "step": 286
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.87e-05,
-      "loss": 7.0776,
+      "learning_rate": 8.609999999999999e-05,
+      "loss": 7.0475,
       "step": 287
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.88e-05,
-      "loss": 6.9919,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 6.9665,
       "step": 288
     },
     {
       "epoch": 0.32,
-      "learning_rate": 2.8899999999999998e-05,
-      "loss": 7.1449,
+      "learning_rate": 8.669999999999998e-05,
+      "loss": 7.1427,
       "step": 289
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.9e-05,
-      "loss": 7.1869,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 7.1855,
       "step": 290
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.91e-05,
-      "loss": 7.0797,
+      "learning_rate": 8.729999999999998e-05,
+      "loss": 7.0805,
       "step": 291
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.9199999999999998e-05,
-      "loss": 6.9812,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 6.969,
       "step": 292
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.93e-05,
-      "loss": 6.9154,
+      "learning_rate": 8.789999999999998e-05,
+      "loss": 6.9259,
       "step": 293
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.94e-05,
-      "loss": 6.9036,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 6.9057,
       "step": 294
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.95e-05,
-      "loss": 6.8639,
+      "learning_rate": 8.849999999999998e-05,
+      "loss": 6.8287,
       "step": 295
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.96e-05,
-      "loss": 6.907,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 6.8847,
       "step": 296
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.97e-05,
-      "loss": 6.8725,
+      "learning_rate": 8.909999999999998e-05,
+      "loss": 6.861,
       "step": 297
     },
     {
       "epoch": 0.33,
-      "learning_rate": 2.98e-05,
-      "loss": 6.6699,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 6.6287,
       "step": 298
     },
     {
       "epoch": 0.34,
-      "learning_rate": 2.9900000000000002e-05,
-      "loss": 6.7538,
+      "learning_rate": 8.969999999999998e-05,
+      "loss": 6.7466,
       "step": 299
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3e-05,
-      "loss": 6.4545,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 6.4532,
       "step": 300
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.01e-05,
-      "loss": 7.1777,
+      "learning_rate": 9.029999999999999e-05,
+      "loss": 7.1479,
       "step": 301
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.02e-05,
-      "loss": 7.1952,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 7.1655,
       "step": 302
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.03e-05,
-      "loss": 7.1275,
+      "learning_rate": 9.089999999999999e-05,
+      "loss": 7.0742,
       "step": 303
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.04e-05,
-      "loss": 7.1978,
+      "learning_rate": 9.12e-05,
+      "loss": 7.1582,
       "step": 304
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.05e-05,
-      "loss": 6.9796,
+      "learning_rate": 9.149999999999999e-05,
+      "loss": 6.9212,
       "step": 305
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.06e-05,
-      "loss": 7.0552,
+      "learning_rate": 9.18e-05,
+      "loss": 7.0174,
       "step": 306
     },
     {
       "epoch": 0.34,
-      "learning_rate": 3.07e-05,
-      "loss": 7.0538,
+      "learning_rate": 9.209999999999999e-05,
+      "loss": 7.0374,
       "step": 307
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.08e-05,
-      "loss": 7.0572,
+      "learning_rate": 9.24e-05,
+      "loss": 7.0562,
       "step": 308
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.09e-05,
-      "loss": 6.9961,
+      "learning_rate": 9.269999999999999e-05,
+      "loss": 6.9826,
       "step": 309
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.1e-05,
-      "loss": 7.0675,
+      "learning_rate": 9.3e-05,
+      "loss": 7.0855,
       "step": 310
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.1100000000000004e-05,
-      "loss": 7.1561,
+      "learning_rate": 9.329999999999999e-05,
+      "loss": 7.1843,
       "step": 311
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.12e-05,
-      "loss": 7.0579,
+      "learning_rate": 9.36e-05,
+      "loss": 7.0568,
       "step": 312
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.13e-05,
-      "loss": 7.0917,
+      "learning_rate": 9.389999999999999e-05,
+      "loss": 7.1016,
       "step": 313
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.1400000000000004e-05,
-      "loss": 7.0052,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 7.0264,
       "step": 314
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.15e-05,
-      "loss": 6.916,
+      "learning_rate": 9.449999999999999e-05,
+      "loss": 6.9181,
       "step": 315
     },
     {
       "epoch": 0.35,
-      "learning_rate": 3.16e-05,
-      "loss": 7.1408,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 7.16,
       "step": 316
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.1700000000000005e-05,
-      "loss": 6.9869,
+      "learning_rate": 9.51e-05,
+      "loss": 6.9908,
       "step": 317
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.18e-05,
-      "loss": 6.9017,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 6.9221,
       "step": 318
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.19e-05,
-      "loss": 6.8957,
+      "learning_rate": 9.57e-05,
+      "loss": 6.9029,
       "step": 319
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2000000000000005e-05,
-      "loss": 7.0838,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 7.0939,
       "step": 320
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.21e-05,
-      "loss": 7.006,
+      "learning_rate": 9.63e-05,
+      "loss": 7.0058,
       "step": 321
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2200000000000003e-05,
-      "loss": 6.9025,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 6.9112,
       "step": 322
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2300000000000006e-05,
-      "loss": 6.9927,
+      "learning_rate": 9.69e-05,
+      "loss": 7.0021,
       "step": 323
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.24e-05,
-      "loss": 7.0682,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 7.0628,
       "step": 324
     },
     {
       "epoch": 0.36,
-      "learning_rate": 3.2500000000000004e-05,
-      "loss": 7.0397,
+      "learning_rate": 9.75e-05,
+      "loss": 7.0523,
       "step": 325
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.26e-05,
-      "loss": 7.0221,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 7.0324,
       "step": 326
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.27e-05,
-      "loss": 6.9221,
+      "learning_rate": 9.81e-05,
+      "loss": 6.9263,
       "step": 327
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.2800000000000004e-05,
-      "loss": 7.0354,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 7.0431,
       "step": 328
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.29e-05,
-      "loss": 7.097,
+      "learning_rate": 9.87e-05,
+      "loss": 7.1019,
       "step": 329
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.3e-05,
-      "loss": 7.0337,
+      "learning_rate": 9.9e-05,
+      "loss": 7.0567,
       "step": 330
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.3100000000000005e-05,
-      "loss": 7.0913,
+      "learning_rate": 9.93e-05,
+      "loss": 7.1136,
       "step": 331
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.32e-05,
-      "loss": 6.7985,
+      "learning_rate": 9.96e-05,
+      "loss": 6.8044,
       "step": 332
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.33e-05,
-      "loss": 6.9769,
+      "learning_rate": 9.99e-05,
+      "loss": 6.9848,
       "step": 333
     },
     {
       "epoch": 0.37,
-      "learning_rate": 3.3400000000000005e-05,
-      "loss": 7.0259,
+      "learning_rate": 0.0001002,
+      "loss": 7.0483,
       "step": 334
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.35e-05,
-      "loss": 7.0789,
+      "learning_rate": 0.0001005,
+      "loss": 7.0885,
       "step": 335
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.3600000000000004e-05,
-      "loss": 6.8872,
+      "learning_rate": 0.0001008,
+      "loss": 6.9142,
       "step": 336
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.3700000000000006e-05,
-      "loss": 6.876,
+      "learning_rate": 0.0001011,
+      "loss": 6.8915,
       "step": 337
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.38e-05,
-      "loss": 7.0223,
+      "learning_rate": 0.0001014,
+      "loss": 7.0468,
       "step": 338
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.3900000000000004e-05,
-      "loss": 7.0551,
+      "learning_rate": 0.00010169999999999999,
+      "loss": 7.0937,
       "step": 339
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.4000000000000007e-05,
-      "loss": 6.9647,
+      "learning_rate": 0.000102,
+      "loss": 6.9816,
       "step": 340
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.41e-05,
-      "loss": 7.0563,
+      "learning_rate": 0.00010229999999999999,
+      "loss": 7.0624,
       "step": 341
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.4200000000000005e-05,
-      "loss": 6.9318,
+      "learning_rate": 0.0001026,
+      "loss": 6.9742,
       "step": 342
     },
     {
       "epoch": 0.38,
-      "learning_rate": 3.430000000000001e-05,
-      "loss": 6.9198,
+      "learning_rate": 0.0001029,
+      "loss": 6.9457,
       "step": 343
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.4399999999999996e-05,
-      "loss": 6.8966,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 6.9374,
       "step": 344
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.45e-05,
-      "loss": 6.7652,
+      "learning_rate": 0.00010349999999999998,
+      "loss": 6.806,
       "step": 345
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.46e-05,
-      "loss": 6.8923,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 6.921,
       "step": 346
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.4699999999999996e-05,
-      "loss": 6.8213,
+      "learning_rate": 0.00010409999999999998,
+      "loss": 6.8681,
       "step": 347
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.48e-05,
-      "loss": 6.8801,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 6.911,
       "step": 348
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.49e-05,
-      "loss": 6.7339,
+      "learning_rate": 0.00010469999999999998,
+      "loss": 6.7701,
       "step": 349
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.5e-05,
-      "loss": 6.5925,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 6.6044,
       "step": 350
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.51e-05,
-      "loss": 6.9905,
+      "learning_rate": 0.00010529999999999998,
+      "loss": 7.0543,
       "step": 351
     },
     {
       "epoch": 0.39,
-      "learning_rate": 3.52e-05,
-      "loss": 7.0017,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 7.1145,
       "step": 352
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.53e-05,
-      "loss": 6.9757,
+      "learning_rate": 0.00010589999999999999,
+      "loss": 7.0522,
       "step": 353
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.54e-05,
-      "loss": 6.9631,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 7.0546,
       "step": 354
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.55e-05,
-      "loss": 7.0725,
+      "learning_rate": 0.00010649999999999999,
+      "loss": 7.1509,
       "step": 355
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.56e-05,
-      "loss": 6.9315,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 7.0003,
       "step": 356
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.57e-05,
-      "loss": 7.1701,
+      "learning_rate": 0.00010709999999999999,
+      "loss": 7.2155,
       "step": 357
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.58e-05,
-      "loss": 6.9165,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 6.9667,
       "step": 358
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.59e-05,
-      "loss": 7.0283,
+      "learning_rate": 0.00010769999999999999,
+      "loss": 7.0813,
       "step": 359
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.6e-05,
-      "loss": 6.811,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 6.8606,
       "step": 360
     },
     {
       "epoch": 0.4,
-      "learning_rate": 3.61e-05,
-      "loss": 6.9093,
+      "learning_rate": 0.00010829999999999999,
+      "loss": 6.9562,
       "step": 361
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.62e-05,
-      "loss": 6.9873,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 7.0227,
       "step": 362
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.63e-05,
-      "loss": 6.9259,
+      "learning_rate": 0.00010889999999999999,
+      "loss": 6.9763,
       "step": 363
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.6400000000000004e-05,
-      "loss": 6.9824,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 7.0286,
       "step": 364
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.65e-05,
-      "loss": 6.9479,
+      "learning_rate": 0.00010949999999999999,
+      "loss": 7.021,
       "step": 365
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.66e-05,
-      "loss": 6.8664,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 6.9146,
       "step": 366
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.6700000000000004e-05,
-      "loss": 6.9216,
+      "learning_rate": 0.00011009999999999999,
+      "loss": 6.9761,
       "step": 367
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.68e-05,
-      "loss": 6.9015,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 6.9522,
       "step": 368
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.69e-05,
-      "loss": 6.8386,
+      "learning_rate": 0.0001107,
+      "loss": 6.8934,
       "step": 369
     },
     {
       "epoch": 0.41,
-      "learning_rate": 3.7e-05,
-      "loss": 7.0073,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 7.0491,
       "step": 370
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.71e-05,
-      "loss": 6.932,
+      "learning_rate": 0.0001113,
+      "loss": 6.9931,
       "step": 371
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.72e-05,
-      "loss": 7.094,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 7.1533,
       "step": 372
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.73e-05,
-      "loss": 6.9379,
+      "learning_rate": 0.0001119,
+      "loss": 6.9872,
       "step": 373
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.74e-05,
-      "loss": 7.0351,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 7.0959,
       "step": 374
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.7500000000000003e-05,
-      "loss": 6.8194,
+      "learning_rate": 0.0001125,
+      "loss": 6.8676,
       "step": 375
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.76e-05,
-      "loss": 6.9407,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 6.9895,
       "step": 376
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.77e-05,
-      "loss": 6.8181,
+      "learning_rate": 0.00011309999999999998,
+      "loss": 6.8687,
       "step": 377
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.7800000000000004e-05,
-      "loss": 6.9171,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 6.9762,
       "step": 378
     },
     {
       "epoch": 0.42,
-      "learning_rate": 3.79e-05,
-      "loss": 6.9143,
+      "learning_rate": 0.00011369999999999999,
+      "loss": 6.9781,
       "step": 379
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.8e-05,
-      "loss": 6.9766,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 7.0334,
       "step": 380
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.8100000000000005e-05,
-      "loss": 7.0032,
+      "learning_rate": 0.00011429999999999999,
+      "loss": 7.0634,
       "step": 381
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.82e-05,
-      "loss": 6.8646,
+      "learning_rate": 0.0001146,
+      "loss": 6.9348,
       "step": 382
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.83e-05,
-      "loss": 7.0346,
+      "learning_rate": 0.00011489999999999999,
+      "loss": 7.0884,
       "step": 383
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.8400000000000005e-05,
-      "loss": 6.9878,
+      "learning_rate": 0.0001152,
+      "loss": 7.0428,
       "step": 384
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.85e-05,
-      "loss": 6.8673,
+      "learning_rate": 0.00011549999999999999,
+      "loss": 6.9321,
       "step": 385
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.86e-05,
-      "loss": 6.9269,
+      "learning_rate": 0.0001158,
+      "loss": 6.9785,
       "step": 386
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.8700000000000006e-05,
-      "loss": 6.902,
+      "learning_rate": 0.00011609999999999999,
+      "loss": 6.9532,
       "step": 387
     },
     {
       "epoch": 0.43,
-      "learning_rate": 3.88e-05,
-      "loss": 6.9011,
+      "learning_rate": 0.0001164,
+      "loss": 6.955,
       "step": 388
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.8900000000000004e-05,
-      "loss": 6.8772,
+      "learning_rate": 0.00011669999999999999,
+      "loss": 6.9162,
       "step": 389
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.9000000000000006e-05,
-      "loss": 7.1375,
+      "learning_rate": 0.000117,
+      "loss": 7.2093,
       "step": 390
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.91e-05,
-      "loss": 6.9559,
+      "learning_rate": 0.00011729999999999999,
+      "loss": 7.0259,
       "step": 391
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.9200000000000004e-05,
-      "loss": 6.9275,
+      "learning_rate": 0.0001176,
+      "loss": 7.0094,
       "step": 392
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.9300000000000007e-05,
-      "loss": 6.8662,
+      "learning_rate": 0.00011789999999999999,
+      "loss": 6.9411,
       "step": 393
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.94e-05,
-      "loss": 7.0261,
+      "learning_rate": 0.0001182,
+      "loss": 7.1313,
       "step": 394
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.9500000000000005e-05,
-      "loss": 6.8534,
+      "learning_rate": 0.0001185,
+      "loss": 6.9356,
       "step": 395
     },
     {
       "epoch": 0.44,
-      "learning_rate": 3.960000000000001e-05,
-      "loss": 6.6708,
+      "learning_rate": 0.0001188,
+      "loss": 6.7395,
       "step": 396
     },
     {
       "epoch": 0.45,
-      "learning_rate": 3.97e-05,
-      "loss": 6.6056,
+      "learning_rate": 0.0001191,
+      "loss": 6.6755,
       "step": 397
     },
     {
       "epoch": 0.45,
-      "learning_rate": 3.9800000000000005e-05,
-      "loss": 6.6856,
+      "learning_rate": 0.0001194,
+      "loss": 6.8141,
       "step": 398
     },
     {
       "epoch": 0.45,
-      "learning_rate": 3.99e-05,
-      "loss": 6.6275,
+      "learning_rate": 0.0001197,
+      "loss": 6.7271,
       "step": 399
     },
     {
       "epoch": 0.45,
-      "learning_rate": 4e-05,
-      "loss": 6.4524,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 6.484,
       "step": 400
     },
     {
       "epoch": 0.45,
-      "learning_rate": 4.0100000000000006e-05,
-      "loss": 7.2072,
+      "learning_rate": 0.0001203,
+      "loss": 7.2112,
       "step": 401
     },
     {
       "epoch": 0.45,
-      "learning_rate": 4.02e-05,
-      "loss": 7.2003,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 7.1384,
       "step": 402
     },
     {
       "epoch": 0.45,
-      "learning_rate": 4.0300000000000004e-05,
-      "loss": 7.1376,
+      "learning_rate": 0.0001209,
+      "loss": 7.1291,
       "step": 403
     },
     {
       "epoch": 0.45,
-      "learning_rate": 4.0400000000000006e-05,
-      "loss": 7.0276,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 7.0676,
       "step": 404
     },
     {
       "epoch": 0.45,
-      "learning_rate": 4.05e-05,
-      "loss": 7.1858,
+      "learning_rate": 0.0001215,
+      "loss": 7.2225,
       "step": 405
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.0600000000000004e-05,
-      "loss": 6.9825,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 7.0608,
       "step": 406
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.07e-05,
-      "loss": 6.9807,
+      "learning_rate": 0.00012209999999999999,
+      "loss": 7.0342,
       "step": 407
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.08e-05,
-      "loss": 6.8985,
+      "learning_rate": 0.0001224,
+      "loss": 6.987,
       "step": 408
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.09e-05,
-      "loss": 7.1197,
+      "learning_rate": 0.00012269999999999997,
+      "loss": 7.2347,
       "step": 409
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.1e-05,
-      "loss": 6.9975,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 7.0631,
       "step": 410
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.11e-05,
-      "loss": 6.9675,
+      "learning_rate": 0.0001233,
+      "loss": 7.0332,
       "step": 411
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.12e-05,
-      "loss": 6.9361,
+      "learning_rate": 0.0001236,
+      "loss": 7.0126,
       "step": 412
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.13e-05,
-      "loss": 6.9257,
+      "learning_rate": 0.00012389999999999998,
+      "loss": 7.0139,
       "step": 413
     },
     {
       "epoch": 0.46,
-      "learning_rate": 4.14e-05,
-      "loss": 6.8091,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 6.8703,
       "step": 414
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.15e-05,
-      "loss": 6.879,
+      "learning_rate": 0.0001245,
+      "loss": 6.9452,
       "step": 415
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.16e-05,
-      "loss": 6.9704,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 7.0457,
       "step": 416
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.17e-05,
-      "loss": 6.8838,
+      "learning_rate": 0.00012509999999999998,
+      "loss": 6.9499,
       "step": 417
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.18e-05,
-      "loss": 6.972,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 7.0225,
       "step": 418
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.19e-05,
-      "loss": 6.9315,
+      "learning_rate": 0.0001257,
+      "loss": 7.0287,
       "step": 419
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.2e-05,
-      "loss": 6.9479,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 7.0214,
       "step": 420
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.21e-05,
-      "loss": 6.9612,
+      "learning_rate": 0.00012629999999999998,
+      "loss": 7.0265,
       "step": 421
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.22e-05,
-      "loss": 7.0222,
+      "learning_rate": 0.0001266,
+      "loss": 7.0998,
       "step": 422
     },
     {
       "epoch": 0.47,
-      "learning_rate": 4.23e-05,
-      "loss": 6.9353,
+      "learning_rate": 0.0001269,
+      "loss": 7.0013,
       "step": 423
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.24e-05,
-      "loss": 6.9481,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 7.0043,
       "step": 424
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.25e-05,
-      "loss": 6.9778,
+      "learning_rate": 0.00012749999999999998,
+      "loss": 7.0486,
       "step": 425
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.26e-05,
-      "loss": 7.0824,
+      "learning_rate": 0.0001278,
+      "loss": 7.1821,
       "step": 426
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.27e-05,
-      "loss": 6.8798,
+      "learning_rate": 0.0001281,
+      "loss": 6.9276,
       "step": 427
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.2800000000000004e-05,
-      "loss": 7.0481,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 7.1228,
       "step": 428
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.29e-05,
-      "loss": 6.9147,
+      "learning_rate": 0.00012869999999999998,
+      "loss": 6.9813,
       "step": 429
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.3e-05,
-      "loss": 6.9341,
+      "learning_rate": 0.000129,
+      "loss": 7.0056,
       "step": 430
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.3100000000000004e-05,
-      "loss": 7.0707,
+      "learning_rate": 0.0001293,
+      "loss": 7.1426,
       "step": 431
     },
     {
       "epoch": 0.48,
-      "learning_rate": 4.32e-05,
-      "loss": 7.0151,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 7.0902,
       "step": 432
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.33e-05,
-      "loss": 6.6748,
+      "learning_rate": 0.00012989999999999999,
+      "loss": 6.7767,
       "step": 433
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.3400000000000005e-05,
-      "loss": 6.9164,
+      "learning_rate": 0.0001302,
+      "loss": 6.9957,
       "step": 434
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.35e-05,
-      "loss": 6.8328,
+      "learning_rate": 0.0001305,
+      "loss": 6.9224,
       "step": 435
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.36e-05,
-      "loss": 7.0409,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 7.1117,
       "step": 436
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.3700000000000005e-05,
-      "loss": 6.9415,
+      "learning_rate": 0.0001311,
+      "loss": 7.0286,
       "step": 437
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.38e-05,
-      "loss": 6.9958,
+      "learning_rate": 0.0001314,
+      "loss": 7.0744,
       "step": 438
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.39e-05,
-      "loss": 6.7835,
+      "learning_rate": 0.00013169999999999998,
+      "loss": 6.8978,
       "step": 439
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.4000000000000006e-05,
-      "loss": 6.9306,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 7.0032,
       "step": 440
     },
     {
       "epoch": 0.49,
-      "learning_rate": 4.41e-05,
-      "loss": 6.9877,
+      "learning_rate": 0.0001323,
+      "loss": 7.1087,
       "step": 441
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.4200000000000004e-05,
-      "loss": 6.8843,
+      "learning_rate": 0.0001326,
+      "loss": 6.979,
       "step": 442
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.43e-05,
-      "loss": 6.8446,
+      "learning_rate": 0.00013289999999999998,
+      "loss": 6.9107,
       "step": 443
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.44e-05,
-      "loss": 7.0288,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 7.0875,
       "step": 444
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.4500000000000004e-05,
-      "loss": 6.7124,
+      "learning_rate": 0.0001335,
+      "loss": 6.7743,
       "step": 445
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.46e-05,
-      "loss": 6.7445,
+      "learning_rate": 0.0001338,
+      "loss": 6.8467,
       "step": 446
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.47e-05,
-      "loss": 6.7771,
+      "learning_rate": 0.00013409999999999998,
+      "loss": 6.8433,
       "step": 447
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.4800000000000005e-05,
-      "loss": 6.741,
+      "learning_rate": 0.0001344,
+      "loss": 6.8407,
       "step": 448
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.49e-05,
-      "loss": 6.4875,
+      "learning_rate": 0.0001347,
+      "loss": 6.6201,
       "step": 449
     },
     {
       "epoch": 0.5,
-      "learning_rate": 4.5e-05,
-      "loss": 6.4227,
+      "learning_rate": 0.000135,
+      "loss": 6.5578,
       "step": 450
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.5100000000000005e-05,
-      "loss": 6.9795,
+      "learning_rate": 0.00013529999999999998,
+      "loss": 7.0756,
       "step": 451
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.52e-05,
-      "loss": 7.0996,
+      "learning_rate": 0.0001356,
+      "loss": 7.1535,
       "step": 452
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.53e-05,
-      "loss": 7.1341,
+      "learning_rate": 0.0001359,
+      "loss": 7.1773,
       "step": 453
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.5400000000000006e-05,
-      "loss": 6.9488,
+      "learning_rate": 0.0001362,
+      "loss": 7.0272,
       "step": 454
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.55e-05,
-      "loss": 7.0556,
+      "learning_rate": 0.00013649999999999998,
+      "loss": 7.1369,
       "step": 455
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.5600000000000004e-05,
-      "loss": 7.0871,
+      "learning_rate": 0.0001368,
+      "loss": 7.1887,
       "step": 456
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.5700000000000006e-05,
-      "loss": 6.8609,
+      "learning_rate": 0.0001371,
+      "loss": 6.9843,
       "step": 457
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.58e-05,
-      "loss": 6.9567,
+      "learning_rate": 0.0001374,
+      "loss": 7.0757,
       "step": 458
     },
     {
       "epoch": 0.51,
-      "learning_rate": 4.5900000000000004e-05,
-      "loss": 6.933,
+      "learning_rate": 0.00013769999999999999,
+      "loss": 7.0141,
       "step": 459
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.600000000000001e-05,
-      "loss": 7.0277,
+      "learning_rate": 0.000138,
+      "loss": 7.1129,
       "step": 460
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.61e-05,
-      "loss": 6.852,
+      "learning_rate": 0.0001383,
+      "loss": 6.9482,
       "step": 461
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.6200000000000005e-05,
-      "loss": 6.997,
+      "learning_rate": 0.0001386,
+      "loss": 7.0756,
       "step": 462
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.630000000000001e-05,
-      "loss": 6.8578,
+      "learning_rate": 0.0001389,
+      "loss": 6.9589,
       "step": 463
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.64e-05,
-      "loss": 6.836,
+      "learning_rate": 0.0001392,
+      "loss": 6.9225,
       "step": 464
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.6500000000000005e-05,
-      "loss": 6.9535,
+      "learning_rate": 0.0001395,
+      "loss": 7.0474,
       "step": 465
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.660000000000001e-05,
-      "loss": 6.9268,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 7.0023,
       "step": 466
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.6700000000000003e-05,
-      "loss": 6.8603,
+      "learning_rate": 0.0001401,
+      "loss": 6.9858,
       "step": 467
     },
     {
       "epoch": 0.52,
-      "learning_rate": 4.6800000000000006e-05,
-      "loss": 6.9719,
+      "learning_rate": 0.0001404,
+      "loss": 7.0471,
       "step": 468
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.69e-05,
-      "loss": 6.8866,
+      "learning_rate": 0.00014069999999999998,
+      "loss": 6.9687,
       "step": 469
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.7e-05,
-      "loss": 6.9621,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 7.0465,
       "step": 470
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.71e-05,
-      "loss": 6.8627,
+      "learning_rate": 0.0001413,
+      "loss": 6.966,
       "step": 471
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.72e-05,
-      "loss": 6.8953,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 6.9786,
       "step": 472
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.73e-05,
-      "loss": 6.9193,
+      "learning_rate": 0.00014189999999999998,
+      "loss": 6.9801,
       "step": 473
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.74e-05,
-      "loss": 6.8248,
+      "learning_rate": 0.0001422,
+      "loss": 6.909,
       "step": 474
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.75e-05,
-      "loss": 6.9785,
+      "learning_rate": 0.0001425,
+      "loss": 7.0728,
       "step": 475
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.76e-05,
-      "loss": 6.9762,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 7.0654,
       "step": 476
     },
     {
       "epoch": 0.53,
-      "learning_rate": 4.77e-05,
-      "loss": 7.073,
+      "learning_rate": 0.00014309999999999998,
+      "loss": 7.181,
       "step": 477
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.78e-05,
-      "loss": 6.809,
+      "learning_rate": 0.0001434,
+      "loss": 6.9118,
       "step": 478
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.79e-05,
-      "loss": 6.8333,
+      "learning_rate": 0.00014369999999999997,
+      "loss": 6.9076,
       "step": 479
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.8e-05,
-      "loss": 6.8699,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 6.9658,
       "step": 480
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.8100000000000004e-05,
-      "loss": 6.8412,
+      "learning_rate": 0.00014429999999999998,
+      "loss": 6.9397,
       "step": 481
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.82e-05,
-      "loss": 6.8619,
+      "learning_rate": 0.0001446,
+      "loss": 6.9348,
       "step": 482
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.83e-05,
-      "loss": 6.8868,
+      "learning_rate": 0.00014489999999999997,
+      "loss": 6.9841,
       "step": 483
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.8400000000000004e-05,
-      "loss": 6.9498,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 7.0631,
       "step": 484
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.85e-05,
-      "loss": 6.9514,
+      "learning_rate": 0.00014549999999999999,
+      "loss": 7.0487,
       "step": 485
     },
     {
       "epoch": 0.54,
-      "learning_rate": 4.86e-05,
-      "loss": 6.8857,
+      "learning_rate": 0.0001458,
+      "loss": 6.9595,
       "step": 486
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.87e-05,
-      "loss": 6.9468,
+      "learning_rate": 0.00014609999999999997,
+      "loss": 7.0519,
       "step": 487
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.88e-05,
-      "loss": 6.8264,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 6.901,
       "step": 488
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.89e-05,
-      "loss": 7.0155,
+      "learning_rate": 0.0001467,
+      "loss": 7.1449,
       "step": 489
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.9e-05,
-      "loss": 7.0414,
+      "learning_rate": 0.000147,
+      "loss": 7.1872,
       "step": 490
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.91e-05,
-      "loss": 6.846,
+      "learning_rate": 0.00014729999999999998,
+      "loss": 6.9438,
       "step": 491
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.92e-05,
-      "loss": 6.9121,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 6.9725,
       "step": 492
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.93e-05,
-      "loss": 6.8727,
+      "learning_rate": 0.0001479,
+      "loss": 6.9609,
       "step": 493
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.94e-05,
-      "loss": 6.8631,
+      "learning_rate": 0.0001482,
+      "loss": 6.9902,
       "step": 494
     },
     {
       "epoch": 0.55,
-      "learning_rate": 4.9500000000000004e-05,
-      "loss": 6.913,
+      "learning_rate": 0.00014849999999999998,
+      "loss": 7.039,
       "step": 495
     },
     {
       "epoch": 0.56,
-      "learning_rate": 4.96e-05,
-      "loss": 6.6503,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 6.7782,
       "step": 496
     },
     {
       "epoch": 0.56,
-      "learning_rate": 4.97e-05,
-      "loss": 6.3894,
+      "learning_rate": 0.0001491,
+      "loss": 6.6103,
       "step": 497
     },
     {
       "epoch": 0.56,
-      "learning_rate": 4.9800000000000004e-05,
-      "loss": 6.9782,
+      "learning_rate": 0.0001494,
+      "loss": 7.1955,
       "step": 498
     },
     {
       "epoch": 0.56,
-      "learning_rate": 4.99e-05,
-      "loss": 6.7164,
+      "learning_rate": 0.00014969999999999998,
+      "loss": 6.8568,
       "step": 499
     },
     {
       "epoch": 0.56,
-      "learning_rate": 5e-05,
-      "loss": 6.3206,
+      "learning_rate": 0.00015,
+      "loss": 6.5145,
       "step": 500
     },
     {
       "epoch": 0.56,
-      "learning_rate": 5.0100000000000005e-05,
-      "loss": 7.389,
+      "learning_rate": 0.0001503,
+      "loss": 7.3303,
       "step": 501
     },
     {
       "epoch": 0.56,
-      "learning_rate": 5.02e-05,
-      "loss": 7.0552,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 6.9687,
       "step": 502
     },
     {
       "epoch": 0.56,
-      "learning_rate": 5.03e-05,
-      "loss": 7.199,
+      "learning_rate": 0.00015089999999999998,
+      "loss": 7.0817,
       "step": 503
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.0400000000000005e-05,
-      "loss": 7.242,
+      "learning_rate": 0.0001512,
+      "loss": 7.1751,
       "step": 504
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.05e-05,
-      "loss": 7.1975,
+      "learning_rate": 0.0001515,
+      "loss": 7.1587,
       "step": 505
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.0600000000000003e-05,
-      "loss": 7.1445,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 7.1619,
       "step": 506
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.0700000000000006e-05,
-      "loss": 7.0869,
+      "learning_rate": 0.00015209999999999998,
+      "loss": 7.1337,
       "step": 507
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.08e-05,
-      "loss": 6.9375,
+      "learning_rate": 0.0001524,
+      "loss": 7.0232,
       "step": 508
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.0900000000000004e-05,
-      "loss": 7.0081,
+      "learning_rate": 0.0001527,
+      "loss": 7.0737,
       "step": 509
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.1000000000000006e-05,
-      "loss": 6.9671,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 7.041,
       "step": 510
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.11e-05,
-      "loss": 6.9103,
+      "learning_rate": 0.00015329999999999999,
+      "loss": 6.9894,
       "step": 511
     },
     {
       "epoch": 0.57,
-      "learning_rate": 5.1200000000000004e-05,
-      "loss": 6.952,
+      "learning_rate": 0.0001536,
+      "loss": 7.0401,
       "step": 512
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.130000000000001e-05,
-      "loss": 7.0824,
+      "learning_rate": 0.0001539,
+      "loss": 7.1902,
       "step": 513
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.14e-05,
-      "loss": 7.0204,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 7.0797,
       "step": 514
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.1500000000000005e-05,
-      "loss": 7.0265,
+      "learning_rate": 0.0001545,
+      "loss": 7.1158,
       "step": 515
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.16e-05,
-      "loss": 6.8061,
+      "learning_rate": 0.0001548,
+      "loss": 6.9046,
       "step": 516
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.17e-05,
-      "loss": 7.0333,
+      "learning_rate": 0.0001551,
+      "loss": 7.1145,
       "step": 517
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.1800000000000005e-05,
-      "loss": 6.8753,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 6.9434,
       "step": 518
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.19e-05,
-      "loss": 7.0494,
+      "learning_rate": 0.0001557,
+      "loss": 7.1319,
       "step": 519
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.2000000000000004e-05,
-      "loss": 7.0285,
+      "learning_rate": 0.000156,
+      "loss": 7.1087,
       "step": 520
     },
     {
       "epoch": 0.58,
-      "learning_rate": 5.2100000000000006e-05,
-      "loss": 6.872,
+      "learning_rate": 0.0001563,
+      "loss": 6.9623,
       "step": 521
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.22e-05,
-      "loss": 6.8672,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 6.9232,
       "step": 522
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.2300000000000004e-05,
-      "loss": 6.769,
+      "learning_rate": 0.0001569,
+      "loss": 6.9014,
       "step": 523
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.2400000000000007e-05,
-      "loss": 6.8619,
+      "learning_rate": 0.0001572,
+      "loss": 6.9578,
       "step": 524
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.25e-05,
-      "loss": 7.0126,
+      "learning_rate": 0.00015749999999999998,
+      "loss": 7.0869,
       "step": 525
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.2600000000000005e-05,
-      "loss": 6.8324,
+      "learning_rate": 0.0001578,
+      "loss": 6.9331,
       "step": 526
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.270000000000001e-05,
-      "loss": 6.9212,
+      "learning_rate": 0.0001581,
+      "loss": 7.0333,
       "step": 527
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.28e-05,
-      "loss": 6.9677,
+      "learning_rate": 0.0001584,
+      "loss": 7.097,
       "step": 528
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.2900000000000005e-05,
-      "loss": 6.9959,
+      "learning_rate": 0.00015869999999999998,
+      "loss": 7.0979,
       "step": 529
     },
     {
       "epoch": 0.59,
-      "learning_rate": 5.300000000000001e-05,
-      "loss": 7.0283,
+      "learning_rate": 0.000159,
+      "loss": 7.1414,
       "step": 530
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.31e-05,
-      "loss": 6.9021,
+      "learning_rate": 0.0001593,
+      "loss": 6.9939,
       "step": 531
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.3200000000000006e-05,
-      "loss": 6.8198,
+      "learning_rate": 0.0001596,
+      "loss": 6.954,
       "step": 532
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.330000000000001e-05,
-      "loss": 6.836,
+      "learning_rate": 0.00015989999999999998,
+      "loss": 6.948,
       "step": 533
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.3400000000000004e-05,
-      "loss": 6.8677,
+      "learning_rate": 0.0001602,
+      "loss": 6.9871,
       "step": 534
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.3500000000000006e-05,
-      "loss": 6.8605,
+      "learning_rate": 0.0001605,
+      "loss": 6.9491,
       "step": 535
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.360000000000001e-05,
-      "loss": 6.9906,
+      "learning_rate": 0.0001608,
+      "loss": 7.122,
       "step": 536
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.3700000000000004e-05,
-      "loss": 6.9186,
+      "learning_rate": 0.00016109999999999999,
+      "loss": 7.0287,
       "step": 537
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.380000000000001e-05,
-      "loss": 7.0049,
+      "learning_rate": 0.0001614,
+      "loss": 7.138,
       "step": 538
     },
     {
       "epoch": 0.6,
-      "learning_rate": 5.390000000000001e-05,
-      "loss": 6.9883,
+      "learning_rate": 0.0001617,
+      "loss": 7.0906,
       "step": 539
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.4000000000000005e-05,
-      "loss": 6.9206,
+      "learning_rate": 0.000162,
+      "loss": 7.0475,
       "step": 540
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.410000000000001e-05,
-      "loss": 7.024,
+      "learning_rate": 0.0001623,
+      "loss": 7.1261,
       "step": 541
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.420000000000001e-05,
-      "loss": 6.8775,
+      "learning_rate": 0.0001626,
+      "loss": 7.002,
       "step": 542
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.4300000000000005e-05,
-      "loss": 6.7833,
+      "learning_rate": 0.0001629,
+      "loss": 6.9215,
       "step": 543
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.440000000000001e-05,
-      "loss": 6.6842,
+      "learning_rate": 0.0001632,
+      "loss": 6.8397,
       "step": 544
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.45e-05,
-      "loss": 6.7713,
+      "learning_rate": 0.0001635,
+      "loss": 6.8969,
       "step": 545
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.4600000000000006e-05,
-      "loss": 6.7058,
+      "learning_rate": 0.0001638,
+      "loss": 6.8427,
       "step": 546
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.470000000000001e-05,
-      "loss": 6.924,
+      "learning_rate": 0.0001641,
+      "loss": 7.0921,
       "step": 547
     },
     {
       "epoch": 0.61,
-      "learning_rate": 5.4800000000000004e-05,
-      "loss": 6.7637,
+      "learning_rate": 0.0001644,
+      "loss": 6.9436,
       "step": 548
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.4900000000000006e-05,
-      "loss": 6.3353,
+      "learning_rate": 0.0001647,
+      "loss": 6.5667,
       "step": 549
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 6.6002,
+      "learning_rate": 0.000165,
+      "loss": 6.8137,
       "step": 550
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.5100000000000004e-05,
-      "loss": 7.0803,
+      "learning_rate": 0.0001653,
+      "loss": 7.0627,
       "step": 551
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.520000000000001e-05,
-      "loss": 7.2943,
+      "learning_rate": 0.0001656,
+      "loss": 7.2371,
       "step": 552
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.530000000000001e-05,
-      "loss": 7.1373,
+      "learning_rate": 0.0001659,
+      "loss": 7.0944,
       "step": 553
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.5400000000000005e-05,
-      "loss": 7.1008,
+      "learning_rate": 0.0001662,
+      "loss": 7.0779,
       "step": 554
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.550000000000001e-05,
-      "loss": 7.0603,
+      "learning_rate": 0.0001665,
+      "loss": 7.0816,
       "step": 555
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.560000000000001e-05,
-      "loss": 7.0148,
+      "learning_rate": 0.0001668,
+      "loss": 7.0683,
       "step": 556
     },
     {
       "epoch": 0.62,
-      "learning_rate": 5.5700000000000005e-05,
-      "loss": 6.9658,
+      "learning_rate": 0.0001671,
+      "loss": 6.993,
       "step": 557
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.580000000000001e-05,
-      "loss": 7.0148,
+      "learning_rate": 0.0001674,
+      "loss": 7.0657,
       "step": 558
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.590000000000001e-05,
-      "loss": 6.9468,
+      "learning_rate": 0.0001677,
+      "loss": 7.0238,
       "step": 559
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.6000000000000006e-05,
-      "loss": 6.8661,
+      "learning_rate": 0.000168,
+      "loss": 6.9663,
       "step": 560
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.610000000000001e-05,
-      "loss": 6.9911,
+      "learning_rate": 0.0001683,
+      "loss": 7.0719,
       "step": 561
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.620000000000001e-05,
-      "loss": 6.8454,
+      "learning_rate": 0.0001686,
+      "loss": 6.938,
       "step": 562
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.63e-05,
-      "loss": 6.9046,
+      "learning_rate": 0.00016889999999999996,
+      "loss": 6.9823,
       "step": 563
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.6399999999999995e-05,
-      "loss": 6.914,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 7.0229,
       "step": 564
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.65e-05,
-      "loss": 7.0916,
+      "learning_rate": 0.00016949999999999997,
+      "loss": 7.1811,
       "step": 565
     },
     {
       "epoch": 0.63,
-      "learning_rate": 5.66e-05,
-      "loss": 6.7813,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 6.8668,
       "step": 566
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.6699999999999996e-05,
-      "loss": 6.9587,
+      "learning_rate": 0.00017009999999999996,
+      "loss": 7.0338,
       "step": 567
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.68e-05,
-      "loss": 6.8033,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 6.9176,
       "step": 568
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.69e-05,
-      "loss": 6.8687,
+      "learning_rate": 0.00017069999999999998,
+      "loss": 6.9411,
       "step": 569
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.6999999999999996e-05,
-      "loss": 6.998,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 7.0967,
       "step": 570
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.71e-05,
-      "loss": 6.8671,
+      "learning_rate": 0.00017129999999999996,
+      "loss": 6.971,
       "step": 571
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.72e-05,
-      "loss": 6.9932,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 7.1118,
       "step": 572
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.73e-05,
-      "loss": 6.8657,
+      "learning_rate": 0.00017189999999999998,
+      "loss": 6.9696,
       "step": 573
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.74e-05,
-      "loss": 6.7437,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 6.8589,
       "step": 574
     },
     {
       "epoch": 0.64,
-      "learning_rate": 5.7499999999999995e-05,
-      "loss": 6.9499,
+      "learning_rate": 0.00017249999999999996,
+      "loss": 7.0333,
       "step": 575
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.76e-05,
-      "loss": 6.7804,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 6.8388,
       "step": 576
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.77e-05,
-      "loss": 6.9061,
+      "learning_rate": 0.00017309999999999998,
+      "loss": 6.9863,
       "step": 577
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.7799999999999995e-05,
-      "loss": 6.9602,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 7.0773,
       "step": 578
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.79e-05,
-      "loss": 6.8117,
+      "learning_rate": 0.00017369999999999997,
+      "loss": 6.9109,
       "step": 579
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.8e-05,
-      "loss": 7.0433,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 7.1806,
       "step": 580
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.8099999999999996e-05,
-      "loss": 7.0144,
+      "learning_rate": 0.00017429999999999998,
+      "loss": 7.1043,
       "step": 581
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.82e-05,
-      "loss": 6.8911,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 7.0122,
       "step": 582
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.83e-05,
-      "loss": 6.7979,
+      "learning_rate": 0.00017489999999999997,
+      "loss": 6.9128,
       "step": 583
     },
     {
       "epoch": 0.65,
-      "learning_rate": 5.8399999999999997e-05,
-      "loss": 6.9271,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 7.0354,
       "step": 584
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.85e-05,
-      "loss": 6.9344,
+      "learning_rate": 0.00017549999999999998,
+      "loss": 7.0603,
       "step": 585
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.86e-05,
-      "loss": 6.9409,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 7.0723,
       "step": 586
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.87e-05,
-      "loss": 6.9363,
+      "learning_rate": 0.00017609999999999997,
+      "loss": 7.0581,
       "step": 587
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.88e-05,
-      "loss": 6.8768,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 7.0059,
       "step": 588
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.89e-05,
-      "loss": 6.8265,
+      "learning_rate": 0.00017669999999999999,
+      "loss": 6.9349,
       "step": 589
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.9e-05,
-      "loss": 6.8594,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 6.9985,
       "step": 590
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.91e-05,
-      "loss": 6.8713,
+      "learning_rate": 0.00017729999999999997,
+      "loss": 7.0269,
       "step": 591
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.92e-05,
-      "loss": 6.8877,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 7.0656,
       "step": 592
     },
     {
       "epoch": 0.66,
-      "learning_rate": 5.93e-05,
-      "loss": 6.9978,
+      "learning_rate": 0.0001779,
+      "loss": 7.1344,
       "step": 593
     },
     {
       "epoch": 0.67,
-      "learning_rate": 5.94e-05,
-      "loss": 6.682,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 6.834,
       "step": 594
     },
     {
       "epoch": 0.67,
-      "learning_rate": 5.95e-05,
-      "loss": 6.7372,
+      "learning_rate": 0.00017849999999999997,
+      "loss": 6.9316,
       "step": 595
     },
     {
       "epoch": 0.67,
-      "learning_rate": 5.96e-05,
-      "loss": 6.6122,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 6.7667,
       "step": 596
     },
     {
       "epoch": 0.67,
-      "learning_rate": 5.97e-05,
-      "loss": 6.6228,
+      "learning_rate": 0.0001791,
+      "loss": 6.7764,
       "step": 597
     },
     {
       "epoch": 0.67,
-      "learning_rate": 5.9800000000000003e-05,
-      "loss": 6.5612,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 6.7705,
       "step": 598
     },
     {
       "epoch": 0.67,
-      "learning_rate": 5.99e-05,
-      "loss": 6.4961,
+      "learning_rate": 0.00017969999999999998,
+      "loss": 6.7109,
       "step": 599
     },
     {
       "epoch": 0.67,
-      "learning_rate": 6e-05,
-      "loss": 6.1368,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 6.3912,
       "step": 600
     },
     {
       "epoch": 0.67,
-      "learning_rate": 6.0100000000000004e-05,
-      "loss": 7.1029,
+      "learning_rate": 0.00018029999999999996,
+      "loss": 7.1668,
       "step": 601
     },
     {
       "epoch": 0.67,
-      "learning_rate": 6.02e-05,
-      "loss": 6.9588,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 6.9761,
       "step": 602
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.03e-05,
-      "loss": 7.0916,
+      "learning_rate": 0.00018089999999999998,
+      "loss": 7.113,
       "step": 603
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.04e-05,
-      "loss": 7.0819,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 7.0443,
       "step": 604
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.05e-05,
-      "loss": 7.1343,
+      "learning_rate": 0.00018149999999999997,
+      "loss": 7.1449,
       "step": 605
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.06e-05,
-      "loss": 6.8571,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 6.8956,
       "step": 606
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.07e-05,
-      "loss": 6.9788,
+      "learning_rate": 0.00018209999999999998,
+      "loss": 7.0026,
       "step": 607
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.08e-05,
-      "loss": 7.001,
+      "learning_rate": 0.0001824,
+      "loss": 7.0658,
       "step": 608
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.09e-05,
-      "loss": 6.9375,
+      "learning_rate": 0.00018269999999999997,
+      "loss": 7.0272,
       "step": 609
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.1e-05,
-      "loss": 6.9305,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 7.0043,
       "step": 610
     },
     {
       "epoch": 0.68,
-      "learning_rate": 6.110000000000001e-05,
-      "loss": 6.9091,
+      "learning_rate": 0.00018329999999999998,
+      "loss": 6.9796,
       "step": 611
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.12e-05,
-      "loss": 6.9292,
+      "learning_rate": 0.0001836,
+      "loss": 7.0331,
       "step": 612
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.13e-05,
-      "loss": 6.846,
+      "learning_rate": 0.00018389999999999997,
+      "loss": 6.9296,
       "step": 613
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.14e-05,
-      "loss": 6.8987,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 7.0106,
       "step": 614
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.15e-05,
-      "loss": 7.097,
+      "learning_rate": 0.00018449999999999999,
+      "loss": 7.1683,
       "step": 615
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.16e-05,
-      "loss": 6.8019,
+      "learning_rate": 0.0001848,
+      "loss": 6.9167,
       "step": 616
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.170000000000001e-05,
-      "loss": 6.8996,
+      "learning_rate": 0.00018509999999999997,
+      "loss": 7.005,
       "step": 617
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.18e-05,
-      "loss": 6.7595,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 6.8711,
       "step": 618
     },
     {
       "epoch": 0.69,
-      "learning_rate": 6.19e-05,
-      "loss": 6.8725,
+      "learning_rate": 0.0001857,
+      "loss": 6.9648,
       "step": 619
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.2e-05,
-      "loss": 6.8947,
+      "learning_rate": 0.000186,
+      "loss": 7.0431,
       "step": 620
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.21e-05,
-      "loss": 6.9275,
+      "learning_rate": 0.00018629999999999997,
+      "loss": 7.063,
       "step": 621
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.220000000000001e-05,
-      "loss": 6.8547,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 6.9664,
       "step": 622
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.23e-05,
-      "loss": 6.965,
+      "learning_rate": 0.0001869,
+      "loss": 7.0641,
       "step": 623
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.24e-05,
-      "loss": 6.8124,
+      "learning_rate": 0.0001872,
+      "loss": 6.9357,
       "step": 624
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.25e-05,
-      "loss": 7.0048,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 7.1105,
       "step": 625
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.26e-05,
-      "loss": 6.8393,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 6.9633,
       "step": 626
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.27e-05,
-      "loss": 6.9773,
+      "learning_rate": 0.0001881,
+      "loss": 7.061,
       "step": 627
     },
     {
       "epoch": 0.7,
-      "learning_rate": 6.280000000000001e-05,
-      "loss": 6.958,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 7.049,
       "step": 628
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.29e-05,
-      "loss": 6.9277,
+      "learning_rate": 0.00018869999999999998,
+      "loss": 7.0243,
       "step": 629
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.3e-05,
-      "loss": 7.1124,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 7.224,
       "step": 630
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.31e-05,
-      "loss": 6.8434,
+      "learning_rate": 0.0001893,
+      "loss": 6.9557,
       "step": 631
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.32e-05,
-      "loss": 6.7448,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 6.8371,
       "step": 632
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.330000000000001e-05,
-      "loss": 6.9454,
+      "learning_rate": 0.00018989999999999998,
+      "loss": 7.0601,
       "step": 633
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.340000000000001e-05,
-      "loss": 6.8538,
+      "learning_rate": 0.0001902,
+      "loss": 6.9688,
       "step": 634
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.35e-05,
-      "loss": 6.9701,
+      "learning_rate": 0.0001905,
+      "loss": 7.0765,
       "step": 635
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.36e-05,
-      "loss": 6.9616,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 7.0849,
       "step": 636
     },
     {
       "epoch": 0.71,
-      "learning_rate": 6.37e-05,
-      "loss": 6.9076,
+      "learning_rate": 0.00019109999999999998,
+      "loss": 7.035,
       "step": 637
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.38e-05,
-      "loss": 6.7661,
+      "learning_rate": 0.0001914,
+      "loss": 6.8834,
       "step": 638
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.390000000000001e-05,
-      "loss": 7.0355,
+      "learning_rate": 0.0001917,
+      "loss": 7.161,
       "step": 639
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.400000000000001e-05,
-      "loss": 6.7776,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 6.9392,
       "step": 640
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.41e-05,
-      "loss": 6.8674,
+      "learning_rate": 0.00019229999999999999,
+      "loss": 6.9935,
       "step": 641
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.42e-05,
-      "loss": 6.99,
+      "learning_rate": 0.0001926,
+      "loss": 7.1394,
       "step": 642
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.43e-05,
-      "loss": 6.9037,
+      "learning_rate": 0.0001929,
+      "loss": 7.0425,
       "step": 643
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.440000000000001e-05,
-      "loss": 6.6599,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 6.8281,
       "step": 644
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.450000000000001e-05,
-      "loss": 6.6884,
+      "learning_rate": 0.0001935,
+      "loss": 6.862,
       "step": 645
     },
     {
       "epoch": 0.72,
-      "learning_rate": 6.460000000000001e-05,
-      "loss": 6.8665,
+      "learning_rate": 0.0001938,
+      "loss": 7.009,
       "step": 646
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.47e-05,
-      "loss": 6.6411,
+      "learning_rate": 0.0001941,
+      "loss": 6.7937,
       "step": 647
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.48e-05,
-      "loss": 6.5276,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 6.6993,
       "step": 648
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.49e-05,
-      "loss": 6.7051,
+      "learning_rate": 0.0001947,
+      "loss": 6.9282,
       "step": 649
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 6.4109,
+      "learning_rate": 0.000195,
+      "loss": 6.6272,
       "step": 650
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.510000000000001e-05,
-      "loss": 7.161,
+      "learning_rate": 0.00019529999999999998,
+      "loss": 7.1857,
       "step": 651
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.52e-05,
-      "loss": 7.148,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 7.1703,
       "step": 652
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.53e-05,
-      "loss": 7.1023,
+      "learning_rate": 0.0001959,
+      "loss": 7.1475,
       "step": 653
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.54e-05,
-      "loss": 6.9783,
+      "learning_rate": 0.0001962,
+      "loss": 7.0216,
       "step": 654
     },
     {
       "epoch": 0.73,
-      "learning_rate": 6.55e-05,
-      "loss": 7.0161,
+      "learning_rate": 0.00019649999999999998,
+      "loss": 7.1311,
       "step": 655
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.560000000000001e-05,
-      "loss": 7.0116,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 7.08,
       "step": 656
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.570000000000001e-05,
-      "loss": 6.787,
+      "learning_rate": 0.0001971,
+      "loss": 6.877,
       "step": 657
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.58e-05,
-      "loss": 7.1259,
+      "learning_rate": 0.0001974,
+      "loss": 7.2532,
       "step": 658
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.59e-05,
-      "loss": 6.8224,
+      "learning_rate": 0.00019769999999999998,
+      "loss": 6.9486,
       "step": 659
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.6e-05,
-      "loss": 6.8399,
+      "learning_rate": 0.000198,
+      "loss": 6.9518,
       "step": 660
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.610000000000001e-05,
-      "loss": 6.8718,
+      "learning_rate": 0.0001983,
+      "loss": 6.9861,
       "step": 661
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.620000000000001e-05,
-      "loss": 6.8949,
+      "learning_rate": 0.0001986,
+      "loss": 7.0336,
       "step": 662
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.630000000000001e-05,
-      "loss": 6.9755,
+      "learning_rate": 0.00019889999999999998,
+      "loss": 7.0591,
       "step": 663
     },
     {
       "epoch": 0.74,
-      "learning_rate": 6.64e-05,
-      "loss": 6.7638,
+      "learning_rate": 0.0001992,
+      "loss": 6.9311,
       "step": 664
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.65e-05,
-      "loss": 6.8473,
+      "learning_rate": 0.0001995,
+      "loss": 6.9483,
       "step": 665
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.66e-05,
-      "loss": 6.967,
+      "learning_rate": 0.0001998,
+      "loss": 7.0858,
       "step": 666
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.670000000000001e-05,
-      "loss": 6.868,
+      "learning_rate": 0.00020009999999999998,
+      "loss": 7.0208,
       "step": 667
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.680000000000001e-05,
-      "loss": 6.7926,
+      "learning_rate": 0.0002004,
+      "loss": 6.9166,
       "step": 668
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.690000000000001e-05,
-      "loss": 6.9591,
+      "learning_rate": 0.0002007,
+      "loss": 7.0713,
       "step": 669
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.7e-05,
-      "loss": 6.8435,
+      "learning_rate": 0.000201,
+      "loss": 6.9962,
       "step": 670
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.71e-05,
-      "loss": 6.7617,
+      "learning_rate": 0.0002013,
+      "loss": 6.8684,
       "step": 671
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.720000000000001e-05,
-      "loss": 6.8906,
+      "learning_rate": 0.0002016,
+      "loss": 7.021,
       "step": 672
     },
     {
       "epoch": 0.75,
-      "learning_rate": 6.730000000000001e-05,
-      "loss": 6.8486,
+      "learning_rate": 0.0002019,
+      "loss": 6.9878,
       "step": 673
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.740000000000001e-05,
-      "loss": 6.8124,
+      "learning_rate": 0.0002022,
+      "loss": 6.9211,
       "step": 674
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.750000000000001e-05,
-      "loss": 6.9221,
+      "learning_rate": 0.0002025,
+      "loss": 7.0502,
       "step": 675
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.76e-05,
-      "loss": 6.9739,
+      "learning_rate": 0.0002028,
+      "loss": 7.0849,
       "step": 676
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.77e-05,
-      "loss": 6.8727,
+      "learning_rate": 0.0002031,
+      "loss": 6.9855,
       "step": 677
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.780000000000001e-05,
-      "loss": 6.6738,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 6.8207,
       "step": 678
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.790000000000001e-05,
-      "loss": 6.8753,
+      "learning_rate": 0.0002037,
+      "loss": 7.0393,
       "step": 679
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.800000000000001e-05,
-      "loss": 6.8175,
+      "learning_rate": 0.000204,
+      "loss": 6.9265,
       "step": 680
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.81e-05,
-      "loss": 6.9644,
+      "learning_rate": 0.0002043,
+      "loss": 7.1252,
       "step": 681
     },
     {
       "epoch": 0.76,
-      "learning_rate": 6.82e-05,
-      "loss": 6.8539,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 6.9898,
       "step": 682
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.83e-05,
-      "loss": 6.8748,
+      "learning_rate": 0.0002049,
+      "loss": 6.9969,
       "step": 683
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.840000000000001e-05,
-      "loss": 6.9203,
+      "learning_rate": 0.0002052,
+      "loss": 7.0967,
       "step": 684
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.850000000000001e-05,
-      "loss": 6.9565,
+      "learning_rate": 0.0002055,
+      "loss": 7.0823,
       "step": 685
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.860000000000001e-05,
-      "loss": 6.7741,
+      "learning_rate": 0.0002058,
+      "loss": 6.9386,
       "step": 686
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.87e-05,
-      "loss": 6.7707,
+      "learning_rate": 0.0002061,
+      "loss": 6.897,
       "step": 687
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.879999999999999e-05,
-      "loss": 6.9914,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 7.1253,
       "step": 688
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.89e-05,
-      "loss": 6.868,
+      "learning_rate": 0.00020669999999999996,
+      "loss": 7.0263,
       "step": 689
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.9e-05,
-      "loss": 6.9214,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 7.0817,
       "step": 690
     },
     {
       "epoch": 0.77,
-      "learning_rate": 6.91e-05,
-      "loss": 6.7253,
+      "learning_rate": 0.00020729999999999997,
+      "loss": 6.9291,
       "step": 691
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.92e-05,
-      "loss": 6.8301,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 7.0098,
       "step": 692
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.93e-05,
-      "loss": 6.6455,
+      "learning_rate": 0.00020789999999999996,
+      "loss": 6.8191,
       "step": 693
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.939999999999999e-05,
-      "loss": 6.826,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 6.9643,
       "step": 694
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.95e-05,
-      "loss": 6.7612,
+      "learning_rate": 0.00020849999999999997,
+      "loss": 6.9201,
       "step": 695
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.96e-05,
-      "loss": 6.7723,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 6.9128,
       "step": 696
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.97e-05,
-      "loss": 6.6301,
+      "learning_rate": 0.00020909999999999996,
+      "loss": 6.8013,
       "step": 697
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.98e-05,
-      "loss": 6.4185,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 6.6455,
       "step": 698
     },
     {
       "epoch": 0.78,
-      "learning_rate": 6.99e-05,
-      "loss": 6.4977,
+      "learning_rate": 0.00020969999999999997,
+      "loss": 6.7599,
       "step": 699
     },
     {
       "epoch": 0.78,
-      "learning_rate": 7e-05,
-      "loss": 6.0836,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 6.4055,
       "step": 700
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.01e-05,
-      "loss": 7.0687,
+      "learning_rate": 0.00021029999999999996,
+      "loss": 7.12,
       "step": 701
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.02e-05,
-      "loss": 7.0183,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 7.0562,
       "step": 702
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.03e-05,
-      "loss": 6.9961,
+      "learning_rate": 0.00021089999999999998,
+      "loss": 7.022,
       "step": 703
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.04e-05,
-      "loss": 6.8668,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 6.9395,
       "step": 704
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.05e-05,
-      "loss": 6.9719,
+      "learning_rate": 0.00021149999999999996,
+      "loss": 7.0,
       "step": 705
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.06e-05,
-      "loss": 7.0097,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 7.1132,
       "step": 706
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.07e-05,
-      "loss": 7.0058,
+      "learning_rate": 0.00021209999999999998,
+      "loss": 7.0707,
       "step": 707
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.08e-05,
-      "loss": 7.1084,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 7.1993,
       "step": 708
     },
     {
       "epoch": 0.79,
-      "learning_rate": 7.09e-05,
-      "loss": 6.9168,
+      "learning_rate": 0.00021269999999999997,
+      "loss": 7.0121,
       "step": 709
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.1e-05,
-      "loss": 6.8363,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 7.0074,
       "step": 710
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.11e-05,
-      "loss": 6.8857,
+      "learning_rate": 0.00021329999999999998,
+      "loss": 7.0015,
       "step": 711
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.12e-05,
-      "loss": 6.743,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 6.8713,
       "step": 712
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.13e-05,
-      "loss": 6.8171,
+      "learning_rate": 0.00021389999999999997,
+      "loss": 6.9104,
       "step": 713
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.14e-05,
-      "loss": 6.8889,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 7.0322,
       "step": 714
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.15e-05,
-      "loss": 6.9596,
+      "learning_rate": 0.00021449999999999998,
+      "loss": 7.0898,
       "step": 715
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.16e-05,
-      "loss": 6.9834,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 7.0912,
       "step": 716
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.17e-05,
-      "loss": 7.1085,
+      "learning_rate": 0.00021509999999999997,
+      "loss": 7.2334,
       "step": 717
     },
     {
       "epoch": 0.8,
-      "learning_rate": 7.18e-05,
-      "loss": 6.8621,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 6.9729,
       "step": 718
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.19e-05,
-      "loss": 6.7457,
+      "learning_rate": 0.00021569999999999998,
+      "loss": 6.8601,
       "step": 719
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.2e-05,
-      "loss": 6.8237,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 6.9348,
       "step": 720
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.21e-05,
-      "loss": 6.8725,
+      "learning_rate": 0.00021629999999999997,
+      "loss": 7.015,
       "step": 721
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.22e-05,
-      "loss": 6.8154,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 6.9484,
       "step": 722
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.23e-05,
-      "loss": 6.786,
+      "learning_rate": 0.0002169,
+      "loss": 6.9264,
       "step": 723
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.24e-05,
-      "loss": 6.942,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 7.0595,
       "step": 724
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.25e-05,
-      "loss": 6.8645,
+      "learning_rate": 0.00021749999999999997,
+      "loss": 6.9979,
       "step": 725
     },
     {
       "epoch": 0.81,
-      "learning_rate": 7.26e-05,
-      "loss": 6.9925,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 7.1654,
       "step": 726
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.27e-05,
-      "loss": 6.8994,
+      "learning_rate": 0.00021809999999999996,
+      "loss": 7.0373,
       "step": 727
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.280000000000001e-05,
-      "loss": 6.7838,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 6.9379,
       "step": 728
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.29e-05,
-      "loss": 6.8165,
+      "learning_rate": 0.00021869999999999998,
+      "loss": 6.9828,
       "step": 729
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.3e-05,
-      "loss": 6.9149,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 7.0468,
       "step": 730
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.31e-05,
-      "loss": 7.0936,
+      "learning_rate": 0.00021929999999999996,
+      "loss": 7.2338,
       "step": 731
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.32e-05,
-      "loss": 6.9341,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 7.0789,
       "step": 732
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.33e-05,
-      "loss": 6.8379,
+      "learning_rate": 0.00021989999999999998,
+      "loss": 6.9872,
       "step": 733
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.340000000000001e-05,
-      "loss": 7.0602,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 7.1955,
       "step": 734
     },
     {
       "epoch": 0.82,
-      "learning_rate": 7.35e-05,
-      "loss": 6.8103,
+      "learning_rate": 0.00022049999999999997,
+      "loss": 6.9709,
       "step": 735
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.36e-05,
-      "loss": 6.885,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 7.0432,
       "step": 736
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.37e-05,
-      "loss": 6.857,
+      "learning_rate": 0.00022109999999999998,
+      "loss": 7.0212,
       "step": 737
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.38e-05,
-      "loss": 6.8849,
+      "learning_rate": 0.0002214,
+      "loss": 7.0663,
       "step": 738
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.390000000000001e-05,
-      "loss": 6.7723,
+      "learning_rate": 0.00022169999999999997,
+      "loss": 6.9312,
       "step": 739
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.4e-05,
-      "loss": 6.8779,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 7.0369,
       "step": 740
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.41e-05,
-      "loss": 6.8765,
+      "learning_rate": 0.00022229999999999998,
+      "loss": 7.0389,
       "step": 741
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.42e-05,
-      "loss": 6.8951,
+      "learning_rate": 0.0002226,
+      "loss": 7.0575,
       "step": 742
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.43e-05,
-      "loss": 6.7084,
+      "learning_rate": 0.00022289999999999997,
+      "loss": 6.8998,
       "step": 743
     },
     {
       "epoch": 0.83,
-      "learning_rate": 7.44e-05,
-      "loss": 6.853,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 6.9934,
       "step": 744
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.450000000000001e-05,
-      "loss": 6.711,
+      "learning_rate": 0.00022349999999999998,
+      "loss": 6.8997,
       "step": 745
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.46e-05,
-      "loss": 6.8895,
+      "learning_rate": 0.0002238,
+      "loss": 7.1195,
       "step": 746
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.47e-05,
-      "loss": 6.5634,
+      "learning_rate": 0.00022409999999999997,
+      "loss": 6.835,
       "step": 747
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.48e-05,
-      "loss": 6.578,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 6.7377,
       "step": 748
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.49e-05,
-      "loss": 6.4924,
+      "learning_rate": 0.0002247,
+      "loss": 6.7668,
       "step": 749
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 6.3147,
+      "learning_rate": 0.000225,
+      "loss": 6.6092,
       "step": 750
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.510000000000001e-05,
-      "loss": 7.0194,
+      "learning_rate": 0.00022529999999999997,
+      "loss": 7.0277,
       "step": 751
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.52e-05,
-      "loss": 7.0678,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 7.0592,
       "step": 752
     },
     {
       "epoch": 0.84,
-      "learning_rate": 7.53e-05,
-      "loss": 7.0482,
+      "learning_rate": 0.0002259,
+      "loss": 7.061,
       "step": 753
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.54e-05,
-      "loss": 6.9445,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 6.9947,
       "step": 754
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.55e-05,
-      "loss": 6.9443,
+      "learning_rate": 0.00022649999999999998,
+      "loss": 6.9703,
       "step": 755
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.560000000000001e-05,
-      "loss": 7.0371,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 7.0917,
       "step": 756
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.570000000000001e-05,
-      "loss": 6.9075,
+      "learning_rate": 0.0002271,
+      "loss": 7.0171,
       "step": 757
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.58e-05,
-      "loss": 6.8103,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 6.9377,
       "step": 758
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.59e-05,
-      "loss": 6.8983,
+      "learning_rate": 0.00022769999999999998,
+      "loss": 7.0249,
       "step": 759
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.6e-05,
-      "loss": 6.8274,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 6.9691,
       "step": 760
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.61e-05,
-      "loss": 6.9065,
+      "learning_rate": 0.0002283,
+      "loss": 7.0327,
       "step": 761
     },
     {
       "epoch": 0.85,
-      "learning_rate": 7.620000000000001e-05,
-      "loss": 6.8471,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 6.9876,
       "step": 762
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.630000000000001e-05,
-      "loss": 6.7048,
+      "learning_rate": 0.00022889999999999998,
+      "loss": 6.8536,
       "step": 763
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.64e-05,
-      "loss": 6.7163,
+      "learning_rate": 0.0002292,
+      "loss": 6.8258,
       "step": 764
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.65e-05,
-      "loss": 6.8448,
+      "learning_rate": 0.0002295,
+      "loss": 7.0073,
       "step": 765
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.66e-05,
-      "loss": 6.7835,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 6.9741,
       "step": 766
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.670000000000001e-05,
-      "loss": 6.7844,
+      "learning_rate": 0.00023009999999999998,
+      "loss": 6.9401,
       "step": 767
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.680000000000001e-05,
-      "loss": 6.8388,
+      "learning_rate": 0.0002304,
+      "loss": 6.9683,
       "step": 768
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.69e-05,
-      "loss": 6.9558,
+      "learning_rate": 0.0002307,
+      "loss": 7.0594,
       "step": 769
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.7e-05,
-      "loss": 6.753,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 6.914,
       "step": 770
     },
     {
       "epoch": 0.86,
-      "learning_rate": 7.71e-05,
-      "loss": 6.9372,
+      "learning_rate": 0.00023129999999999998,
+      "loss": 7.0689,
       "step": 771
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.72e-05,
-      "loss": 6.9966,
+      "learning_rate": 0.0002316,
+      "loss": 7.1266,
       "step": 772
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.730000000000001e-05,
-      "loss": 6.9252,
+      "learning_rate": 0.0002319,
+      "loss": 7.0636,
       "step": 773
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.740000000000001e-05,
-      "loss": 6.7173,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 6.9092,
       "step": 774
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.75e-05,
-      "loss": 6.6336,
+      "learning_rate": 0.00023249999999999999,
+      "loss": 6.768,
       "step": 775
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.76e-05,
-      "loss": 6.76,
+      "learning_rate": 0.0002328,
+      "loss": 6.8916,
       "step": 776
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.77e-05,
-      "loss": 6.6931,
+      "learning_rate": 0.00023309999999999997,
+      "loss": 6.8454,
       "step": 777
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.780000000000001e-05,
-      "loss": 6.8892,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 7.0369,
       "step": 778
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.790000000000001e-05,
-      "loss": 6.8471,
+      "learning_rate": 0.0002337,
+      "loss": 6.9757,
       "step": 779
     },
     {
       "epoch": 0.87,
-      "learning_rate": 7.800000000000001e-05,
-      "loss": 6.8263,
+      "learning_rate": 0.000234,
+      "loss": 6.991,
       "step": 780
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.81e-05,
-      "loss": 6.8019,
+      "learning_rate": 0.00023429999999999998,
+      "loss": 6.9953,
       "step": 781
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.82e-05,
-      "loss": 6.7788,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 6.9079,
       "step": 782
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.83e-05,
-      "loss": 6.9041,
+      "learning_rate": 0.0002349,
+      "loss": 7.0673,
       "step": 783
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.840000000000001e-05,
-      "loss": 6.7189,
+      "learning_rate": 0.0002352,
+      "loss": 6.918,
       "step": 784
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.850000000000001e-05,
-      "loss": 6.878,
+      "learning_rate": 0.00023549999999999998,
+      "loss": 7.0351,
       "step": 785
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.860000000000001e-05,
-      "loss": 6.7892,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 7.0,
       "step": 786
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.87e-05,
-      "loss": 6.8262,
+      "learning_rate": 0.0002361,
+      "loss": 6.9544,
       "step": 787
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.88e-05,
-      "loss": 6.8662,
+      "learning_rate": 0.0002364,
+      "loss": 7.0578,
       "step": 788
     },
     {
       "epoch": 0.88,
-      "learning_rate": 7.890000000000001e-05,
-      "loss": 6.8205,
+      "learning_rate": 0.00023669999999999998,
+      "loss": 6.9423,
       "step": 789
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.900000000000001e-05,
-      "loss": 6.8276,
+      "learning_rate": 0.000237,
+      "loss": 6.9214,
       "step": 790
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.910000000000001e-05,
-      "loss": 6.7078,
+      "learning_rate": 0.0002373,
+      "loss": 6.9172,
       "step": 791
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.920000000000001e-05,
-      "loss": 6.6245,
+      "learning_rate": 0.0002376,
+      "loss": 6.7942,
       "step": 792
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.93e-05,
-      "loss": 6.7751,
+      "learning_rate": 0.00023789999999999998,
+      "loss": 6.9887,
       "step": 793
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.94e-05,
-      "loss": 6.6748,
+      "learning_rate": 0.0002382,
+      "loss": 6.8503,
       "step": 794
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.950000000000001e-05,
-      "loss": 6.7857,
+      "learning_rate": 0.0002385,
+      "loss": 6.9834,
       "step": 795
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.960000000000001e-05,
-      "loss": 6.6818,
+      "learning_rate": 0.0002388,
+      "loss": 6.8774,
       "step": 796
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.970000000000001e-05,
-      "loss": 6.6669,
+      "learning_rate": 0.00023909999999999998,
+      "loss": 6.8861,
       "step": 797
     },
     {
       "epoch": 0.89,
-      "learning_rate": 7.98e-05,
-      "loss": 6.6309,
+      "learning_rate": 0.0002394,
+      "loss": 6.8851,
       "step": 798
     },
     {
       "epoch": 0.9,
-      "learning_rate": 7.99e-05,
-      "loss": 6.7343,
+      "learning_rate": 0.0002397,
+      "loss": 7.0403,
       "step": 799
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8e-05,
-      "loss": 6.105,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 6.5031,
       "step": 800
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8.010000000000001e-05,
-      "loss": 7.0499,
+      "learning_rate": 0.00024029999999999999,
+      "loss": 7.0173,
       "step": 801
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8.020000000000001e-05,
-      "loss": 6.9876,
+      "learning_rate": 0.0002406,
+      "loss": 7.0207,
       "step": 802
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8.030000000000001e-05,
-      "loss": 7.05,
+      "learning_rate": 0.0002409,
+      "loss": 7.0631,
       "step": 803
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8.04e-05,
-      "loss": 7.0119,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 7.0727,
       "step": 804
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8.05e-05,
-      "loss": 6.7699,
+      "learning_rate": 0.0002415,
+      "loss": 6.8426,
       "step": 805
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8.060000000000001e-05,
-      "loss": 6.9616,
+      "learning_rate": 0.0002418,
+      "loss": 7.0447,
       "step": 806
     },
     {
       "epoch": 0.9,
-      "learning_rate": 8.070000000000001e-05,
-      "loss": 6.9359,
+      "learning_rate": 0.0002421,
+      "loss": 7.0486,
       "step": 807
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.080000000000001e-05,
-      "loss": 6.7282,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 6.8542,
       "step": 808
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.090000000000001e-05,
-      "loss": 6.8661,
+      "learning_rate": 0.0002427,
+      "loss": 7.0249,
       "step": 809
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.1e-05,
-      "loss": 6.7949,
+      "learning_rate": 0.000243,
+      "loss": 6.9624,
       "step": 810
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.11e-05,
-      "loss": 6.8966,
+      "learning_rate": 0.0002433,
+      "loss": 7.0352,
       "step": 811
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.120000000000001e-05,
-      "loss": 6.9528,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 7.0877,
       "step": 812
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.13e-05,
-      "loss": 6.7941,
+      "learning_rate": 0.00024389999999999997,
+      "loss": 6.9089,
       "step": 813
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.14e-05,
-      "loss": 6.8723,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 6.9718,
       "step": 814
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.15e-05,
-      "loss": 6.897,
+      "learning_rate": 0.0002445,
+      "loss": 7.0001,
       "step": 815
     },
     {
       "epoch": 0.91,
-      "learning_rate": 8.16e-05,
-      "loss": 6.7847,
+      "learning_rate": 0.0002448,
+      "loss": 6.9555,
       "step": 816
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.17e-05,
-      "loss": 6.9361,
+      "learning_rate": 0.00024509999999999994,
+      "loss": 7.0658,
       "step": 817
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.18e-05,
-      "loss": 6.8414,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 6.9888,
       "step": 818
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.19e-05,
-      "loss": 6.7702,
+      "learning_rate": 0.00024569999999999995,
+      "loss": 6.922,
       "step": 819
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.2e-05,
-      "loss": 6.9013,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 7.0558,
       "step": 820
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.21e-05,
-      "loss": 6.9063,
+      "learning_rate": 0.00024629999999999997,
+      "loss": 7.0735,
       "step": 821
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.22e-05,
-      "loss": 6.8081,
+      "learning_rate": 0.0002466,
+      "loss": 6.9265,
       "step": 822
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.23e-05,
-      "loss": 6.9144,
+      "learning_rate": 0.0002469,
+      "loss": 7.0573,
       "step": 823
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.24e-05,
-      "loss": 6.7956,
+      "learning_rate": 0.0002472,
+      "loss": 6.9422,
       "step": 824
     },
     {
       "epoch": 0.92,
-      "learning_rate": 8.25e-05,
-      "loss": 7.0556,
+      "learning_rate": 0.00024749999999999994,
+      "loss": 7.1936,
       "step": 825
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.26e-05,
-      "loss": 6.9566,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 7.0872,
       "step": 826
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.27e-05,
-      "loss": 6.8352,
+      "learning_rate": 0.00024809999999999996,
+      "loss": 6.9692,
       "step": 827
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.28e-05,
-      "loss": 6.75,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 6.8901,
       "step": 828
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.29e-05,
-      "loss": 6.7798,
+      "learning_rate": 0.0002487,
+      "loss": 6.9535,
       "step": 829
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.3e-05,
-      "loss": 6.8131,
+      "learning_rate": 0.000249,
+      "loss": 6.9952,
       "step": 830
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.31e-05,
-      "loss": 6.745,
+      "learning_rate": 0.0002493,
+      "loss": 6.8827,
       "step": 831
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.32e-05,
-      "loss": 6.8108,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 6.9398,
       "step": 832
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.33e-05,
-      "loss": 6.8198,
+      "learning_rate": 0.00024989999999999995,
+      "loss": 6.9726,
       "step": 833
     },
     {
       "epoch": 0.93,
-      "learning_rate": 8.34e-05,
-      "loss": 6.8443,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 6.9834,
       "step": 834
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.35e-05,
-      "loss": 6.8043,
+      "learning_rate": 0.00025049999999999996,
+      "loss": 6.9625,
       "step": 835
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.36e-05,
-      "loss": 6.9266,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 7.0948,
       "step": 836
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.37e-05,
-      "loss": 6.8271,
+      "learning_rate": 0.0002511,
+      "loss": 6.9981,
       "step": 837
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.38e-05,
-      "loss": 6.9342,
+      "learning_rate": 0.0002514,
+      "loss": 7.0934,
       "step": 838
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.39e-05,
-      "loss": 6.7495,
+      "learning_rate": 0.0002517,
+      "loss": 6.9185,
       "step": 839
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.4e-05,
-      "loss": 6.6958,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 6.8492,
       "step": 840
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.41e-05,
-      "loss": 6.9511,
+      "learning_rate": 0.00025229999999999995,
+      "loss": 7.1197,
       "step": 841
     },
     {
       "epoch": 0.94,
-      "learning_rate": 8.42e-05,
-      "loss": 6.7255,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 6.9109,
       "step": 842
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.43e-05,
-      "loss": 6.9484,
+      "learning_rate": 0.00025289999999999997,
+      "loss": 7.1512,
       "step": 843
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.44e-05,
-      "loss": 6.7221,
+      "learning_rate": 0.0002532,
+      "loss": 6.9191,
       "step": 844
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.450000000000001e-05,
-      "loss": 6.7095,
+      "learning_rate": 0.0002535,
+      "loss": 6.9814,
       "step": 845
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.46e-05,
-      "loss": 6.564,
+      "learning_rate": 0.0002538,
+      "loss": 6.8625,
       "step": 846
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.47e-05,
-      "loss": 6.4785,
+      "learning_rate": 0.0002541,
+      "loss": 6.7735,
       "step": 847
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.48e-05,
-      "loss": 6.7376,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 6.9799,
       "step": 848
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.49e-05,
-      "loss": 6.4674,
+      "learning_rate": 0.00025469999999999996,
+      "loss": 6.6976,
       "step": 849
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.5e-05,
-      "loss": 6.132,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 6.4269,
       "step": 850
     },
     {
       "epoch": 0.95,
-      "learning_rate": 8.510000000000001e-05,
-      "loss": 7.1849,
+      "learning_rate": 0.00025529999999999997,
+      "loss": 7.0714,
       "step": 851
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.52e-05,
-      "loss": 7.1504,
+      "learning_rate": 0.0002556,
+      "loss": 7.041,
       "step": 852
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.53e-05,
-      "loss": 7.2565,
+      "learning_rate": 0.0002559,
+      "loss": 7.1556,
       "step": 853
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.54e-05,
-      "loss": 7.109,
+      "learning_rate": 0.0002562,
+      "loss": 7.1243,
       "step": 854
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.55e-05,
-      "loss": 6.9596,
+      "learning_rate": 0.00025649999999999995,
+      "loss": 7.0138,
       "step": 855
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.560000000000001e-05,
-      "loss": 7.029,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 7.1105,
       "step": 856
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.57e-05,
-      "loss": 7.0794,
+      "learning_rate": 0.00025709999999999996,
+      "loss": 7.1453,
       "step": 857
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.58e-05,
-      "loss": 6.9885,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 7.0781,
       "step": 858
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.59e-05,
-      "loss": 6.7794,
+      "learning_rate": 0.0002577,
+      "loss": 6.9069,
       "step": 859
     },
     {
       "epoch": 0.96,
-      "learning_rate": 8.6e-05,
-      "loss": 6.8992,
+      "learning_rate": 0.000258,
+      "loss": 6.9845,
       "step": 860
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.61e-05,
-      "loss": 7.0632,
+      "learning_rate": 0.0002583,
+      "loss": 7.1599,
       "step": 861
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.620000000000001e-05,
-      "loss": 6.9089,
+      "learning_rate": 0.0002586,
+      "loss": 7.0317,
       "step": 862
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.63e-05,
-      "loss": 6.8525,
+      "learning_rate": 0.00025889999999999995,
+      "loss": 7.0027,
       "step": 863
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.64e-05,
-      "loss": 7.0363,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 7.1109,
       "step": 864
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.65e-05,
-      "loss": 6.9645,
+      "learning_rate": 0.00025949999999999997,
+      "loss": 7.1168,
       "step": 865
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.66e-05,
-      "loss": 6.7923,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 6.9035,
       "step": 866
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.67e-05,
-      "loss": 7.0068,
+      "learning_rate": 0.0002601,
+      "loss": 7.1186,
       "step": 867
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.680000000000001e-05,
-      "loss": 6.7872,
+      "learning_rate": 0.0002604,
+      "loss": 6.887,
       "step": 868
     },
     {
       "epoch": 0.97,
-      "learning_rate": 8.69e-05,
-      "loss": 6.8906,
+      "learning_rate": 0.0002607,
+      "loss": 7.0143,
       "step": 869
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.7e-05,
-      "loss": 6.8413,
+      "learning_rate": 0.000261,
+      "loss": 6.979,
       "step": 870
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.71e-05,
-      "loss": 6.8052,
+      "learning_rate": 0.00026129999999999995,
+      "loss": 6.9135,
       "step": 871
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.72e-05,
-      "loss": 6.8843,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 7.0007,
       "step": 872
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.730000000000001e-05,
-      "loss": 6.8487,
+      "learning_rate": 0.00026189999999999997,
+      "loss": 6.9807,
       "step": 873
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.740000000000001e-05,
-      "loss": 6.9423,
+      "learning_rate": 0.0002622,
+      "loss": 7.0661,
       "step": 874
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.75e-05,
-      "loss": 6.9963,
+      "learning_rate": 0.0002625,
+      "loss": 7.1212,
       "step": 875
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.76e-05,
-      "loss": 6.7368,
+      "learning_rate": 0.0002628,
+      "loss": 6.8602,
       "step": 876
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.77e-05,
-      "loss": 6.694,
+      "learning_rate": 0.0002631,
+      "loss": 6.8811,
       "step": 877
     },
     {
       "epoch": 0.98,
-      "learning_rate": 8.78e-05,
-      "loss": 6.8147,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 6.935,
       "step": 878
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.790000000000001e-05,
-      "loss": 6.8343,
+      "learning_rate": 0.00026369999999999996,
+      "loss": 6.969,
       "step": 879
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.800000000000001e-05,
-      "loss": 6.9569,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 7.0835,
       "step": 880
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.81e-05,
-      "loss": 6.6338,
+      "learning_rate": 0.0002643,
+      "loss": 6.7698,
       "step": 881
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.82e-05,
-      "loss": 6.8634,
+      "learning_rate": 0.0002646,
+      "loss": 7.0172,
       "step": 882
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.83e-05,
-      "loss": 6.7133,
+      "learning_rate": 0.0002649,
+      "loss": 6.8346,
       "step": 883
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.840000000000001e-05,
-      "loss": 6.8771,
+      "learning_rate": 0.0002652,
+      "loss": 7.0038,
       "step": 884
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.850000000000001e-05,
-      "loss": 6.8077,
+      "learning_rate": 0.0002655,
+      "loss": 6.9791,
       "step": 885
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.86e-05,
-      "loss": 6.8339,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 6.9784,
       "step": 886
     },
     {
       "epoch": 0.99,
-      "learning_rate": 8.87e-05,
-      "loss": 6.5647,
+      "learning_rate": 0.00026609999999999996,
+      "loss": 6.7272,
       "step": 887
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.88e-05,
-      "loss": 6.7102,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 6.906,
       "step": 888
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.89e-05,
-      "loss": 6.6053,
+      "learning_rate": 0.0002667,
+      "loss": 6.804,
       "step": 889
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.900000000000001e-05,
-      "loss": 6.5223,
+      "learning_rate": 0.000267,
+      "loss": 6.7746,
       "step": 890
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.910000000000001e-05,
-      "loss": 6.3557,
+      "learning_rate": 0.0002673,
+      "loss": 6.5674,
       "step": 891
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.92e-05,
-      "loss": 6.024,
+      "learning_rate": 0.0002676,
+      "loss": 6.2881,
       "step": 892
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.93e-05,
-      "loss": 7.1035,
+      "learning_rate": 0.0002679,
+      "loss": 7.1967,
       "step": 893
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.94e-05,
-      "loss": 7.1815,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 7.2593,
       "step": 894
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.950000000000001e-05,
-      "loss": 6.84,
+      "learning_rate": 0.00026849999999999997,
+      "loss": 6.953,
       "step": 895
     },
     {
       "epoch": 1.0,
-      "learning_rate": 8.960000000000001e-05,
-      "loss": 6.851,
+      "learning_rate": 0.0002688,
+      "loss": 6.963,
       "step": 896
     },
     {
       "epoch": 1.01,
-      "learning_rate": 8.970000000000001e-05,
-      "loss": 6.8529,
+      "learning_rate": 0.0002691,
+      "loss": 6.9957,
       "step": 897
     },
     {
       "epoch": 1.01,
-      "learning_rate": 8.98e-05,
-      "loss": 6.9552,
+      "learning_rate": 0.0002694,
+      "loss": 7.0339,
       "step": 898
     },
     {
       "epoch": 1.01,
-      "learning_rate": 8.99e-05,
-      "loss": 6.8332,
+      "learning_rate": 0.0002697,
+      "loss": 6.9958,
       "step": 899
     },
     {
       "epoch": 1.01,
-      "learning_rate": 9e-05,
-      "loss": 6.9165,
+      "learning_rate": 0.00027,
+      "loss": 7.0439,
       "step": 900
     },
     {
       "epoch": 1.01,
-      "learning_rate": 9.010000000000001e-05,
-      "loss": 6.9338,
+      "learning_rate": 0.00027029999999999996,
+      "loss": 7.0787,
       "step": 901
     },
     {
       "epoch": 1.01,
-      "learning_rate": 9.020000000000001e-05,
-      "loss": 6.9225,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 7.0502,
       "step": 902
     },
     {
       "epoch": 1.01,
-      "learning_rate": 9.030000000000001e-05,
-      "loss": 6.7139,
+      "learning_rate": 0.00027089999999999997,
+      "loss": 6.8449,
       "step": 903
     },
     {
       "epoch": 1.01,
-      "learning_rate": 9.04e-05,
-      "loss": 6.8658,
+      "learning_rate": 0.0002712,
+      "loss": 7.0154,
       "step": 904
     },
     {
       "epoch": 1.01,
-      "learning_rate": 9.05e-05,
-      "loss": 6.7132,
+      "learning_rate": 0.0002715,
+      "loss": 6.9135,
       "step": 905
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.06e-05,
-      "loss": 6.7589,
+      "learning_rate": 0.0002718,
+      "loss": 6.9461,
       "step": 906
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.070000000000001e-05,
-      "loss": 6.9746,
+      "learning_rate": 0.0002721,
+      "loss": 7.134,
       "step": 907
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.080000000000001e-05,
-      "loss": 6.9491,
+      "learning_rate": 0.0002724,
+      "loss": 7.1246,
       "step": 908
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.090000000000001e-05,
-      "loss": 6.8044,
+      "learning_rate": 0.00027269999999999996,
+      "loss": 6.9381,
       "step": 909
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.1e-05,
-      "loss": 6.6508,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 6.8045,
       "step": 910
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.11e-05,
-      "loss": 6.7558,
+      "learning_rate": 0.0002733,
+      "loss": 6.9037,
       "step": 911
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.120000000000001e-05,
-      "loss": 6.7664,
+      "learning_rate": 0.0002736,
+      "loss": 6.9561,
       "step": 912
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.130000000000001e-05,
-      "loss": 6.8548,
+      "learning_rate": 0.0002739,
+      "loss": 7.0199,
       "step": 913
     },
     {
       "epoch": 1.02,
-      "learning_rate": 9.140000000000001e-05,
-      "loss": 6.8507,
+      "learning_rate": 0.0002742,
+      "loss": 7.0162,
       "step": 914
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.15e-05,
-      "loss": 6.6343,
+      "learning_rate": 0.0002745,
+      "loss": 6.8227,
       "step": 915
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.16e-05,
-      "loss": 6.7585,
+      "learning_rate": 0.0002748,
+      "loss": 6.943,
       "step": 916
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.17e-05,
-      "loss": 6.725,
+      "learning_rate": 0.00027509999999999996,
+      "loss": 6.9006,
       "step": 917
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.180000000000001e-05,
-      "loss": 6.8298,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 6.977,
       "step": 918
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.190000000000001e-05,
-      "loss": 6.8499,
+      "learning_rate": 0.0002757,
+      "loss": 7.0232,
       "step": 919
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.200000000000001e-05,
-      "loss": 6.7042,
+      "learning_rate": 0.000276,
+      "loss": 6.9054,
       "step": 920
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.21e-05,
-      "loss": 6.7175,
+      "learning_rate": 0.0002763,
+      "loss": 6.8834,
       "step": 921
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.22e-05,
-      "loss": 6.9401,
+      "learning_rate": 0.0002766,
+      "loss": 7.1196,
       "step": 922
     },
     {
       "epoch": 1.03,
-      "learning_rate": 9.230000000000001e-05,
-      "loss": 6.8277,
+      "learning_rate": 0.0002769,
+      "loss": 7.0089,
       "step": 923
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.240000000000001e-05,
-      "loss": 6.8287,
+      "learning_rate": 0.0002772,
+      "loss": 7.0042,
       "step": 924
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.250000000000001e-05,
-      "loss": 6.8019,
+      "learning_rate": 0.00027749999999999997,
+      "loss": 6.9491,
       "step": 925
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.260000000000001e-05,
-      "loss": 6.7668,
+      "learning_rate": 0.0002778,
+      "loss": 6.9522,
       "step": 926
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.27e-05,
-      "loss": 6.8041,
+      "learning_rate": 0.0002781,
+      "loss": 6.9624,
       "step": 927
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.28e-05,
-      "loss": 6.7687,
+      "learning_rate": 0.0002784,
+      "loss": 6.9376,
       "step": 928
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.290000000000001e-05,
-      "loss": 6.8275,
+      "learning_rate": 0.0002787,
+      "loss": 6.9789,
       "step": 929
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.300000000000001e-05,
-      "loss": 6.9067,
+      "learning_rate": 0.000279,
+      "loss": 7.1098,
       "step": 930
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.310000000000001e-05,
-      "loss": 6.7715,
+      "learning_rate": 0.0002793,
+      "loss": 6.9675,
       "step": 931
     },
     {
       "epoch": 1.04,
-      "learning_rate": 9.320000000000002e-05,
-      "loss": 6.7018,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 6.8483,
       "step": 932
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.33e-05,
-      "loss": 6.7703,
+      "learning_rate": 0.0002799,
+      "loss": 6.9588,
       "step": 933
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.340000000000001e-05,
-      "loss": 6.8784,
+      "learning_rate": 0.0002802,
+      "loss": 7.1063,
       "step": 934
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.350000000000001e-05,
-      "loss": 6.6638,
+      "learning_rate": 0.0002805,
+      "loss": 6.8421,
       "step": 935
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.360000000000001e-05,
-      "loss": 6.617,
+      "learning_rate": 0.0002808,
+      "loss": 6.8531,
       "step": 936
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.370000000000001e-05,
-      "loss": 6.7837,
+      "learning_rate": 0.0002811,
+      "loss": 6.9779,
       "step": 937
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.38e-05,
-      "loss": 6.5596,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 6.8109,
       "step": 938
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.39e-05,
-      "loss": 6.3513,
+      "learning_rate": 0.00028169999999999996,
+      "loss": 6.6152,
       "step": 939
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.4e-05,
-      "loss": 6.5328,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 6.8361,
       "step": 940
     },
     {
       "epoch": 1.05,
-      "learning_rate": 9.41e-05,
-      "loss": 6.301,
+      "learning_rate": 0.0002823,
+      "loss": 6.6731,
       "step": 941
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.42e-05,
-      "loss": 6.3741,
+      "learning_rate": 0.0002826,
+      "loss": 6.8068,
       "step": 942
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.43e-05,
-      "loss": 6.9741,
+      "learning_rate": 0.00028289999999999994,
+      "loss": 6.9916,
       "step": 943
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.44e-05,
-      "loss": 6.8678,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 6.9033,
       "step": 944
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.449999999999999e-05,
-      "loss": 7.0961,
+      "learning_rate": 0.00028349999999999995,
+      "loss": 7.1316,
       "step": 945
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.46e-05,
-      "loss": 6.8965,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 6.9697,
       "step": 946
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.47e-05,
-      "loss": 7.0457,
+      "learning_rate": 0.00028409999999999997,
+      "loss": 7.1034,
       "step": 947
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.48e-05,
-      "loss": 6.9916,
+      "learning_rate": 0.0002844,
+      "loss": 7.106,
       "step": 948
     },
     {
       "epoch": 1.06,
-      "learning_rate": 9.49e-05,
-      "loss": 6.8464,
+      "learning_rate": 0.0002847,
+      "loss": 7.0118,
       "step": 949
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.5e-05,
-      "loss": 6.9537,
+      "learning_rate": 0.000285,
+      "loss": 7.1486,
       "step": 950
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.51e-05,
-      "loss": 6.6696,
+      "learning_rate": 0.00028529999999999994,
+      "loss": 6.8273,
       "step": 951
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.52e-05,
-      "loss": 6.8794,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 7.0408,
       "step": 952
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.53e-05,
-      "loss": 6.8356,
+      "learning_rate": 0.00028589999999999996,
+      "loss": 6.9993,
       "step": 953
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.54e-05,
-      "loss": 6.7324,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 6.924,
       "step": 954
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.55e-05,
-      "loss": 7.022,
+      "learning_rate": 0.00028649999999999997,
+      "loss": 7.1493,
       "step": 955
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.56e-05,
-      "loss": 6.8157,
+      "learning_rate": 0.0002868,
+      "loss": 7.0309,
       "step": 956
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.57e-05,
-      "loss": 6.8341,
+      "learning_rate": 0.0002871,
+      "loss": 6.9915,
       "step": 957
     },
     {
       "epoch": 1.07,
-      "learning_rate": 9.58e-05,
-      "loss": 6.8282,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 6.9844,
       "step": 958
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.59e-05,
-      "loss": 6.9325,
+      "learning_rate": 0.00028769999999999995,
+      "loss": 7.0603,
       "step": 959
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.6e-05,
-      "loss": 6.7229,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 6.8874,
       "step": 960
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.61e-05,
-      "loss": 6.7774,
+      "learning_rate": 0.00028829999999999996,
+      "loss": 6.9932,
       "step": 961
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.620000000000001e-05,
-      "loss": 6.8715,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 7.0442,
       "step": 962
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.63e-05,
-      "loss": 6.8987,
+      "learning_rate": 0.0002889,
+      "loss": 7.0798,
       "step": 963
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.64e-05,
-      "loss": 6.8257,
+      "learning_rate": 0.0002892,
+      "loss": 6.9777,
       "step": 964
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.65e-05,
-      "loss": 6.7984,
+      "learning_rate": 0.0002895,
+      "loss": 6.9684,
       "step": 965
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.66e-05,
-      "loss": 6.8233,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 6.9836,
       "step": 966
     },
     {
       "epoch": 1.08,
-      "learning_rate": 9.67e-05,
-      "loss": 6.7268,
+      "learning_rate": 0.00029009999999999995,
+      "loss": 6.9024,
       "step": 967
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.680000000000001e-05,
-      "loss": 6.6949,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 6.8622,
       "step": 968
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.69e-05,
-      "loss": 6.9409,
+      "learning_rate": 0.00029069999999999996,
+      "loss": 7.0725,
       "step": 969
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.7e-05,
-      "loss": 6.7298,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 6.9589,
       "step": 970
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.71e-05,
-      "loss": 6.8041,
+      "learning_rate": 0.0002913,
+      "loss": 7.0023,
       "step": 971
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.72e-05,
-      "loss": 6.7982,
+      "learning_rate": 0.0002916,
+      "loss": 6.967,
       "step": 972
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.730000000000001e-05,
-      "loss": 6.8162,
+      "learning_rate": 0.0002919,
+      "loss": 7.036,
       "step": 973
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.74e-05,
-      "loss": 6.902,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 7.0541,
       "step": 974
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.75e-05,
-      "loss": 6.7075,
+      "learning_rate": 0.00029249999999999995,
+      "loss": 6.9392,
       "step": 975
     },
     {
       "epoch": 1.09,
-      "learning_rate": 9.76e-05,
-      "loss": 6.7672,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 7.0188,
       "step": 976
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.77e-05,
-      "loss": 6.9432,
+      "learning_rate": 0.00029309999999999997,
+      "loss": 7.0982,
       "step": 977
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.78e-05,
-      "loss": 6.6737,
+      "learning_rate": 0.0002934,
+      "loss": 6.8706,
       "step": 978
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.790000000000001e-05,
-      "loss": 6.7456,
+      "learning_rate": 0.0002937,
+      "loss": 6.9848,
       "step": 979
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.8e-05,
-      "loss": 6.7291,
+      "learning_rate": 0.000294,
+      "loss": 6.9211,
       "step": 980
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.81e-05,
-      "loss": 6.708,
+      "learning_rate": 0.00029429999999999994,
+      "loss": 6.9283,
       "step": 981
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.82e-05,
-      "loss": 6.9339,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 7.1223,
       "step": 982
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.83e-05,
-      "loss": 7.016,
+      "learning_rate": 0.00029489999999999996,
+      "loss": 7.2257,
       "step": 983
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.84e-05,
-      "loss": 6.7448,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 6.938,
       "step": 984
     },
     {
       "epoch": 1.1,
-      "learning_rate": 9.850000000000001e-05,
-      "loss": 6.5576,
+      "learning_rate": 0.00029549999999999997,
+      "loss": 6.8249,
       "step": 985
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.86e-05,
-      "loss": 6.8618,
+      "learning_rate": 0.0002958,
+      "loss": 6.9414,
       "step": 986
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.87e-05,
-      "loss": 6.5983,
+      "learning_rate": 0.0002961,
+      "loss": 6.8512,
       "step": 987
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.88e-05,
-      "loss": 6.4557,
+      "learning_rate": 0.0002964,
+      "loss": 6.7299,
       "step": 988
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.89e-05,
-      "loss": 6.3591,
+      "learning_rate": 0.00029669999999999995,
+      "loss": 6.672,
       "step": 989
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.900000000000001e-05,
-      "loss": 6.5566,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 6.8087,
       "step": 990
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.910000000000001e-05,
-      "loss": 6.1116,
+      "learning_rate": 0.00029729999999999996,
+      "loss": 6.4623,
       "step": 991
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.92e-05,
-      "loss": 5.961,
+      "learning_rate": 0.00029759999999999997,
+      "loss": 6.3527,
       "step": 992
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.93e-05,
-      "loss": 6.8168,
+      "learning_rate": 0.0002979,
+      "loss": 6.9607,
       "step": 993
     },
     {
       "epoch": 1.11,
-      "learning_rate": 9.94e-05,
-      "loss": 6.9493,
+      "learning_rate": 0.0002982,
+      "loss": 7.0054,
       "step": 994
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.95e-05,
-      "loss": 6.9552,
+      "learning_rate": 0.0002985,
+      "loss": 7.0058,
       "step": 995
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.960000000000001e-05,
-      "loss": 6.975,
+      "learning_rate": 0.0002988,
+      "loss": 7.0884,
       "step": 996
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.970000000000001e-05,
-      "loss": 6.7764,
+      "learning_rate": 0.00029909999999999995,
+      "loss": 6.9105,
       "step": 997
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.98e-05,
-      "loss": 6.989,
+      "learning_rate": 0.00029939999999999996,
+      "loss": 7.1735,
       "step": 998
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.99e-05,
-      "loss": 6.8327,
+      "learning_rate": 0.00029969999999999997,
+      "loss": 6.9737,
       "step": 999
     },
     {
       "epoch": 1.12,
-      "learning_rate": 0.0001,
-      "loss": 6.6442,
+      "learning_rate": 0.0003,
+      "loss": 6.8231,
       "step": 1000
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.997109826589596e-05,
-      "loss": 6.8448,
+      "learning_rate": 0.00029991329479768785,
+      "loss": 7.001,
       "step": 1001
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.994219653179191e-05,
-      "loss": 6.8876,
+      "learning_rate": 0.00029982658959537567,
+      "loss": 7.026,
       "step": 1002
     },
     {
       "epoch": 1.12,
-      "learning_rate": 9.991329479768787e-05,
-      "loss": 6.7732,
+      "learning_rate": 0.00029973988439306354,
+      "loss": 6.9912,
       "step": 1003
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.988439306358382e-05,
-      "loss": 6.8916,
+      "learning_rate": 0.0002996531791907514,
+      "loss": 7.0669,
       "step": 1004
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.985549132947977e-05,
-      "loss": 6.9549,
+      "learning_rate": 0.0002995664739884393,
+      "loss": 7.0908,
       "step": 1005
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.982658959537573e-05,
-      "loss": 6.7951,
+      "learning_rate": 0.00029947976878612716,
+      "loss": 6.971,
       "step": 1006
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.979768786127168e-05,
-      "loss": 6.7415,
+      "learning_rate": 0.000299393063583815,
+      "loss": 6.9087,
       "step": 1007
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.976878612716763e-05,
-      "loss": 6.8201,
+      "learning_rate": 0.00029930635838150286,
+      "loss": 7.0051,
       "step": 1008
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.973988439306359e-05,
-      "loss": 6.8066,
+      "learning_rate": 0.00029921965317919073,
+      "loss": 6.9887,
       "step": 1009
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.971098265895954e-05,
-      "loss": 6.8587,
+      "learning_rate": 0.0002991329479768786,
+      "loss": 7.0412,
       "step": 1010
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.968208092485549e-05,
-      "loss": 6.8517,
+      "learning_rate": 0.0002990462427745664,
+      "loss": 7.0099,
       "step": 1011
     },
     {
       "epoch": 1.13,
-      "learning_rate": 9.965317919075145e-05,
-      "loss": 6.9332,
+      "learning_rate": 0.0002989595375722543,
+      "loss": 7.0963,
       "step": 1012
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.962427745664741e-05,
-      "loss": 6.7091,
+      "learning_rate": 0.00029887283236994217,
+      "loss": 6.9095,
       "step": 1013
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.959537572254337e-05,
-      "loss": 6.8574,
+      "learning_rate": 0.00029878612716763005,
+      "loss": 6.9932,
       "step": 1014
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.95664739884393e-05,
-      "loss": 6.6962,
+      "learning_rate": 0.0002986994219653179,
+      "loss": 6.8215,
       "step": 1015
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.953757225433526e-05,
-      "loss": 6.767,
+      "learning_rate": 0.00029861271676300574,
+      "loss": 6.9744,
       "step": 1016
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.950867052023121e-05,
-      "loss": 6.7628,
+      "learning_rate": 0.0002985260115606936,
+      "loss": 6.9133,
       "step": 1017
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.947976878612717e-05,
-      "loss": 6.8646,
+      "learning_rate": 0.0002984393063583815,
+      "loss": 7.0339,
       "step": 1018
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.945086705202312e-05,
-      "loss": 6.8293,
+      "learning_rate": 0.0002983526011560693,
+      "loss": 6.9964,
       "step": 1019
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.942196531791907e-05,
-      "loss": 6.9041,
+      "learning_rate": 0.0002982658959537572,
+      "loss": 7.0872,
       "step": 1020
     },
     {
       "epoch": 1.14,
-      "learning_rate": 9.939306358381504e-05,
-      "loss": 6.8285,
+      "learning_rate": 0.00029817919075144505,
+      "loss": 7.0037,
       "step": 1021
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.9364161849711e-05,
-      "loss": 6.7863,
+      "learning_rate": 0.00029809248554913293,
+      "loss": 6.971,
       "step": 1022
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.933526011560695e-05,
-      "loss": 6.7558,
+      "learning_rate": 0.0002980057803468208,
+      "loss": 6.9502,
       "step": 1023
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.93063583815029e-05,
-      "loss": 6.8689,
+      "learning_rate": 0.0002979190751445086,
+      "loss": 7.0391,
       "step": 1024
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.927745664739884e-05,
-      "loss": 6.6102,
+      "learning_rate": 0.0002978323699421965,
+      "loss": 6.8286,
       "step": 1025
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.92485549132948e-05,
-      "loss": 6.6934,
+      "learning_rate": 0.00029774566473988437,
+      "loss": 6.8919,
       "step": 1026
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.921965317919075e-05,
-      "loss": 6.9347,
+      "learning_rate": 0.00029765895953757224,
+      "loss": 7.096,
       "step": 1027
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.91907514450867e-05,
-      "loss": 6.6877,
+      "learning_rate": 0.00029757225433526006,
+      "loss": 6.8504,
       "step": 1028
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.916184971098267e-05,
-      "loss": 6.5464,
+      "learning_rate": 0.000297485549132948,
+      "loss": 6.7955,
       "step": 1029
     },
     {
       "epoch": 1.15,
-      "learning_rate": 9.913294797687862e-05,
-      "loss": 6.8075,
+      "learning_rate": 0.0002973988439306358,
+      "loss": 7.0063,
       "step": 1030
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.910404624277458e-05,
-      "loss": 6.638,
+      "learning_rate": 0.0002973121387283237,
+      "loss": 6.8344,
       "step": 1031
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.907514450867053e-05,
-      "loss": 6.6992,
+      "learning_rate": 0.00029722543352601156,
+      "loss": 6.9072,
       "step": 1032
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.904624277456648e-05,
-      "loss": 6.7608,
+      "learning_rate": 0.0002971387283236994,
+      "loss": 6.982,
       "step": 1033
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.901734104046244e-05,
-      "loss": 6.7228,
+      "learning_rate": 0.00029705202312138725,
+      "loss": 6.9852,
       "step": 1034
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.898843930635839e-05,
-      "loss": 6.8435,
+      "learning_rate": 0.0002969653179190751,
+      "loss": 7.0498,
       "step": 1035
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.895953757225433e-05,
-      "loss": 6.7073,
+      "learning_rate": 0.00029687861271676295,
+      "loss": 6.9429,
       "step": 1036
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.89306358381503e-05,
-      "loss": 6.6076,
+      "learning_rate": 0.0002967919075144509,
+      "loss": 6.894,
       "step": 1037
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.890173410404625e-05,
-      "loss": 6.6714,
+      "learning_rate": 0.0002967052023121387,
+      "loss": 6.9482,
       "step": 1038
     },
     {
       "epoch": 1.16,
-      "learning_rate": 9.88728323699422e-05,
-      "loss": 6.539,
+      "learning_rate": 0.00029661849710982657,
+      "loss": 6.7851,
       "step": 1039
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.884393063583816e-05,
-      "loss": 6.5178,
+      "learning_rate": 0.00029653179190751444,
+      "loss": 6.79,
       "step": 1040
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.881502890173411e-05,
-      "loss": 6.2161,
+      "learning_rate": 0.00029644508670520226,
+      "loss": 6.601,
       "step": 1041
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.878612716763006e-05,
-      "loss": 6.0046,
+      "learning_rate": 0.00029635838150289014,
+      "loss": 6.3764,
       "step": 1042
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.875722543352602e-05,
-      "loss": 7.0683,
+      "learning_rate": 0.000296271676300578,
+      "loss": 7.0497,
       "step": 1043
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.872832369942197e-05,
-      "loss": 7.0125,
+      "learning_rate": 0.00029618497109826583,
+      "loss": 6.9928,
       "step": 1044
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.869942196531792e-05,
-      "loss": 7.0778,
+      "learning_rate": 0.00029609826589595376,
+      "loss": 7.0496,
       "step": 1045
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.867052023121388e-05,
-      "loss": 6.9903,
+      "learning_rate": 0.0002960115606936416,
+      "loss": 6.9892,
       "step": 1046
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.864161849710983e-05,
-      "loss": 6.9986,
+      "learning_rate": 0.00029592485549132945,
+      "loss": 7.0501,
       "step": 1047
     },
     {
       "epoch": 1.17,
-      "learning_rate": 9.861271676300578e-05,
-      "loss": 6.893,
+      "learning_rate": 0.0002958381502890173,
+      "loss": 6.9571,
       "step": 1048
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.858381502890174e-05,
-      "loss": 6.9212,
+      "learning_rate": 0.0002957514450867052,
+      "loss": 7.0141,
       "step": 1049
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.855491329479769e-05,
-      "loss": 6.8722,
+      "learning_rate": 0.000295664739884393,
+      "loss": 6.9528,
       "step": 1050
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.852601156069364e-05,
-      "loss": 6.8042,
+      "learning_rate": 0.0002955780346820809,
+      "loss": 6.9328,
       "step": 1051
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.84971098265896e-05,
-      "loss": 7.0224,
+      "learning_rate": 0.00029549132947976877,
+      "loss": 7.122,
       "step": 1052
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.846820809248556e-05,
-      "loss": 6.8286,
+      "learning_rate": 0.00029540462427745664,
+      "loss": 6.9903,
       "step": 1053
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.84393063583815e-05,
-      "loss": 6.9792,
+      "learning_rate": 0.0002953179190751445,
+      "loss": 7.1228,
       "step": 1054
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.841040462427746e-05,
-      "loss": 6.7868,
+      "learning_rate": 0.00029523121387283233,
+      "loss": 6.9285,
       "step": 1055
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.838150289017341e-05,
-      "loss": 6.7968,
+      "learning_rate": 0.0002951445086705202,
+      "loss": 6.9491,
       "step": 1056
     },
     {
       "epoch": 1.18,
-      "learning_rate": 9.835260115606936e-05,
-      "loss": 6.961,
+      "learning_rate": 0.0002950578034682081,
+      "loss": 7.07,
       "step": 1057
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.832369942196532e-05,
-      "loss": 6.9985,
+      "learning_rate": 0.0002949710982658959,
+      "loss": 7.1385,
       "step": 1058
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.829479768786127e-05,
-      "loss": 6.7562,
+      "learning_rate": 0.0002948843930635838,
+      "loss": 6.9367,
       "step": 1059
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.826589595375723e-05,
-      "loss": 6.7406,
+      "learning_rate": 0.00029479768786127165,
+      "loss": 6.9371,
       "step": 1060
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.823699421965319e-05,
-      "loss": 6.7705,
+      "learning_rate": 0.0002947109826589595,
+      "loss": 6.9534,
       "step": 1061
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.820809248554915e-05,
-      "loss": 6.8379,
+      "learning_rate": 0.0002946242774566474,
+      "loss": 6.9856,
       "step": 1062
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.81791907514451e-05,
-      "loss": 6.7659,
+      "learning_rate": 0.0002945375722543352,
+      "loss": 6.9414,
       "step": 1063
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.815028901734105e-05,
-      "loss": 6.7486,
+      "learning_rate": 0.0002944508670520231,
+      "loss": 6.9301,
       "step": 1064
     },
     {
       "epoch": 1.19,
-      "learning_rate": 9.812138728323699e-05,
-      "loss": 6.8629,
+      "learning_rate": 0.00029436416184971096,
+      "loss": 7.0531,
       "step": 1065
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.809248554913295e-05,
-      "loss": 6.78,
+      "learning_rate": 0.00029427745664739884,
+      "loss": 6.9637,
       "step": 1066
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.80635838150289e-05,
-      "loss": 6.7193,
+      "learning_rate": 0.00029419075144508666,
+      "loss": 6.8698,
       "step": 1067
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.803468208092485e-05,
-      "loss": 6.7284,
+      "learning_rate": 0.00029410404624277453,
+      "loss": 6.8937,
       "step": 1068
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.800578034682082e-05,
-      "loss": 6.9339,
+      "learning_rate": 0.0002940173410404624,
+      "loss": 7.1158,
       "step": 1069
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.797687861271677e-05,
-      "loss": 6.799,
+      "learning_rate": 0.0002939306358381503,
+      "loss": 7.0062,
       "step": 1070
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.794797687861273e-05,
-      "loss": 6.796,
+      "learning_rate": 0.00029384393063583815,
+      "loss": 6.9955,
       "step": 1071
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.791907514450868e-05,
-      "loss": 6.8769,
+      "learning_rate": 0.00029375722543352597,
+      "loss": 7.0734,
       "step": 1072
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.789017341040463e-05,
-      "loss": 6.7186,
+      "learning_rate": 0.00029367052023121385,
+      "loss": 6.9385,
       "step": 1073
     },
     {
       "epoch": 1.2,
-      "learning_rate": 9.786127167630059e-05,
-      "loss": 6.6335,
+      "learning_rate": 0.0002935838150289017,
+      "loss": 6.8934,
       "step": 1074
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.783236994219653e-05,
-      "loss": 6.6763,
+      "learning_rate": 0.00029349710982658954,
+      "loss": 6.8655,
       "step": 1075
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.780346820809248e-05,
-      "loss": 6.7582,
+      "learning_rate": 0.0002934104046242774,
+      "loss": 6.9261,
       "step": 1076
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.777456647398845e-05,
-      "loss": 6.6409,
+      "learning_rate": 0.0002933236994219653,
+      "loss": 6.8333,
       "step": 1077
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.77456647398844e-05,
-      "loss": 6.8043,
+      "learning_rate": 0.00029323699421965316,
+      "loss": 7.0255,
       "step": 1078
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.771676300578035e-05,
-      "loss": 6.8127,
+      "learning_rate": 0.00029315028901734104,
+      "loss": 6.9974,
       "step": 1079
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.768786127167631e-05,
-      "loss": 6.7294,
+      "learning_rate": 0.00029306358381502886,
+      "loss": 6.9652,
       "step": 1080
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.765895953757226e-05,
-      "loss": 6.7308,
+      "learning_rate": 0.00029297687861271673,
+      "loss": 6.919,
       "step": 1081
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.763005780346821e-05,
-      "loss": 6.7703,
+      "learning_rate": 0.0002928901734104046,
+      "loss": 6.9966,
       "step": 1082
     },
     {
       "epoch": 1.21,
-      "learning_rate": 9.760115606936417e-05,
-      "loss": 6.7657,
+      "learning_rate": 0.0002928034682080925,
+      "loss": 6.9917,
       "step": 1083
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.757225433526012e-05,
-      "loss": 6.649,
+      "learning_rate": 0.0002927167630057803,
+      "loss": 6.8685,
       "step": 1084
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.754335260115607e-05,
-      "loss": 6.6804,
+      "learning_rate": 0.0002926300578034682,
+      "loss": 6.9399,
       "step": 1085
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.751445086705203e-05,
-      "loss": 6.6432,
+      "learning_rate": 0.00029254335260115604,
+      "loss": 6.924,
       "step": 1086
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.748554913294798e-05,
-      "loss": 6.5757,
+      "learning_rate": 0.0002924566473988439,
+      "loss": 6.8392,
       "step": 1087
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.745664739884394e-05,
-      "loss": 6.709,
+      "learning_rate": 0.0002923699421965318,
+      "loss": 7.0557,
       "step": 1088
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.742774566473989e-05,
-      "loss": 6.5361,
+      "learning_rate": 0.0002922832369942196,
+      "loss": 6.8386,
       "step": 1089
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.739884393063584e-05,
-      "loss": 6.3655,
+      "learning_rate": 0.0002921965317919075,
+      "loss": 6.7111,
       "step": 1090
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.73699421965318e-05,
-      "loss": 6.3244,
+      "learning_rate": 0.00029210982658959536,
+      "loss": 6.6707,
       "step": 1091
     },
     {
       "epoch": 1.22,
-      "learning_rate": 9.734104046242775e-05,
-      "loss": 6.0537,
+      "learning_rate": 0.0002920231213872832,
+      "loss": 6.5014,
       "step": 1092
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.731213872832372e-05,
-      "loss": 7.0556,
+      "learning_rate": 0.0002919364161849711,
+      "loss": 7.0419,
       "step": 1093
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.728323699421966e-05,
-      "loss": 6.8468,
+      "learning_rate": 0.0002918497109826589,
+      "loss": 6.9151,
       "step": 1094
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.725433526011561e-05,
-      "loss": 6.9464,
+      "learning_rate": 0.0002917630057803468,
+      "loss": 7.0245,
       "step": 1095
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.722543352601156e-05,
-      "loss": 7.0257,
+      "learning_rate": 0.0002916763005780347,
+      "loss": 7.0573,
       "step": 1096
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.719653179190752e-05,
-      "loss": 6.9944,
+      "learning_rate": 0.0002915895953757225,
+      "loss": 7.0576,
       "step": 1097
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.716763005780347e-05,
-      "loss": 6.8981,
+      "learning_rate": 0.00029150289017341037,
+      "loss": 7.0022,
       "step": 1098
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.713872832369942e-05,
-      "loss": 6.9311,
+      "learning_rate": 0.00029141618497109824,
+      "loss": 7.0864,
       "step": 1099
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.710982658959538e-05,
-      "loss": 7.0534,
+      "learning_rate": 0.0002913294797687861,
+      "loss": 7.1691,
       "step": 1100
     },
     {
       "epoch": 1.23,
-      "learning_rate": 9.708092485549133e-05,
-      "loss": 6.8323,
+      "learning_rate": 0.00029124277456647394,
+      "loss": 6.9565,
       "step": 1101
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.70520231213873e-05,
-      "loss": 6.7655,
+      "learning_rate": 0.00029115606936416186,
+      "loss": 6.8883,
       "step": 1102
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.702312138728325e-05,
-      "loss": 6.6748,
+      "learning_rate": 0.0002910693641618497,
+      "loss": 6.8195,
       "step": 1103
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.699421965317919e-05,
-      "loss": 6.8738,
+      "learning_rate": 0.00029098265895953756,
+      "loss": 7.0411,
       "step": 1104
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.696531791907514e-05,
-      "loss": 6.8083,
+      "learning_rate": 0.00029089595375722543,
+      "loss": 6.9553,
       "step": 1105
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.69364161849711e-05,
-      "loss": 6.7345,
+      "learning_rate": 0.00029080924855491325,
+      "loss": 6.9584,
       "step": 1106
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.690751445086705e-05,
-      "loss": 6.7465,
+      "learning_rate": 0.0002907225433526011,
+      "loss": 6.9157,
       "step": 1107
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.6878612716763e-05,
-      "loss": 6.7133,
+      "learning_rate": 0.000290635838150289,
+      "loss": 6.8947,
       "step": 1108
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.684971098265896e-05,
-      "loss": 6.7263,
+      "learning_rate": 0.0002905491329479768,
+      "loss": 6.8699,
       "step": 1109
     },
     {
       "epoch": 1.24,
-      "learning_rate": 9.682080924855492e-05,
-      "loss": 6.7653,
+      "learning_rate": 0.00029046242774566475,
+      "loss": 6.9365,
       "step": 1110
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.679190751445088e-05,
-      "loss": 6.8678,
+      "learning_rate": 0.00029037572254335257,
+      "loss": 7.0318,
       "step": 1111
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.676300578034683e-05,
-      "loss": 6.7971,
+      "learning_rate": 0.00029028901734104044,
+      "loss": 6.9425,
       "step": 1112
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.673410404624278e-05,
-      "loss": 6.8226,
+      "learning_rate": 0.0002902023121387283,
+      "loss": 7.0044,
       "step": 1113
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.670520231213874e-05,
-      "loss": 6.7696,
+      "learning_rate": 0.00029011560693641613,
+      "loss": 6.973,
       "step": 1114
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.667630057803468e-05,
-      "loss": 6.8388,
+      "learning_rate": 0.000290028901734104,
+      "loss": 7.017,
       "step": 1115
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.664739884393063e-05,
-      "loss": 6.6757,
+      "learning_rate": 0.0002899421965317919,
+      "loss": 6.8651,
       "step": 1116
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.661849710982659e-05,
-      "loss": 6.7663,
+      "learning_rate": 0.0002898554913294797,
+      "loss": 6.9558,
       "step": 1117
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.658959537572255e-05,
-      "loss": 6.8615,
+      "learning_rate": 0.00028976878612716763,
+      "loss": 7.0653,
       "step": 1118
     },
     {
       "epoch": 1.25,
-      "learning_rate": 9.65606936416185e-05,
-      "loss": 6.7429,
+      "learning_rate": 0.00028968208092485545,
+      "loss": 6.9378,
       "step": 1119
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.653179190751446e-05,
-      "loss": 6.727,
+      "learning_rate": 0.0002895953757225433,
+      "loss": 6.9479,
       "step": 1120
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.650289017341041e-05,
-      "loss": 6.802,
+      "learning_rate": 0.0002895086705202312,
+      "loss": 7.0026,
       "step": 1121
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.647398843930637e-05,
-      "loss": 6.8809,
+      "learning_rate": 0.00028942196531791907,
+      "loss": 7.0523,
       "step": 1122
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.644508670520232e-05,
-      "loss": 6.7435,
+      "learning_rate": 0.0002893352601156069,
+      "loss": 6.9269,
       "step": 1123
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.641618497109827e-05,
-      "loss": 6.8051,
+      "learning_rate": 0.00028924855491329476,
+      "loss": 6.9884,
       "step": 1124
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.638728323699421e-05,
-      "loss": 6.7915,
+      "learning_rate": 0.00028916184971098264,
+      "loss": 7.0234,
       "step": 1125
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.635838150289018e-05,
-      "loss": 6.9002,
+      "learning_rate": 0.0002890751445086705,
+      "loss": 7.1123,
       "step": 1126
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.632947976878613e-05,
-      "loss": 6.7489,
+      "learning_rate": 0.0002889884393063584,
+      "loss": 6.9365,
       "step": 1127
     },
     {
       "epoch": 1.26,
-      "learning_rate": 9.630057803468209e-05,
-      "loss": 6.7078,
+      "learning_rate": 0.0002889017341040462,
+      "loss": 6.9098,
       "step": 1128
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.627167630057804e-05,
-      "loss": 6.8439,
+      "learning_rate": 0.0002888150289017341,
+      "loss": 7.0374,
       "step": 1129
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.6242774566474e-05,
-      "loss": 6.7135,
+      "learning_rate": 0.00028872832369942195,
+      "loss": 6.93,
       "step": 1130
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.621387283236995e-05,
-      "loss": 6.7495,
+      "learning_rate": 0.0002886416184971098,
+      "loss": 6.9537,
       "step": 1131
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.61849710982659e-05,
-      "loss": 6.6711,
+      "learning_rate": 0.00028855491329479765,
+      "loss": 6.9154,
       "step": 1132
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.615606936416185e-05,
-      "loss": 6.7818,
+      "learning_rate": 0.0002884682080924855,
+      "loss": 7.0262,
       "step": 1133
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.612716763005781e-05,
-      "loss": 6.8237,
+      "learning_rate": 0.0002883815028901734,
+      "loss": 7.0627,
       "step": 1134
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.609826589595376e-05,
-      "loss": 6.3679,
+      "learning_rate": 0.00028829479768786127,
+      "loss": 6.5888,
       "step": 1135
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.606936416184971e-05,
-      "loss": 6.4992,
+      "learning_rate": 0.0002882080924855491,
+      "loss": 6.787,
       "step": 1136
     },
     {
       "epoch": 1.27,
-      "learning_rate": 9.604046242774567e-05,
-      "loss": 6.7425,
+      "learning_rate": 0.00028812138728323696,
+      "loss": 7.0181,
       "step": 1137
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.601156069364162e-05,
-      "loss": 6.3646,
+      "learning_rate": 0.00028803468208092484,
+      "loss": 6.6948,
       "step": 1138
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.598265895953757e-05,
-      "loss": 6.4325,
+      "learning_rate": 0.0002879479768786127,
+      "loss": 6.7467,
       "step": 1139
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.595375722543353e-05,
-      "loss": 6.4202,
+      "learning_rate": 0.00028786127167630053,
+      "loss": 6.7896,
       "step": 1140
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.592485549132948e-05,
-      "loss": 6.3996,
+      "learning_rate": 0.0002877745664739884,
+      "loss": 6.7665,
       "step": 1141
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.589595375722545e-05,
-      "loss": 6.4195,
+      "learning_rate": 0.0002876878612716763,
+      "loss": 6.8433,
       "step": 1142
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.58670520231214e-05,
-      "loss": 7.0123,
+      "learning_rate": 0.00028760115606936415,
+      "loss": 7.097,
       "step": 1143
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.583815028901734e-05,
-      "loss": 6.7802,
+      "learning_rate": 0.000287514450867052,
+      "loss": 6.8719,
       "step": 1144
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.58092485549133e-05,
-      "loss": 6.7964,
+      "learning_rate": 0.00028742774566473984,
+      "loss": 6.9585,
       "step": 1145
     },
     {
       "epoch": 1.28,
-      "learning_rate": 9.578034682080925e-05,
-      "loss": 6.9757,
+      "learning_rate": 0.0002873410404624277,
+      "loss": 7.084,
       "step": 1146
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.57514450867052e-05,
-      "loss": 6.8304,
+      "learning_rate": 0.0002872543352601156,
+      "loss": 6.966,
       "step": 1147
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.572254335260116e-05,
-      "loss": 6.914,
+      "learning_rate": 0.0002871676300578034,
+      "loss": 7.0419,
       "step": 1148
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.569364161849711e-05,
-      "loss": 6.8427,
+      "learning_rate": 0.0002870809248554913,
+      "loss": 6.9644,
       "step": 1149
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.566473988439308e-05,
-      "loss": 6.8501,
+      "learning_rate": 0.00028699421965317916,
+      "loss": 6.9897,
       "step": 1150
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.563583815028903e-05,
-      "loss": 6.7317,
+      "learning_rate": 0.00028690751445086703,
+      "loss": 6.9256,
       "step": 1151
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.560693641618498e-05,
-      "loss": 6.6758,
+      "learning_rate": 0.0002868208092485549,
+      "loss": 6.8939,
       "step": 1152
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.557803468208094e-05,
-      "loss": 6.8055,
+      "learning_rate": 0.00028673410404624273,
+      "loss": 6.9593,
       "step": 1153
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.554913294797688e-05,
-      "loss": 6.8828,
+      "learning_rate": 0.0002866473988439306,
+      "loss": 7.0732,
       "step": 1154
     },
     {
       "epoch": 1.29,
-      "learning_rate": 9.552023121387283e-05,
-      "loss": 6.8182,
+      "learning_rate": 0.0002865606936416185,
+      "loss": 6.9682,
       "step": 1155
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.549132947976878e-05,
-      "loss": 6.7379,
+      "learning_rate": 0.00028647398843930635,
+      "loss": 6.9555,
       "step": 1156
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.546242774566474e-05,
-      "loss": 6.7678,
+      "learning_rate": 0.00028638728323699417,
+      "loss": 6.968,
       "step": 1157
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.54335260115607e-05,
-      "loss": 7.0244,
+      "learning_rate": 0.0002863005780346821,
+      "loss": 7.1803,
       "step": 1158
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.540462427745666e-05,
-      "loss": 6.8244,
+      "learning_rate": 0.0002862138728323699,
+      "loss": 7.0058,
       "step": 1159
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.537572254335261e-05,
-      "loss": 6.7123,
+      "learning_rate": 0.0002861271676300578,
+      "loss": 6.8984,
       "step": 1160
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.534682080924856e-05,
-      "loss": 6.8363,
+      "learning_rate": 0.00028604046242774566,
+      "loss": 7.042,
       "step": 1161
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.531791907514452e-05,
-      "loss": 6.8558,
+      "learning_rate": 0.0002859537572254335,
+      "loss": 7.029,
       "step": 1162
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.528901734104047e-05,
-      "loss": 6.6985,
+      "learning_rate": 0.00028586705202312136,
+      "loss": 6.8847,
       "step": 1163
     },
     {
       "epoch": 1.3,
-      "learning_rate": 9.526011560693642e-05,
-      "loss": 6.6407,
+      "learning_rate": 0.00028578034682080923,
+      "loss": 6.8427,
       "step": 1164
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.523121387283236e-05,
-      "loss": 6.7582,
+      "learning_rate": 0.00028569364161849705,
+      "loss": 6.9903,
       "step": 1165
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.520231213872833e-05,
-      "loss": 6.612,
+      "learning_rate": 0.000285606936416185,
+      "loss": 6.7861,
       "step": 1166
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.517341040462428e-05,
-      "loss": 6.9583,
+      "learning_rate": 0.0002855202312138728,
+      "loss": 7.1773,
       "step": 1167
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.514450867052024e-05,
-      "loss": 6.815,
+      "learning_rate": 0.0002854335260115607,
+      "loss": 6.9984,
       "step": 1168
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.511560693641619e-05,
-      "loss": 6.9203,
+      "learning_rate": 0.00028534682080924855,
+      "loss": 7.1351,
       "step": 1169
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.508670520231214e-05,
-      "loss": 6.9201,
+      "learning_rate": 0.00028526011560693637,
+      "loss": 7.114,
       "step": 1170
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.50578034682081e-05,
-      "loss": 6.8136,
+      "learning_rate": 0.00028517341040462424,
+      "loss": 6.985,
       "step": 1171
     },
     {
       "epoch": 1.31,
-      "learning_rate": 9.502890173410405e-05,
-      "loss": 6.8682,
+      "learning_rate": 0.0002850867052023121,
+      "loss": 7.0578,
       "step": 1172
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.5e-05,
-      "loss": 6.7984,
+      "learning_rate": 0.000285,
+      "loss": 7.0342,
       "step": 1173
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.497109826589596e-05,
-      "loss": 6.7778,
+      "learning_rate": 0.00028491329479768786,
+      "loss": 6.9936,
       "step": 1174
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.494219653179191e-05,
-      "loss": 6.7791,
+      "learning_rate": 0.00028482658959537574,
+      "loss": 6.9926,
       "step": 1175
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.491329479768787e-05,
-      "loss": 6.8255,
+      "learning_rate": 0.00028473988439306356,
+      "loss": 7.0866,
       "step": 1176
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.488439306358382e-05,
-      "loss": 6.8335,
+      "learning_rate": 0.00028465317919075143,
+      "loss": 7.0623,
       "step": 1177
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.485549132947977e-05,
-      "loss": 6.7124,
+      "learning_rate": 0.0002845664739884393,
+      "loss": 6.9769,
       "step": 1178
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.482658959537573e-05,
-      "loss": 6.8192,
+      "learning_rate": 0.0002844797687861271,
+      "loss": 7.0429,
       "step": 1179
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.479768786127168e-05,
-      "loss": 6.6875,
+      "learning_rate": 0.000284393063583815,
+      "loss": 6.9067,
       "step": 1180
     },
     {
       "epoch": 1.32,
-      "learning_rate": 9.476878612716763e-05,
-      "loss": 6.7327,
+      "learning_rate": 0.00028430635838150287,
+      "loss": 6.9767,
       "step": 1181
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.47398843930636e-05,
-      "loss": 6.8375,
+      "learning_rate": 0.00028421965317919075,
+      "loss": 7.1155,
       "step": 1182
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.471098265895954e-05,
-      "loss": 6.8496,
+      "learning_rate": 0.0002841329479768786,
+      "loss": 7.1142,
       "step": 1183
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.468208092485549e-05,
-      "loss": 6.7939,
+      "learning_rate": 0.00028404624277456644,
+      "loss": 7.0375,
       "step": 1184
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.465317919075145e-05,
-      "loss": 6.5131,
+      "learning_rate": 0.0002839595375722543,
+      "loss": 6.841,
       "step": 1185
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.46242774566474e-05,
-      "loss": 6.4426,
+      "learning_rate": 0.0002838728323699422,
+      "loss": 6.7433,
       "step": 1186
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.459537572254335e-05,
-      "loss": 6.4412,
+      "learning_rate": 0.00028378612716763,
+      "loss": 6.7174,
       "step": 1187
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.456647398843931e-05,
-      "loss": 6.5164,
+      "learning_rate": 0.0002836994219653179,
+      "loss": 6.8676,
       "step": 1188
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.453757225433526e-05,
-      "loss": 6.1505,
+      "learning_rate": 0.00028361271676300575,
+      "loss": 6.5716,
       "step": 1189
     },
     {
       "epoch": 1.33,
-      "learning_rate": 9.450867052023123e-05,
-      "loss": 6.1173,
+      "learning_rate": 0.00028352601156069363,
+      "loss": 6.5459,
       "step": 1190
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.447976878612718e-05,
-      "loss": 6.3324,
+      "learning_rate": 0.0002834393063583815,
+      "loss": 6.8077,
       "step": 1191
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.445086705202313e-05,
-      "loss": 5.91,
+      "learning_rate": 0.0002833526011560693,
+      "loss": 6.4026,
       "step": 1192
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.442196531791909e-05,
-      "loss": 7.0624,
+      "learning_rate": 0.0002832658959537572,
+      "loss": 7.118,
       "step": 1193
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.439306358381503e-05,
-      "loss": 7.0578,
+      "learning_rate": 0.00028317919075144507,
+      "loss": 7.1025,
       "step": 1194
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.436416184971098e-05,
-      "loss": 6.9605,
+      "learning_rate": 0.00028309248554913294,
+      "loss": 6.998,
       "step": 1195
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.433526011560693e-05,
-      "loss": 7.0197,
+      "learning_rate": 0.00028300578034682076,
+      "loss": 7.0846,
       "step": 1196
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.430635838150289e-05,
-      "loss": 6.9692,
+      "learning_rate": 0.00028291907514450864,
+      "loss": 7.0731,
       "step": 1197
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.427745664739884e-05,
-      "loss": 6.8511,
+      "learning_rate": 0.0002828323699421965,
+      "loss": 6.9913,
       "step": 1198
     },
     {
       "epoch": 1.34,
-      "learning_rate": 9.424855491329481e-05,
-      "loss": 6.9713,
+      "learning_rate": 0.0002827456647398844,
+      "loss": 7.0561,
       "step": 1199
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.421965317919076e-05,
-      "loss": 6.933,
+      "learning_rate": 0.00028265895953757226,
+      "loss": 7.0612,
       "step": 1200
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.419075144508672e-05,
-      "loss": 6.8514,
+      "learning_rate": 0.0002825722543352601,
+      "loss": 6.9925,
       "step": 1201
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.416184971098267e-05,
-      "loss": 6.8906,
+      "learning_rate": 0.00028248554913294795,
+      "loss": 7.0225,
       "step": 1202
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.413294797687862e-05,
-      "loss": 6.817,
+      "learning_rate": 0.0002823988439306358,
+      "loss": 6.9911,
       "step": 1203
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.410404624277456e-05,
-      "loss": 6.781,
+      "learning_rate": 0.00028231213872832365,
+      "loss": 6.9188,
       "step": 1204
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.407514450867052e-05,
-      "loss": 6.761,
+      "learning_rate": 0.0002822254335260115,
+      "loss": 6.9353,
       "step": 1205
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.404624277456647e-05,
-      "loss": 6.7112,
+      "learning_rate": 0.0002821387283236994,
+      "loss": 6.897,
       "step": 1206
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.401734104046244e-05,
-      "loss": 6.7875,
+      "learning_rate": 0.00028205202312138727,
+      "loss": 6.9758,
       "step": 1207
     },
     {
       "epoch": 1.35,
-      "learning_rate": 9.398843930635839e-05,
-      "loss": 6.7841,
+      "learning_rate": 0.00028196531791907514,
+      "loss": 6.9944,
       "step": 1208
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.395953757225434e-05,
-      "loss": 6.752,
+      "learning_rate": 0.00028187861271676296,
+      "loss": 6.8819,
       "step": 1209
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.39306358381503e-05,
-      "loss": 6.8957,
+      "learning_rate": 0.00028179190751445083,
+      "loss": 7.0381,
       "step": 1210
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.390173410404625e-05,
-      "loss": 6.8899,
+      "learning_rate": 0.0002817052023121387,
+      "loss": 7.0779,
       "step": 1211
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.38728323699422e-05,
-      "loss": 6.8648,
+      "learning_rate": 0.0002816184971098266,
+      "loss": 6.9963,
       "step": 1212
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.384393063583816e-05,
-      "loss": 6.8806,
+      "learning_rate": 0.0002815317919075144,
+      "loss": 7.0521,
       "step": 1213
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.38150289017341e-05,
-      "loss": 6.7568,
+      "learning_rate": 0.0002814450867052023,
+      "loss": 7.0034,
       "step": 1214
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.378612716763006e-05,
-      "loss": 6.9281,
+      "learning_rate": 0.00028135838150289015,
+      "loss": 7.0778,
       "step": 1215
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.375722543352602e-05,
-      "loss": 6.7034,
+      "learning_rate": 0.000281271676300578,
+      "loss": 6.9147,
       "step": 1216
     },
     {
       "epoch": 1.36,
-      "learning_rate": 9.372832369942197e-05,
-      "loss": 6.7528,
+      "learning_rate": 0.0002811849710982659,
+      "loss": 6.9434,
       "step": 1217
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.369942196531792e-05,
-      "loss": 6.8896,
+      "learning_rate": 0.0002810982658959537,
+      "loss": 7.0471,
       "step": 1218
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.367052023121388e-05,
-      "loss": 6.7403,
+      "learning_rate": 0.0002810115606936416,
+      "loss": 6.911,
       "step": 1219
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.364161849710983e-05,
-      "loss": 6.8054,
+      "learning_rate": 0.00028092485549132947,
+      "loss": 6.9871,
       "step": 1220
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.361271676300578e-05,
-      "loss": 6.798,
+      "learning_rate": 0.0002808381502890173,
+      "loss": 7.0308,
       "step": 1221
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.358381502890174e-05,
-      "loss": 6.6997,
+      "learning_rate": 0.00028075144508670516,
+      "loss": 6.9191,
       "step": 1222
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.355491329479769e-05,
-      "loss": 6.7324,
+      "learning_rate": 0.00028066473988439303,
+      "loss": 6.9027,
       "step": 1223
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.352601156069364e-05,
-      "loss": 6.7915,
+      "learning_rate": 0.0002805780346820809,
+      "loss": 7.0065,
       "step": 1224
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.34971098265896e-05,
-      "loss": 6.8616,
+      "learning_rate": 0.0002804913294797688,
+      "loss": 7.0332,
       "step": 1225
     },
     {
       "epoch": 1.37,
-      "learning_rate": 9.346820809248555e-05,
-      "loss": 6.8569,
+      "learning_rate": 0.0002804046242774566,
+      "loss": 7.0724,
       "step": 1226
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.34393063583815e-05,
-      "loss": 6.8253,
+      "learning_rate": 0.0002803179190751445,
+      "loss": 7.0253,
       "step": 1227
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.341040462427746e-05,
-      "loss": 6.7892,
+      "learning_rate": 0.00028023121387283235,
+      "loss": 6.9992,
       "step": 1228
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.338150289017341e-05,
-      "loss": 6.8306,
+      "learning_rate": 0.0002801445086705202,
+      "loss": 7.029,
       "step": 1229
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.335260115606937e-05,
-      "loss": 6.6154,
+      "learning_rate": 0.00028005780346820804,
+      "loss": 6.848,
       "step": 1230
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.332369942196533e-05,
-      "loss": 6.6423,
+      "learning_rate": 0.00027997109826589597,
+      "loss": 6.8728,
       "step": 1231
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.329479768786129e-05,
-      "loss": 6.7778,
+      "learning_rate": 0.0002798843930635838,
+      "loss": 7.0106,
       "step": 1232
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.326589595375723e-05,
-      "loss": 6.7842,
+      "learning_rate": 0.00027979768786127166,
+      "loss": 6.9713,
       "step": 1233
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.323699421965318e-05,
-      "loss": 6.8109,
+      "learning_rate": 0.00027971098265895954,
+      "loss": 7.0476,
       "step": 1234
     },
     {
       "epoch": 1.38,
-      "learning_rate": 9.320809248554913e-05,
-      "loss": 6.7303,
+      "learning_rate": 0.00027962427745664736,
+      "loss": 6.9746,
       "step": 1235
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.317919075144509e-05,
-      "loss": 6.6306,
+      "learning_rate": 0.00027953757225433523,
+      "loss": 6.9131,
       "step": 1236
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.315028901734104e-05,
-      "loss": 6.5929,
+      "learning_rate": 0.0002794508670520231,
+      "loss": 6.8417,
       "step": 1237
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.312138728323699e-05,
-      "loss": 6.5594,
+      "learning_rate": 0.0002793641618497109,
+      "loss": 6.8977,
       "step": 1238
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.309248554913296e-05,
-      "loss": 6.3825,
+      "learning_rate": 0.00027927745664739885,
+      "loss": 6.6746,
       "step": 1239
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.306358381502891e-05,
-      "loss": 6.3209,
+      "learning_rate": 0.00027919075144508667,
+      "loss": 6.7429,
       "step": 1240
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.303468208092487e-05,
-      "loss": 6.1835,
+      "learning_rate": 0.00027910404624277455,
+      "loss": 6.5424,
       "step": 1241
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.300578034682082e-05,
-      "loss": 6.1078,
+      "learning_rate": 0.0002790173410404624,
+      "loss": 6.5673,
       "step": 1242
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.297687861271676e-05,
-      "loss": 7.0026,
+      "learning_rate": 0.00027893063583815024,
+      "loss": 7.1688,
       "step": 1243
     },
     {
       "epoch": 1.39,
-      "learning_rate": 9.294797687861271e-05,
-      "loss": 6.8092,
+      "learning_rate": 0.0002788439306358381,
+      "loss": 6.9321,
       "step": 1244
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.291907514450867e-05,
-      "loss": 6.9385,
+      "learning_rate": 0.000278757225433526,
+      "loss": 7.0772,
       "step": 1245
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.289017341040462e-05,
-      "loss": 6.7991,
+      "learning_rate": 0.00027867052023121386,
+      "loss": 6.983,
       "step": 1246
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.286127167630059e-05,
-      "loss": 6.8815,
+      "learning_rate": 0.00027858381502890174,
+      "loss": 7.0144,
       "step": 1247
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.283236994219654e-05,
-      "loss": 6.8727,
+      "learning_rate": 0.00027849710982658955,
+      "loss": 7.0018,
       "step": 1248
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.28034682080925e-05,
-      "loss": 6.8667,
+      "learning_rate": 0.00027841040462427743,
+      "loss": 6.993,
       "step": 1249
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.277456647398845e-05,
-      "loss": 6.8564,
+      "learning_rate": 0.0002783236994219653,
+      "loss": 6.9988,
       "step": 1250
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.27456647398844e-05,
-      "loss": 6.7964,
+      "learning_rate": 0.0002782369942196532,
+      "loss": 6.9834,
       "step": 1251
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.271676300578035e-05,
-      "loss": 6.61,
+      "learning_rate": 0.000278150289017341,
+      "loss": 6.8022,
       "step": 1252
     },
     {
       "epoch": 1.4,
-      "learning_rate": 9.268786127167631e-05,
-      "loss": 6.5616,
+      "learning_rate": 0.00027806358381502887,
+      "loss": 6.7716,
       "step": 1253
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.265895953757225e-05,
-      "loss": 6.8845,
+      "learning_rate": 0.00027797687861271674,
+      "loss": 7.0998,
       "step": 1254
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.263005780346821e-05,
-      "loss": 6.7727,
+      "learning_rate": 0.0002778901734104046,
+      "loss": 6.9524,
       "step": 1255
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.260115606936417e-05,
-      "loss": 6.6488,
+      "learning_rate": 0.0002778034682080925,
+      "loss": 6.8791,
       "step": 1256
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.257225433526012e-05,
-      "loss": 6.7033,
+      "learning_rate": 0.0002777167630057803,
+      "loss": 6.8933,
       "step": 1257
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.254335260115608e-05,
-      "loss": 6.8093,
+      "learning_rate": 0.0002776300578034682,
+      "loss": 7.0075,
       "step": 1258
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.251445086705203e-05,
-      "loss": 6.7659,
+      "learning_rate": 0.00027754335260115606,
+      "loss": 6.9637,
       "step": 1259
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.248554913294798e-05,
-      "loss": 6.7379,
+      "learning_rate": 0.0002774566473988439,
+      "loss": 6.9448,
       "step": 1260
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.245664739884394e-05,
-      "loss": 6.6826,
+      "learning_rate": 0.00027736994219653175,
+      "loss": 6.941,
       "step": 1261
     },
     {
       "epoch": 1.41,
-      "learning_rate": 9.242774566473989e-05,
-      "loss": 6.8953,
+      "learning_rate": 0.0002772832369942196,
+      "loss": 7.091,
       "step": 1262
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.239884393063584e-05,
-      "loss": 6.5806,
+      "learning_rate": 0.0002771965317919075,
+      "loss": 6.8212,
       "step": 1263
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.23699421965318e-05,
-      "loss": 6.7554,
+      "learning_rate": 0.0002771098265895954,
+      "loss": 6.9691,
       "step": 1264
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.234104046242775e-05,
-      "loss": 6.6997,
+      "learning_rate": 0.0002770231213872832,
+      "loss": 6.9212,
       "step": 1265
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.23121387283237e-05,
-      "loss": 6.6562,
+      "learning_rate": 0.00027693641618497107,
+      "loss": 6.9143,
       "step": 1266
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.228323699421966e-05,
-      "loss": 6.8225,
+      "learning_rate": 0.00027684971098265894,
+      "loss": 7.0751,
       "step": 1267
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.225433526011561e-05,
-      "loss": 6.7031,
+      "learning_rate": 0.0002767630057803468,
+      "loss": 6.9269,
       "step": 1268
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.222543352601156e-05,
-      "loss": 6.6924,
+      "learning_rate": 0.00027667630057803464,
+      "loss": 6.9317,
       "step": 1269
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.219653179190752e-05,
-      "loss": 6.7957,
+      "learning_rate": 0.0002765895953757225,
+      "loss": 7.0037,
       "step": 1270
     },
     {
       "epoch": 1.42,
-      "learning_rate": 9.216763005780348e-05,
-      "loss": 6.6207,
+      "learning_rate": 0.0002765028901734104,
+      "loss": 6.8466,
       "step": 1271
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.213872832369944e-05,
-      "loss": 6.6815,
+      "learning_rate": 0.00027641618497109826,
+      "loss": 6.9358,
       "step": 1272
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.210982658959538e-05,
-      "loss": 6.6672,
+      "learning_rate": 0.00027632947976878613,
+      "loss": 6.9001,
       "step": 1273
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.208092485549133e-05,
-      "loss": 6.6621,
+      "learning_rate": 0.00027624277456647395,
+      "loss": 6.9394,
       "step": 1274
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.205202312138728e-05,
-      "loss": 6.7405,
+      "learning_rate": 0.0002761560693641618,
+      "loss": 6.9867,
       "step": 1275
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.202312138728324e-05,
-      "loss": 6.6532,
+      "learning_rate": 0.0002760693641618497,
+      "loss": 6.9211,
       "step": 1276
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.199421965317919e-05,
-      "loss": 6.9357,
+      "learning_rate": 0.0002759826589595375,
+      "loss": 7.2167,
       "step": 1277
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.196531791907514e-05,
-      "loss": 6.6429,
+      "learning_rate": 0.0002758959537572254,
+      "loss": 6.9552,
       "step": 1278
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.193641618497111e-05,
-      "loss": 6.8137,
+      "learning_rate": 0.00027580924855491327,
+      "loss": 7.0684,
       "step": 1279
     },
     {
       "epoch": 1.43,
-      "learning_rate": 9.190751445086706e-05,
-      "loss": 6.5833,
+      "learning_rate": 0.00027572254335260114,
+      "loss": 6.9158,
       "step": 1280
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.187861271676302e-05,
-      "loss": 6.6832,
+      "learning_rate": 0.000275635838150289,
+      "loss": 6.9651,
       "step": 1281
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.184971098265897e-05,
-      "loss": 6.6882,
+      "learning_rate": 0.00027554913294797683,
+      "loss": 6.9407,
       "step": 1282
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.182080924855491e-05,
-      "loss": 6.7417,
+      "learning_rate": 0.0002754624277456647,
+      "loss": 7.0128,
       "step": 1283
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.179190751445086e-05,
-      "loss": 6.5643,
+      "learning_rate": 0.0002753757225433526,
+      "loss": 6.8905,
       "step": 1284
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.176300578034682e-05,
-      "loss": 6.6627,
+      "learning_rate": 0.00027528901734104045,
+      "loss": 6.9704,
       "step": 1285
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.173410404624277e-05,
-      "loss": 6.7024,
+      "learning_rate": 0.0002752023121387283,
+      "loss": 7.0173,
       "step": 1286
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.170520231213874e-05,
-      "loss": 6.3669,
+      "learning_rate": 0.0002751156069364162,
+      "loss": 6.7661,
       "step": 1287
     },
     {
       "epoch": 1.44,
-      "learning_rate": 9.167630057803469e-05,
-      "loss": 6.6679,
+      "learning_rate": 0.000275028901734104,
+      "loss": 6.9857,
       "step": 1288
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.164739884393065e-05,
-      "loss": 6.5965,
+      "learning_rate": 0.0002749421965317919,
+      "loss": 6.9989,
       "step": 1289
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.16184971098266e-05,
-      "loss": 6.322,
+      "learning_rate": 0.00027485549132947977,
+      "loss": 6.7521,
       "step": 1290
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.158959537572255e-05,
-      "loss": 6.4638,
+      "learning_rate": 0.0002747687861271676,
+      "loss": 7.0413,
       "step": 1291
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.15606936416185e-05,
-      "loss": 6.2358,
+      "learning_rate": 0.00027468208092485546,
+      "loss": 6.8337,
       "step": 1292
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.153179190751445e-05,
-      "loss": 6.8944,
+      "learning_rate": 0.00027459537572254334,
+      "loss": 7.0638,
       "step": 1293
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.15028901734104e-05,
-      "loss": 6.9028,
+      "learning_rate": 0.00027450867052023116,
+      "loss": 7.0662,
       "step": 1294
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.147398843930635e-05,
-      "loss": 6.8633,
+      "learning_rate": 0.00027442196531791903,
+      "loss": 7.0263,
       "step": 1295
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.144508670520232e-05,
-      "loss": 6.7898,
+      "learning_rate": 0.0002743352601156069,
+      "loss": 6.9727,
       "step": 1296
     },
     {
       "epoch": 1.45,
-      "learning_rate": 9.141618497109827e-05,
-      "loss": 6.7085,
+      "learning_rate": 0.0002742485549132948,
+      "loss": 6.8901,
       "step": 1297
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.138728323699423e-05,
-      "loss": 6.8277,
+      "learning_rate": 0.00027416184971098265,
+      "loss": 7.0417,
       "step": 1298
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.135838150289018e-05,
-      "loss": 6.7111,
+      "learning_rate": 0.00027407514450867047,
+      "loss": 6.906,
       "step": 1299
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.132947976878613e-05,
-      "loss": 6.9226,
+      "learning_rate": 0.00027398843930635835,
+      "loss": 7.1088,
       "step": 1300
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.130057803468209e-05,
-      "loss": 6.9142,
+      "learning_rate": 0.0002739017341040462,
+      "loss": 7.0997,
       "step": 1301
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.127167630057804e-05,
-      "loss": 6.6968,
+      "learning_rate": 0.0002738150289017341,
+      "loss": 6.9273,
       "step": 1302
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.1242774566474e-05,
-      "loss": 6.7351,
+      "learning_rate": 0.0002737283236994219,
+      "loss": 6.9724,
       "step": 1303
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.121387283236995e-05,
-      "loss": 6.7411,
+      "learning_rate": 0.00027364161849710984,
+      "loss": 6.9539,
       "step": 1304
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.11849710982659e-05,
-      "loss": 6.7889,
+      "learning_rate": 0.00027355491329479766,
+      "loss": 6.9659,
       "step": 1305
     },
     {
       "epoch": 1.46,
-      "learning_rate": 9.115606936416185e-05,
-      "loss": 6.7902,
+      "learning_rate": 0.00027346820809248554,
+      "loss": 7.0259,
       "step": 1306
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.112716763005781e-05,
-      "loss": 6.6447,
+      "learning_rate": 0.0002733815028901734,
+      "loss": 6.868,
       "step": 1307
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.109826589595376e-05,
-      "loss": 6.8882,
+      "learning_rate": 0.00027329479768786123,
+      "loss": 7.0617,
       "step": 1308
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.106936416184971e-05,
-      "loss": 6.6744,
+      "learning_rate": 0.0002732080924855491,
+      "loss": 6.8809,
       "step": 1309
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.104046242774567e-05,
-      "loss": 6.6745,
+      "learning_rate": 0.000273121387283237,
+      "loss": 6.9181,
       "step": 1310
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.101156069364162e-05,
-      "loss": 6.7324,
+      "learning_rate": 0.0002730346820809248,
+      "loss": 6.9522,
       "step": 1311
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.098265895953757e-05,
-      "loss": 6.8473,
+      "learning_rate": 0.0002729479768786127,
+      "loss": 7.0577,
       "step": 1312
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.095375722543353e-05,
-      "loss": 6.6559,
+      "learning_rate": 0.00027286127167630054,
+      "loss": 6.8543,
       "step": 1313
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.092485549132948e-05,
-      "loss": 6.7266,
+      "learning_rate": 0.0002727745664739884,
+      "loss": 6.9432,
       "step": 1314
     },
     {
       "epoch": 1.47,
-      "learning_rate": 9.089595375722544e-05,
-      "loss": 6.7583,
+      "learning_rate": 0.0002726878612716763,
+      "loss": 6.9966,
       "step": 1315
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.086705202312139e-05,
-      "loss": 6.6478,
+      "learning_rate": 0.0002726011560693641,
+      "loss": 6.907,
       "step": 1316
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.083815028901734e-05,
-      "loss": 6.7939,
+      "learning_rate": 0.000272514450867052,
+      "loss": 6.9958,
       "step": 1317
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.08092485549133e-05,
-      "loss": 6.8075,
+      "learning_rate": 0.00027242774566473986,
+      "loss": 7.0521,
       "step": 1318
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.078034682080925e-05,
-      "loss": 6.8408,
+      "learning_rate": 0.0002723410404624277,
+      "loss": 7.0987,
       "step": 1319
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.075144508670522e-05,
-      "loss": 6.8165,
+      "learning_rate": 0.0002722543352601156,
+      "loss": 7.0385,
       "step": 1320
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.072254335260117e-05,
-      "loss": 6.6541,
+      "learning_rate": 0.00027216763005780343,
+      "loss": 6.9171,
       "step": 1321
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.069364161849711e-05,
-      "loss": 6.7952,
+      "learning_rate": 0.0002720809248554913,
+      "loss": 7.0702,
       "step": 1322
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.066473988439306e-05,
-      "loss": 6.7428,
+      "learning_rate": 0.0002719942196531792,
+      "loss": 6.9908,
       "step": 1323
     },
     {
       "epoch": 1.48,
-      "learning_rate": 9.063583815028902e-05,
-      "loss": 6.7134,
+      "learning_rate": 0.00027190751445086705,
+      "loss": 6.9608,
       "step": 1324
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.060693641618497e-05,
-      "loss": 6.7914,
+      "learning_rate": 0.00027182080924855487,
+      "loss": 7.0523,
       "step": 1325
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.057803468208092e-05,
-      "loss": 6.7064,
+      "learning_rate": 0.00027173410404624274,
+      "loss": 6.9785,
       "step": 1326
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.054913294797688e-05,
-      "loss": 6.8053,
+      "learning_rate": 0.0002716473988439306,
+      "loss": 7.0457,
       "step": 1327
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.052023121387284e-05,
-      "loss": 6.8037,
+      "learning_rate": 0.0002715606936416185,
+      "loss": 7.0444,
       "step": 1328
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.04913294797688e-05,
-      "loss": 6.8797,
+      "learning_rate": 0.00027147398843930636,
+      "loss": 7.1682,
       "step": 1329
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.046242774566475e-05,
-      "loss": 6.7811,
+      "learning_rate": 0.0002713872832369942,
+      "loss": 7.0354,
       "step": 1330
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.04335260115607e-05,
-      "loss": 6.8481,
+      "learning_rate": 0.00027130057803468206,
+      "loss": 7.1295,
       "step": 1331
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.040462427745666e-05,
-      "loss": 6.6852,
+      "learning_rate": 0.00027121387283236993,
+      "loss": 6.9691,
       "step": 1332
     },
     {
       "epoch": 1.49,
-      "learning_rate": 9.03757225433526e-05,
-      "loss": 6.6757,
+      "learning_rate": 0.00027112716763005775,
+      "loss": 6.9992,
       "step": 1333
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.034682080924855e-05,
-      "loss": 6.6631,
+      "learning_rate": 0.0002710404624277456,
+      "loss": 7.0056,
       "step": 1334
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.03179190751445e-05,
-      "loss": 6.6376,
+      "learning_rate": 0.0002709537572254335,
+      "loss": 6.9519,
       "step": 1335
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.028901734104047e-05,
-      "loss": 6.5893,
+      "learning_rate": 0.0002708670520231214,
+      "loss": 6.996,
       "step": 1336
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.026011560693642e-05,
-      "loss": 6.5568,
+      "learning_rate": 0.00027078034682080925,
+      "loss": 6.9799,
       "step": 1337
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.023121387283238e-05,
-      "loss": 6.503,
+      "learning_rate": 0.00027069364161849707,
+      "loss": 6.9062,
       "step": 1338
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.020231213872833e-05,
-      "loss": 6.3305,
+      "learning_rate": 0.00027060693641618494,
+      "loss": 6.7702,
       "step": 1339
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.017341040462428e-05,
-      "loss": 6.3587,
+      "learning_rate": 0.0002705202312138728,
+      "loss": 6.8806,
       "step": 1340
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.014450867052024e-05,
-      "loss": 6.17,
+      "learning_rate": 0.0002704335260115607,
+      "loss": 6.704,
       "step": 1341
     },
     {
       "epoch": 1.5,
-      "learning_rate": 9.011560693641619e-05,
-      "loss": 5.6732,
+      "learning_rate": 0.0002703468208092485,
+      "loss": 6.3873,
       "step": 1342
     },
     {
       "epoch": 1.51,
-      "learning_rate": 9.008670520231213e-05,
-      "loss": 7.1315,
+      "learning_rate": 0.0002702601156069364,
+      "loss": 7.2563,
       "step": 1343
     },
     {
       "epoch": 1.51,
-      "learning_rate": 9.00578034682081e-05,
-      "loss": 6.7074,
+      "learning_rate": 0.00027017341040462426,
+      "loss": 6.8698,
       "step": 1344
     },
     {
       "epoch": 1.51,
-      "learning_rate": 9.002890173410405e-05,
-      "loss": 6.8734,
+      "learning_rate": 0.00027008670520231213,
+      "loss": 7.0548,
       "step": 1345
     },
     {
       "epoch": 1.51,
-      "learning_rate": 9e-05,
-      "loss": 6.7216,
+      "learning_rate": 0.00027,
+      "loss": 6.9819,
       "step": 1346
     },
     {
       "epoch": 1.51,
-      "learning_rate": 8.997109826589596e-05,
-      "loss": 6.7835,
+      "learning_rate": 0.0002699132947976878,
+      "loss": 6.9929,
       "step": 1347
     },
     {
       "epoch": 1.51,
-      "learning_rate": 8.994219653179191e-05,
-      "loss": 6.8531,
+      "learning_rate": 0.0002698265895953757,
+      "loss": 7.0542,
       "step": 1348
     },
     {
       "epoch": 1.51,
-      "learning_rate": 8.991329479768787e-05,
-      "loss": 6.7352,
+      "learning_rate": 0.00026973988439306357,
+      "loss": 6.974,
       "step": 1349
     },
     {
       "epoch": 1.51,
-      "learning_rate": 8.988439306358382e-05,
-      "loss": 6.7006,
+      "learning_rate": 0.0002696531791907514,
+      "loss": 6.9218,
       "step": 1350
     },
     {
       "epoch": 1.51,
-      "learning_rate": 8.985549132947977e-05,
-      "loss": 6.8023,
+      "learning_rate": 0.00026956647398843926,
+      "loss": 6.9982,
       "step": 1351
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.982658959537573e-05,
-      "loss": 6.704,
+      "learning_rate": 0.00026947976878612714,
+      "loss": 6.9549,
       "step": 1352
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.979768786127168e-05,
-      "loss": 6.7159,
+      "learning_rate": 0.000269393063583815,
+      "loss": 6.9479,
       "step": 1353
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.976878612716763e-05,
-      "loss": 6.7931,
+      "learning_rate": 0.0002693063583815029,
+      "loss": 7.0326,
       "step": 1354
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.973988439306359e-05,
-      "loss": 6.7173,
+      "learning_rate": 0.0002692196531791907,
+      "loss": 6.9664,
       "step": 1355
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.971098265895954e-05,
-      "loss": 6.7739,
+      "learning_rate": 0.0002691329479768786,
+      "loss": 7.0377,
       "step": 1356
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.96820809248555e-05,
-      "loss": 6.7204,
+      "learning_rate": 0.00026904624277456645,
+      "loss": 6.9727,
       "step": 1357
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.965317919075145e-05,
-      "loss": 6.8273,
+      "learning_rate": 0.00026895953757225433,
+      "loss": 7.05,
       "step": 1358
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.96242774566474e-05,
-      "loss": 6.6448,
+      "learning_rate": 0.00026887283236994215,
+      "loss": 6.8841,
       "step": 1359
     },
     {
       "epoch": 1.52,
-      "learning_rate": 8.959537572254337e-05,
-      "loss": 6.5788,
+      "learning_rate": 0.0002687861271676301,
+      "loss": 6.898,
       "step": 1360
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.956647398843932e-05,
-      "loss": 6.4342,
+      "learning_rate": 0.0002686994219653179,
+      "loss": 6.7405,
       "step": 1361
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.953757225433526e-05,
-      "loss": 6.8381,
+      "learning_rate": 0.00026861271676300577,
+      "loss": 7.0831,
       "step": 1362
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.950867052023121e-05,
-      "loss": 6.6486,
+      "learning_rate": 0.00026852601156069364,
+      "loss": 6.9367,
       "step": 1363
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.947976878612717e-05,
-      "loss": 6.7101,
+      "learning_rate": 0.00026843930635838146,
+      "loss": 6.9968,
       "step": 1364
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.945086705202312e-05,
-      "loss": 6.773,
+      "learning_rate": 0.00026835260115606934,
+      "loss": 7.0489,
       "step": 1365
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.942196531791907e-05,
-      "loss": 6.829,
+      "learning_rate": 0.0002682658959537572,
+      "loss": 7.0744,
       "step": 1366
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.939306358381503e-05,
-      "loss": 6.6826,
+      "learning_rate": 0.00026817919075144503,
+      "loss": 6.9516,
       "step": 1367
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.9364161849711e-05,
-      "loss": 6.7347,
+      "learning_rate": 0.00026809248554913296,
+      "loss": 7.0033,
       "step": 1368
     },
     {
       "epoch": 1.53,
-      "learning_rate": 8.933526011560695e-05,
-      "loss": 6.6233,
+      "learning_rate": 0.0002680057803468208,
+      "loss": 6.9089,
       "step": 1369
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.93063583815029e-05,
-      "loss": 6.635,
+      "learning_rate": 0.00026791907514450865,
+      "loss": 6.9339,
       "step": 1370
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.927745664739886e-05,
-      "loss": 6.7332,
+      "learning_rate": 0.0002678323699421965,
+      "loss": 6.9986,
       "step": 1371
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.92485549132948e-05,
-      "loss": 6.721,
+      "learning_rate": 0.00026774566473988435,
+      "loss": 7.0025,
       "step": 1372
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.921965317919075e-05,
-      "loss": 6.7175,
+      "learning_rate": 0.0002676589595375722,
+      "loss": 7.0003,
       "step": 1373
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.91907514450867e-05,
-      "loss": 6.6046,
+      "learning_rate": 0.0002675722543352601,
+      "loss": 6.8533,
       "step": 1374
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.916184971098266e-05,
-      "loss": 6.7429,
+      "learning_rate": 0.00026748554913294797,
+      "loss": 6.9581,
       "step": 1375
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.913294797687862e-05,
-      "loss": 6.691,
+      "learning_rate": 0.00026739884393063584,
+      "loss": 6.9366,
       "step": 1376
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.910404624277458e-05,
-      "loss": 6.4398,
+      "learning_rate": 0.0002673121387283237,
+      "loss": 6.7786,
       "step": 1377
     },
     {
       "epoch": 1.54,
-      "learning_rate": 8.907514450867053e-05,
-      "loss": 6.6628,
+      "learning_rate": 0.00026722543352601153,
+      "loss": 6.9563,
       "step": 1378
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.904624277456648e-05,
-      "loss": 6.7791,
+      "learning_rate": 0.0002671387283236994,
+      "loss": 7.072,
       "step": 1379
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.901734104046244e-05,
-      "loss": 6.8101,
+      "learning_rate": 0.0002670520231213873,
+      "loss": 7.1018,
       "step": 1380
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.898843930635839e-05,
-      "loss": 6.6784,
+      "learning_rate": 0.0002669653179190751,
+      "loss": 6.9861,
       "step": 1381
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.895953757225434e-05,
-      "loss": 6.725,
+      "learning_rate": 0.000266878612716763,
+      "loss": 7.0669,
       "step": 1382
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.893063583815028e-05,
-      "loss": 6.5534,
+      "learning_rate": 0.00026679190751445085,
+      "loss": 6.9009,
       "step": 1383
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.890173410404625e-05,
-      "loss": 6.5428,
+      "learning_rate": 0.0002667052023121387,
+      "loss": 6.9451,
       "step": 1384
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.88728323699422e-05,
-      "loss": 6.6342,
+      "learning_rate": 0.0002666184971098266,
+      "loss": 7.0072,
       "step": 1385
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.884393063583816e-05,
-      "loss": 6.3975,
+      "learning_rate": 0.0002665317919075144,
+      "loss": 6.9182,
       "step": 1386
     },
     {
       "epoch": 1.55,
-      "learning_rate": 8.881502890173411e-05,
-      "loss": 6.4607,
+      "learning_rate": 0.0002664450867052023,
+      "loss": 6.9589,
       "step": 1387
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.878612716763006e-05,
-      "loss": 6.6045,
+      "learning_rate": 0.00026635838150289016,
+      "loss": 7.0718,
       "step": 1388
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.875722543352602e-05,
-      "loss": 6.0668,
+      "learning_rate": 0.000266271676300578,
+      "loss": 6.6617,
       "step": 1389
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.872832369942197e-05,
-      "loss": 6.0906,
+      "learning_rate": 0.00026618497109826586,
+      "loss": 6.7721,
       "step": 1390
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.869942196531792e-05,
-      "loss": 6.1648,
+      "learning_rate": 0.00026609826589595373,
+      "loss": 6.7833,
       "step": 1391
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.867052023121388e-05,
-      "loss": 5.5363,
+      "learning_rate": 0.00026601156069364155,
+      "loss": 6.3809,
       "step": 1392
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.864161849710983e-05,
-      "loss": 6.9825,
+      "learning_rate": 0.0002659248554913295,
+      "loss": 7.0456,
       "step": 1393
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.861271676300578e-05,
-      "loss": 6.9885,
+      "learning_rate": 0.0002658381502890173,
+      "loss": 7.0945,
       "step": 1394
     },
     {
       "epoch": 1.56,
-      "learning_rate": 8.858381502890174e-05,
-      "loss": 6.8796,
+      "learning_rate": 0.0002657514450867052,
+      "loss": 7.0478,
       "step": 1395
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.855491329479769e-05,
-      "loss": 6.8141,
+      "learning_rate": 0.00026566473988439305,
+      "loss": 6.9443,
       "step": 1396
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.852601156069364e-05,
-      "loss": 6.7062,
+      "learning_rate": 0.0002655780346820809,
+      "loss": 6.8748,
       "step": 1397
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.84971098265896e-05,
-      "loss": 6.8572,
+      "learning_rate": 0.00026549132947976874,
+      "loss": 7.0025,
       "step": 1398
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.846820809248555e-05,
-      "loss": 6.8146,
+      "learning_rate": 0.0002654046242774566,
+      "loss": 6.9858,
       "step": 1399
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.84393063583815e-05,
-      "loss": 6.841,
+      "learning_rate": 0.0002653179190751445,
+      "loss": 6.9927,
       "step": 1400
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.841040462427746e-05,
-      "loss": 6.7717,
+      "learning_rate": 0.00026523121387283236,
+      "loss": 6.9565,
       "step": 1401
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.838150289017341e-05,
-      "loss": 6.8371,
+      "learning_rate": 0.00026514450867052024,
+      "loss": 7.0021,
       "step": 1402
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.835260115606937e-05,
-      "loss": 6.6904,
+      "learning_rate": 0.00026505780346820806,
+      "loss": 6.9112,
       "step": 1403
     },
     {
       "epoch": 1.57,
-      "learning_rate": 8.832369942196532e-05,
-      "loss": 6.7732,
+      "learning_rate": 0.00026497109826589593,
+      "loss": 7.0751,
       "step": 1404
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.829479768786127e-05,
-      "loss": 6.6941,
+      "learning_rate": 0.0002648843930635838,
+      "loss": 6.9154,
       "step": 1405
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.826589595375723e-05,
-      "loss": 6.794,
+      "learning_rate": 0.0002647976878612716,
+      "loss": 6.9798,
       "step": 1406
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.823699421965318e-05,
-      "loss": 6.7916,
+      "learning_rate": 0.0002647109826589595,
+      "loss": 7.0247,
       "step": 1407
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.820809248554913e-05,
-      "loss": 6.6544,
+      "learning_rate": 0.00026462427745664737,
+      "loss": 6.9248,
       "step": 1408
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.81791907514451e-05,
-      "loss": 6.8014,
+      "learning_rate": 0.00026453757225433525,
+      "loss": 7.0063,
       "step": 1409
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.815028901734105e-05,
-      "loss": 6.7925,
+      "learning_rate": 0.0002644508670520231,
+      "loss": 7.0526,
       "step": 1410
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.812138728323701e-05,
-      "loss": 6.7691,
+      "learning_rate": 0.00026436416184971094,
+      "loss": 7.0152,
       "step": 1411
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.809248554913295e-05,
-      "loss": 6.8132,
+      "learning_rate": 0.0002642774566473988,
+      "loss": 7.0567,
       "step": 1412
     },
     {
       "epoch": 1.58,
-      "learning_rate": 8.80635838150289e-05,
-      "loss": 6.684,
+      "learning_rate": 0.0002641907514450867,
+      "loss": 6.9244,
       "step": 1413
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.803468208092485e-05,
-      "loss": 6.6887,
+      "learning_rate": 0.00026410404624277456,
+      "loss": 6.9438,
       "step": 1414
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.800578034682081e-05,
-      "loss": 6.7164,
+      "learning_rate": 0.0002640173410404624,
+      "loss": 7.0067,
       "step": 1415
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.797687861271676e-05,
-      "loss": 6.58,
+      "learning_rate": 0.00026393063583815025,
+      "loss": 6.869,
       "step": 1416
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.794797687861273e-05,
-      "loss": 6.5732,
+      "learning_rate": 0.00026384393063583813,
+      "loss": 6.8377,
       "step": 1417
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.791907514450868e-05,
-      "loss": 6.5842,
+      "learning_rate": 0.000263757225433526,
+      "loss": 6.9125,
       "step": 1418
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.789017341040463e-05,
-      "loss": 6.6924,
+      "learning_rate": 0.0002636705202312139,
+      "loss": 6.9601,
       "step": 1419
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.786127167630059e-05,
-      "loss": 6.8736,
+      "learning_rate": 0.0002635838150289017,
+      "loss": 7.1062,
       "step": 1420
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.783236994219654e-05,
-      "loss": 6.7743,
+      "learning_rate": 0.00026349710982658957,
+      "loss": 7.0635,
       "step": 1421
     },
     {
       "epoch": 1.59,
-      "learning_rate": 8.780346820809248e-05,
-      "loss": 6.8299,
+      "learning_rate": 0.00026341040462427744,
+      "loss": 7.1106,
       "step": 1422
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.777456647398843e-05,
-      "loss": 6.7007,
+      "learning_rate": 0.00026332369942196526,
+      "loss": 7.014,
       "step": 1423
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.774566473988439e-05,
-      "loss": 6.5897,
+      "learning_rate": 0.00026323699421965314,
+      "loss": 6.9521,
       "step": 1424
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.771676300578036e-05,
-      "loss": 6.5344,
+      "learning_rate": 0.000263150289017341,
+      "loss": 6.8409,
       "step": 1425
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.768786127167631e-05,
-      "loss": 6.5539,
+      "learning_rate": 0.0002630635838150289,
+      "loss": 6.8629,
       "step": 1426
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.765895953757226e-05,
-      "loss": 6.5991,
+      "learning_rate": 0.00026297687861271676,
+      "loss": 6.9166,
       "step": 1427
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.763005780346822e-05,
-      "loss": 6.6142,
+      "learning_rate": 0.0002628901734104046,
+      "loss": 6.9298,
       "step": 1428
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.760115606936417e-05,
-      "loss": 6.5638,
+      "learning_rate": 0.00026280346820809245,
+      "loss": 6.8207,
       "step": 1429
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.757225433526012e-05,
-      "loss": 6.6755,
+      "learning_rate": 0.0002627167630057803,
+      "loss": 7.0026,
       "step": 1430
     },
     {
       "epoch": 1.6,
-      "learning_rate": 8.754335260115608e-05,
-      "loss": 6.6631,
+      "learning_rate": 0.0002626300578034682,
+      "loss": 7.0167,
       "step": 1431
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.751445086705203e-05,
-      "loss": 6.7668,
+      "learning_rate": 0.000262543352601156,
+      "loss": 7.0869,
       "step": 1432
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.748554913294798e-05,
-      "loss": 6.7827,
+      "learning_rate": 0.00026245664739884395,
+      "loss": 7.1264,
       "step": 1433
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.745664739884394e-05,
-      "loss": 6.5325,
+      "learning_rate": 0.00026236994219653177,
+      "loss": 6.8787,
       "step": 1434
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.742774566473989e-05,
-      "loss": 6.594,
+      "learning_rate": 0.00026228323699421964,
+      "loss": 6.9975,
       "step": 1435
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.739884393063584e-05,
-      "loss": 6.6582,
+      "learning_rate": 0.0002621965317919075,
+      "loss": 7.0536,
       "step": 1436
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.73699421965318e-05,
-      "loss": 6.5712,
+      "learning_rate": 0.00026210982658959533,
+      "loss": 6.985,
       "step": 1437
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.734104046242775e-05,
-      "loss": 6.219,
+      "learning_rate": 0.0002620231213872832,
+      "loss": 6.6977,
       "step": 1438
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.73121387283237e-05,
-      "loss": 6.2175,
+      "learning_rate": 0.0002619364161849711,
+      "loss": 6.8462,
       "step": 1439
     },
     {
       "epoch": 1.61,
-      "learning_rate": 8.728323699421966e-05,
-      "loss": 6.1539,
+      "learning_rate": 0.0002618497109826589,
+      "loss": 6.7694,
       "step": 1440
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.725433526011561e-05,
-      "loss": 5.7474,
+      "learning_rate": 0.00026176300578034683,
+      "loss": 6.4729,
       "step": 1441
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.722543352601156e-05,
-      "loss": 5.9744,
+      "learning_rate": 0.00026167630057803465,
+      "loss": 6.7614,
       "step": 1442
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.719653179190752e-05,
-      "loss": 6.8402,
+      "learning_rate": 0.0002615895953757225,
+      "loss": 6.942,
       "step": 1443
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.716763005780347e-05,
-      "loss": 6.7606,
+      "learning_rate": 0.0002615028901734104,
+      "loss": 6.9187,
       "step": 1444
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.713872832369942e-05,
-      "loss": 6.9661,
+      "learning_rate": 0.0002614161849710982,
+      "loss": 7.0935,
       "step": 1445
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.710982658959538e-05,
-      "loss": 6.7477,
+      "learning_rate": 0.0002613294797687861,
+      "loss": 6.8963,
       "step": 1446
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.708092485549133e-05,
-      "loss": 6.8106,
+      "learning_rate": 0.00026124277456647397,
+      "loss": 7.0353,
       "step": 1447
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.705202312138728e-05,
-      "loss": 6.6693,
+      "learning_rate": 0.00026115606936416184,
+      "loss": 6.8463,
       "step": 1448
     },
     {
       "epoch": 1.62,
-      "learning_rate": 8.702312138728325e-05,
-      "loss": 6.7578,
+      "learning_rate": 0.0002610693641618497,
+      "loss": 6.975,
       "step": 1449
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.69942196531792e-05,
-      "loss": 6.863,
+      "learning_rate": 0.0002609826589595376,
+      "loss": 7.0488,
       "step": 1450
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.696531791907514e-05,
-      "loss": 6.7228,
+      "learning_rate": 0.0002608959537572254,
+      "loss": 6.9503,
       "step": 1451
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.69364161849711e-05,
-      "loss": 6.9035,
+      "learning_rate": 0.0002608092485549133,
+      "loss": 7.1028,
       "step": 1452
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.690751445086705e-05,
-      "loss": 6.7297,
+      "learning_rate": 0.00026072254335260115,
+      "loss": 6.956,
       "step": 1453
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.6878612716763e-05,
-      "loss": 6.6187,
+      "learning_rate": 0.000260635838150289,
+      "loss": 6.8677,
       "step": 1454
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.684971098265896e-05,
-      "loss": 6.7969,
+      "learning_rate": 0.00026054913294797685,
+      "loss": 7.0284,
       "step": 1455
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.682080924855491e-05,
-      "loss": 6.6973,
+      "learning_rate": 0.0002604624277456647,
+      "loss": 6.9777,
       "step": 1456
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.679190751445088e-05,
-      "loss": 6.7217,
+      "learning_rate": 0.0002603757225433526,
+      "loss": 7.0098,
       "step": 1457
     },
     {
       "epoch": 1.63,
-      "learning_rate": 8.676300578034683e-05,
-      "loss": 6.6774,
+      "learning_rate": 0.00026028901734104047,
+      "loss": 6.9353,
       "step": 1458
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.673410404624279e-05,
-      "loss": 6.7955,
+      "learning_rate": 0.0002602023121387283,
+      "loss": 7.0551,
       "step": 1459
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.670520231213874e-05,
-      "loss": 6.6913,
+      "learning_rate": 0.00026011560693641616,
+      "loss": 6.965,
       "step": 1460
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.667630057803469e-05,
-      "loss": 6.6226,
+      "learning_rate": 0.00026002890173410404,
+      "loss": 6.9107,
       "step": 1461
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.664739884393063e-05,
-      "loss": 6.749,
+      "learning_rate": 0.00025994219653179186,
+      "loss": 7.0147,
       "step": 1462
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.661849710982659e-05,
-      "loss": 6.6374,
+      "learning_rate": 0.00025985549132947973,
+      "loss": 6.934,
       "step": 1463
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.658959537572254e-05,
-      "loss": 6.8614,
+      "learning_rate": 0.0002597687861271676,
+      "loss": 7.1356,
       "step": 1464
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.65606936416185e-05,
-      "loss": 6.6829,
+      "learning_rate": 0.0002596820809248555,
+      "loss": 6.9668,
       "step": 1465
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.653179190751446e-05,
-      "loss": 6.6748,
+      "learning_rate": 0.00025959537572254335,
+      "loss": 7.0069,
       "step": 1466
     },
     {
       "epoch": 1.64,
-      "learning_rate": 8.650289017341041e-05,
-      "loss": 6.6369,
+      "learning_rate": 0.00025950867052023117,
+      "loss": 6.9394,
       "step": 1467
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.647398843930637e-05,
-      "loss": 6.7427,
+      "learning_rate": 0.00025942196531791905,
+      "loss": 7.0758,
       "step": 1468
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.644508670520232e-05,
-      "loss": 6.6899,
+      "learning_rate": 0.0002593352601156069,
+      "loss": 6.9921,
       "step": 1469
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.641618497109827e-05,
-      "loss": 6.6439,
+      "learning_rate": 0.0002592485549132948,
+      "loss": 6.9199,
       "step": 1470
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.638728323699423e-05,
-      "loss": 6.5822,
+      "learning_rate": 0.0002591618497109826,
+      "loss": 6.8884,
       "step": 1471
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.635838150289017e-05,
-      "loss": 6.7477,
+      "learning_rate": 0.0002590751445086705,
+      "loss": 7.0782,
       "step": 1472
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.632947976878613e-05,
-      "loss": 6.5864,
+      "learning_rate": 0.00025898843930635836,
+      "loss": 6.9281,
       "step": 1473
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.630057803468209e-05,
-      "loss": 6.5783,
+      "learning_rate": 0.00025890173410404624,
+      "loss": 6.9054,
       "step": 1474
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.627167630057804e-05,
-      "loss": 6.4554,
+      "learning_rate": 0.00025890173410404624,
+      "loss": 9.5025,
       "step": 1475
     },
     {
       "epoch": 1.65,
-      "learning_rate": 8.6242774566474e-05,
-      "loss": 6.582,
+      "learning_rate": 0.00025890173410404624,
+      "loss": 9.8928,
       "step": 1476
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.621387283236995e-05,
-      "loss": 6.7103,
+      "learning_rate": 0.0002588150289017341,
+      "loss": 10.2183,
       "step": 1477
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.61849710982659e-05,
-      "loss": 6.5102,
+      "learning_rate": 0.00025872832369942193,
+      "loss": 6.9026,
       "step": 1478
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.615606936416185e-05,
-      "loss": 6.6684,
+      "learning_rate": 0.0002586416184971098,
+      "loss": 6.9966,
       "step": 1479
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.612716763005781e-05,
-      "loss": 6.6518,
+      "learning_rate": 0.0002585549132947977,
+      "loss": 6.9904,
       "step": 1480
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.609826589595376e-05,
-      "loss": 6.4647,
+      "learning_rate": 0.0002584682080924855,
+      "loss": 6.8403,
       "step": 1481
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.606936416184972e-05,
-      "loss": 6.4951,
+      "learning_rate": 0.00025838150289017337,
+      "loss": 6.9223,
       "step": 1482
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.604046242774567e-05,
-      "loss": 6.4654,
+      "learning_rate": 0.00025829479768786124,
+      "loss": 6.8427,
       "step": 1483
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.601156069364162e-05,
-      "loss": 6.6174,
+      "learning_rate": 0.0002582080924855491,
+      "loss": 7.0218,
       "step": 1484
     },
     {
       "epoch": 1.66,
-      "learning_rate": 8.598265895953758e-05,
-      "loss": 6.6163,
+      "learning_rate": 0.000258121387283237,
+      "loss": 7.0305,
       "step": 1485
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.595375722543353e-05,
-      "loss": 6.5572,
+      "learning_rate": 0.0002580346820809248,
+      "loss": 7.0112,
       "step": 1486
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.592485549132948e-05,
-      "loss": 6.4622,
+      "learning_rate": 0.0002579479768786127,
+      "loss": 6.968,
       "step": 1487
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.589595375722544e-05,
-      "loss": 6.4749,
+      "learning_rate": 0.00025786127167630056,
+      "loss": 6.9765,
       "step": 1488
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.586705202312139e-05,
-      "loss": 6.1401,
+      "learning_rate": 0.00025777456647398843,
+      "loss": 6.7469,
       "step": 1489
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.583815028901736e-05,
-      "loss": 6.3766,
+      "learning_rate": 0.00025768786127167625,
+      "loss": 7.0633,
       "step": 1490
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.58092485549133e-05,
-      "loss": 6.0161,
+      "learning_rate": 0.0002576011560693641,
+      "loss": 6.704,
       "step": 1491
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.578034682080925e-05,
-      "loss": 5.9076,
+      "learning_rate": 0.000257514450867052,
+      "loss": 6.7634,
       "step": 1492
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.57514450867052e-05,
-      "loss": 6.8832,
+      "learning_rate": 0.0002574277456647399,
+      "loss": 7.0861,
       "step": 1493
     },
     {
       "epoch": 1.67,
-      "learning_rate": 8.572254335260116e-05,
-      "loss": 6.7631,
+      "learning_rate": 0.00025734104046242775,
+      "loss": 6.9248,
       "step": 1494
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.569364161849711e-05,
-      "loss": 6.74,
+      "learning_rate": 0.00025725433526011557,
+      "loss": 6.9327,
       "step": 1495
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.566473988439306e-05,
-      "loss": 6.7615,
+      "learning_rate": 0.00025716763005780344,
+      "loss": 6.9496,
       "step": 1496
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.563583815028902e-05,
-      "loss": 6.8259,
+      "learning_rate": 0.0002570809248554913,
+      "loss": 7.0318,
       "step": 1497
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.560693641618498e-05,
-      "loss": 6.7028,
+      "learning_rate": 0.00025699421965317914,
+      "loss": 6.9218,
       "step": 1498
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.557803468208094e-05,
-      "loss": 6.8608,
+      "learning_rate": 0.000256907514450867,
+      "loss": 7.0631,
       "step": 1499
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.554913294797689e-05,
-      "loss": 6.6771,
+      "learning_rate": 0.0002568208092485549,
+      "loss": 6.9006,
       "step": 1500
     },
     {
       "epoch": 1.68,
-      "eval_loss": 6.590250492095947,
-      "eval_runtime": 988.737,
-      "eval_samples_per_second": 2.672,
-      "eval_steps_per_second": 0.335,
-      "eval_wer": 1.447883802607298,
+      "eval_loss": 6.950684070587158,
+      "eval_runtime": 984.3687,
+      "eval_samples_per_second": 2.684,
+      "eval_steps_per_second": 0.336,
+      "eval_wer": 2.0097029585094353,
       "step": 1500
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.552023121387283e-05,
-      "loss": 6.7678,
+      "learning_rate": 0.00025673410404624276,
+      "loss": 6.9684,
       "step": 1501
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.549132947976878e-05,
-      "loss": 6.6625,
+      "learning_rate": 0.00025664739884393063,
+      "loss": 6.9068,
       "step": 1502
     },
     {
       "epoch": 1.68,
-      "learning_rate": 8.546242774566474e-05,
-      "loss": 6.5489,
+      "learning_rate": 0.00025656069364161845,
+      "loss": 6.8335,
       "step": 1503
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.543352601156069e-05,
-      "loss": 6.672,
+      "learning_rate": 0.0002564739884393063,
+      "loss": 6.9637,
       "step": 1504
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.540462427745664e-05,
-      "loss": 6.6687,
+      "learning_rate": 0.0002563872832369942,
+      "loss": 7.0084,
       "step": 1505
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.537572254335261e-05,
-      "loss": 6.6571,
+      "learning_rate": 0.00025630057803468207,
+      "loss": 6.9313,
       "step": 1506
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.534682080924856e-05,
-      "loss": 6.5162,
+      "learning_rate": 0.0002562138728323699,
+      "loss": 6.8596,
       "step": 1507
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.531791907514452e-05,
-      "loss": 6.6387,
+      "learning_rate": 0.0002561271676300578,
+      "loss": 6.9095,
       "step": 1508
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.528901734104047e-05,
-      "loss": 6.5701,
+      "learning_rate": 0.00025604046242774564,
+      "loss": 6.837,
       "step": 1509
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.526011560693643e-05,
-      "loss": 6.5151,
+      "learning_rate": 0.0002559537572254335,
+      "loss": 6.8328,
       "step": 1510
     },
     {
       "epoch": 1.69,
-      "learning_rate": 8.523121387283236e-05,
-      "loss": 6.6113,
+      "learning_rate": 0.0002558670520231214,
+      "loss": 6.9344,
       "step": 1511
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.520231213872832e-05,
-      "loss": 6.6368,
+      "learning_rate": 0.0002557803468208092,
+      "loss": 6.9766,
       "step": 1512
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.517341040462427e-05,
-      "loss": 6.744,
+      "learning_rate": 0.0002556936416184971,
+      "loss": 7.0267,
       "step": 1513
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.514450867052024e-05,
-      "loss": 6.6867,
+      "learning_rate": 0.00025560693641618496,
+      "loss": 6.9732,
       "step": 1514
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.511560693641619e-05,
-      "loss": 6.702,
+      "learning_rate": 0.0002555202312138728,
+      "loss": 6.9781,
       "step": 1515
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.508670520231215e-05,
-      "loss": 6.7214,
+      "learning_rate": 0.0002554335260115607,
+      "loss": 7.0235,
       "step": 1516
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.50578034682081e-05,
-      "loss": 6.6063,
+      "learning_rate": 0.0002553468208092485,
+      "loss": 6.9375,
       "step": 1517
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.502890173410405e-05,
-      "loss": 6.6116,
+      "learning_rate": 0.0002552601156069364,
+      "loss": 6.9598,
       "step": 1518
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.5e-05,
-      "loss": 6.5994,
+      "learning_rate": 0.00025517341040462427,
+      "loss": 6.9656,
       "step": 1519
     },
     {
       "epoch": 1.7,
-      "learning_rate": 8.497109826589596e-05,
-      "loss": 6.5855,
+      "learning_rate": 0.0002550867052023121,
+      "loss": 6.9019,
       "step": 1520
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.494219653179191e-05,
-      "loss": 6.7625,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 7.0767,
       "step": 1521
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.491329479768787e-05,
-      "loss": 6.7785,
+      "learning_rate": 0.00025491329479768784,
+      "loss": 7.0776,
       "step": 1522
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.488439306358382e-05,
-      "loss": 6.4986,
+      "learning_rate": 0.0002548265895953757,
+      "loss": 6.8564,
       "step": 1523
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.485549132947977e-05,
-      "loss": 6.6255,
+      "learning_rate": 0.0002547398843930636,
+      "loss": 6.985,
       "step": 1524
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.482658959537573e-05,
-      "loss": 6.7036,
+      "learning_rate": 0.00025465317919075146,
+      "loss": 7.0325,
       "step": 1525
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.479768786127168e-05,
-      "loss": 6.5569,
+      "learning_rate": 0.0002545664739884393,
+      "loss": 6.9548,
       "step": 1526
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.476878612716763e-05,
-      "loss": 6.6511,
+      "learning_rate": 0.00025447976878612715,
+      "loss": 7.0547,
       "step": 1527
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.473988439306359e-05,
-      "loss": 6.6761,
+      "learning_rate": 0.00025439306358381503,
+      "loss": 7.0514,
       "step": 1528
     },
     {
       "epoch": 1.71,
-      "learning_rate": 8.471098265895954e-05,
-      "loss": 6.6102,
+      "learning_rate": 0.00025430635838150285,
+      "loss": 7.0317,
       "step": 1529
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.46820809248555e-05,
-      "loss": 6.6288,
+      "learning_rate": 0.0002542196531791907,
+      "loss": 7.0048,
       "step": 1530
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.465317919075145e-05,
-      "loss": 6.4882,
+      "learning_rate": 0.0002541329479768786,
+      "loss": 6.9406,
       "step": 1531
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.46242774566474e-05,
-      "loss": 6.5146,
+      "learning_rate": 0.00025404624277456647,
+      "loss": 6.9451,
       "step": 1532
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.459537572254335e-05,
-      "loss": 6.4439,
+      "learning_rate": 0.00025395953757225434,
+      "loss": 6.9089,
       "step": 1533
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.456647398843931e-05,
-      "loss": 6.5248,
+      "learning_rate": 0.00025387283236994216,
+      "loss": 6.9334,
       "step": 1534
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.453757225433526e-05,
-      "loss": 6.4031,
+      "learning_rate": 0.00025378612716763004,
+      "loss": 6.8563,
       "step": 1535
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.450867052023121e-05,
-      "loss": 6.5097,
+      "learning_rate": 0.0002536994219653179,
+      "loss": 7.0497,
       "step": 1536
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.447976878612717e-05,
-      "loss": 6.2889,
+      "learning_rate": 0.00025361271676300573,
+      "loss": 6.8794,
       "step": 1537
     },
     {
       "epoch": 1.72,
-      "learning_rate": 8.445086705202314e-05,
-      "loss": 6.267,
+      "learning_rate": 0.0002535260115606936,
+      "loss": 6.8202,
       "step": 1538
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.442196531791909e-05,
-      "loss": 6.1338,
+      "learning_rate": 0.0002534393063583815,
+      "loss": 6.751,
       "step": 1539
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.439306358381503e-05,
-      "loss": 6.1696,
+      "learning_rate": 0.00025335260115606935,
+      "loss": 6.9672,
       "step": 1540
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.436416184971098e-05,
-      "loss": 6.086,
+      "learning_rate": 0.0002532658959537572,
+      "loss": 6.8183,
       "step": 1541
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.433526011560694e-05,
-      "loss": 5.6172,
+      "learning_rate": 0.00025317919075144504,
+      "loss": 6.4965,
       "step": 1542
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.430635838150289e-05,
-      "loss": 6.9292,
+      "learning_rate": 0.0002530924855491329,
+      "loss": 7.0608,
       "step": 1543
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.427745664739884e-05,
-      "loss": 6.9568,
+      "learning_rate": 0.0002530057803468208,
+      "loss": 7.0964,
       "step": 1544
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.42485549132948e-05,
-      "loss": 6.8594,
+      "learning_rate": 0.00025291907514450867,
+      "loss": 7.0435,
       "step": 1545
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.421965317919076e-05,
-      "loss": 6.8073,
+      "learning_rate": 0.0002528323699421965,
+      "loss": 6.9756,
       "step": 1546
     },
     {
       "epoch": 1.73,
-      "learning_rate": 8.419075144508672e-05,
-      "loss": 6.7406,
+      "learning_rate": 0.00025274566473988436,
+      "loss": 6.9558,
       "step": 1547
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.416184971098267e-05,
-      "loss": 6.6528,
+      "learning_rate": 0.00025265895953757223,
+      "loss": 6.9328,
       "step": 1548
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.413294797687862e-05,
-      "loss": 6.8674,
+      "learning_rate": 0.0002525722543352601,
+      "loss": 7.1163,
       "step": 1549
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.410404624277458e-05,
-      "loss": 6.7163,
+      "learning_rate": 0.000252485549132948,
+      "loss": 6.9838,
       "step": 1550
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.407514450867052e-05,
-      "loss": 6.6875,
+      "learning_rate": 0.0002523988439306358,
+      "loss": 6.9539,
       "step": 1551
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.404624277456647e-05,
-      "loss": 6.8431,
+      "learning_rate": 0.0002523121387283237,
+      "loss": 7.0622,
       "step": 1552
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.401734104046242e-05,
-      "loss": 6.7371,
+      "learning_rate": 0.00025222543352601155,
+      "loss": 7.0412,
       "step": 1553
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.398843930635839e-05,
-      "loss": 6.5288,
+      "learning_rate": 0.00025213872832369937,
+      "loss": 6.825,
       "step": 1554
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.395953757225434e-05,
-      "loss": 6.6481,
+      "learning_rate": 0.00025205202312138724,
+      "loss": 6.9309,
       "step": 1555
     },
     {
       "epoch": 1.74,
-      "learning_rate": 8.39306358381503e-05,
-      "loss": 6.6048,
+      "learning_rate": 0.0002519653179190751,
+      "loss": 6.8787,
       "step": 1556
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.390173410404625e-05,
-      "loss": 6.6506,
+      "learning_rate": 0.000251878612716763,
+      "loss": 6.9472,
       "step": 1557
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.38728323699422e-05,
-      "loss": 6.6265,
+      "learning_rate": 0.00025179190751445086,
+      "loss": 6.9533,
       "step": 1558
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.384393063583816e-05,
-      "loss": 6.6603,
+      "learning_rate": 0.0002517052023121387,
+      "loss": 6.9886,
       "step": 1559
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.381502890173411e-05,
-      "loss": 6.6702,
+      "learning_rate": 0.00025161849710982656,
+      "loss": 6.9996,
       "step": 1560
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.378612716763005e-05,
-      "loss": 6.6077,
+      "learning_rate": 0.00025153179190751443,
+      "loss": 6.9173,
       "step": 1561
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.375722543352602e-05,
-      "loss": 6.7751,
+      "learning_rate": 0.0002514450867052023,
+      "loss": 7.0896,
       "step": 1562
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.372832369942197e-05,
-      "loss": 6.6844,
+      "learning_rate": 0.0002513583815028901,
+      "loss": 7.0304,
       "step": 1563
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.369942196531792e-05,
-      "loss": 6.6499,
+      "learning_rate": 0.00025127167630057805,
+      "loss": 6.9784,
       "step": 1564
     },
     {
       "epoch": 1.75,
-      "learning_rate": 8.367052023121388e-05,
-      "loss": 6.4721,
+      "learning_rate": 0.0002511849710982659,
+      "loss": 6.8033,
       "step": 1565
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.364161849710983e-05,
-      "loss": 6.7714,
+      "learning_rate": 0.00025109826589595375,
+      "loss": 7.0936,
       "step": 1566
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.361271676300579e-05,
-      "loss": 6.6632,
+      "learning_rate": 0.0002510115606936416,
+      "loss": 6.987,
       "step": 1567
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.358381502890174e-05,
-      "loss": 6.4433,
+      "learning_rate": 0.00025092485549132944,
+      "loss": 6.8598,
       "step": 1568
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.355491329479769e-05,
-      "loss": 6.5281,
+      "learning_rate": 0.0002508381502890173,
+      "loss": 6.9054,
       "step": 1569
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.352601156069365e-05,
-      "loss": 6.629,
+      "learning_rate": 0.0002507514450867052,
+      "loss": 7.0212,
       "step": 1570
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.34971098265896e-05,
-      "loss": 6.5747,
+      "learning_rate": 0.000250664739884393,
+      "loss": 6.9968,
       "step": 1571
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.346820809248555e-05,
-      "loss": 6.6537,
+      "learning_rate": 0.00025057803468208094,
+      "loss": 7.0157,
       "step": 1572
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.34393063583815e-05,
-      "loss": 6.5937,
+      "learning_rate": 0.00025049132947976876,
+      "loss": 6.9591,
       "step": 1573
     },
     {
       "epoch": 1.76,
-      "learning_rate": 8.341040462427746e-05,
-      "loss": 6.5011,
+      "learning_rate": 0.00025040462427745663,
+      "loss": 6.9147,
       "step": 1574
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.338150289017341e-05,
-      "loss": 6.5943,
+      "learning_rate": 0.0002503179190751445,
+      "loss": 6.9684,
       "step": 1575
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.335260115606937e-05,
-      "loss": 6.6089,
+      "learning_rate": 0.0002502312138728323,
+      "loss": 7.0229,
       "step": 1576
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.332369942196532e-05,
-      "loss": 6.5746,
+      "learning_rate": 0.0002501445086705202,
+      "loss": 6.9602,
       "step": 1577
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.329479768786127e-05,
-      "loss": 6.5245,
+      "learning_rate": 0.00025005780346820807,
+      "loss": 6.9206,
       "step": 1578
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.326589595375724e-05,
-      "loss": 6.5139,
+      "learning_rate": 0.00024997109826589595,
+      "loss": 6.9201,
       "step": 1579
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.323699421965318e-05,
-      "loss": 6.6734,
+      "learning_rate": 0.00024988439306358376,
+      "loss": 7.0534,
       "step": 1580
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.320809248554913e-05,
-      "loss": 6.5112,
+      "learning_rate": 0.0002497976878612717,
+      "loss": 6.93,
       "step": 1581
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.317919075144509e-05,
-      "loss": 6.5576,
+      "learning_rate": 0.0002497109826589595,
+      "loss": 6.999,
       "step": 1582
     },
     {
       "epoch": 1.77,
-      "learning_rate": 8.315028901734104e-05,
-      "loss": 6.5332,
+      "learning_rate": 0.0002496242774566474,
+      "loss": 6.9831,
       "step": 1583
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.3121387283237e-05,
-      "loss": 6.5529,
+      "learning_rate": 0.00024953757225433526,
+      "loss": 7.0028,
       "step": 1584
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.309248554913295e-05,
-      "loss": 6.3277,
+      "learning_rate": 0.0002494508670520231,
+      "loss": 6.8744,
       "step": 1585
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.30635838150289e-05,
-      "loss": 6.3373,
+      "learning_rate": 0.00024936416184971095,
+      "loss": 6.8706,
       "step": 1586
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.303468208092487e-05,
-      "loss": 6.4268,
+      "learning_rate": 0.00024927745664739883,
+      "loss": 6.9726,
       "step": 1587
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.300578034682082e-05,
-      "loss": 6.1773,
+      "learning_rate": 0.00024919075144508665,
+      "loss": 6.7924,
       "step": 1588
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.297687861271677e-05,
-      "loss": 6.3622,
+      "learning_rate": 0.0002491040462427746,
+      "loss": 6.9609,
       "step": 1589
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.294797687861271e-05,
-      "loss": 6.0731,
+      "learning_rate": 0.0002490173410404624,
+      "loss": 6.8327,
       "step": 1590
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.291907514450867e-05,
-      "loss": 6.1328,
+      "learning_rate": 0.00024893063583815027,
+      "loss": 6.8986,
       "step": 1591
     },
     {
       "epoch": 1.78,
-      "learning_rate": 8.289017341040462e-05,
-      "loss": 5.5035,
+      "learning_rate": 0.00024884393063583814,
+      "loss": 6.4239,
       "step": 1592
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.286127167630057e-05,
-      "loss": 7.0565,
+      "learning_rate": 0.00024875722543352596,
+      "loss": 7.0544,
       "step": 1593
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.283236994219653e-05,
-      "loss": 6.986,
+      "learning_rate": 0.00024867052023121384,
+      "loss": 7.0459,
       "step": 1594
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.28034682080925e-05,
-      "loss": 6.8284,
+      "learning_rate": 0.0002485838150289017,
+      "loss": 7.0072,
       "step": 1595
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.277456647398845e-05,
-      "loss": 6.8633,
+      "learning_rate": 0.0002484971098265896,
+      "loss": 6.9776,
       "step": 1596
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.27456647398844e-05,
-      "loss": 6.7321,
+      "learning_rate": 0.00024841040462427746,
+      "loss": 6.9233,
       "step": 1597
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.271676300578036e-05,
-      "loss": 6.7852,
+      "learning_rate": 0.00024832369942196533,
+      "loss": 6.9851,
       "step": 1598
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.268786127167631e-05,
-      "loss": 6.7595,
+      "learning_rate": 0.00024823699421965315,
+      "loss": 6.9801,
       "step": 1599
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.265895953757226e-05,
-      "loss": 6.9473,
+      "learning_rate": 0.000248150289017341,
+      "loss": 7.1636,
       "step": 1600
     },
     {
       "epoch": 1.79,
-      "learning_rate": 8.26300578034682e-05,
-      "loss": 6.7889,
+      "learning_rate": 0.0002480635838150289,
+      "loss": 7.0498,
       "step": 1601
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.260115606936416e-05,
-      "loss": 6.8197,
+      "learning_rate": 0.0002479768786127167,
+      "loss": 7.0476,
       "step": 1602
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.257225433526012e-05,
-      "loss": 6.7356,
+      "learning_rate": 0.0002478901734104046,
+      "loss": 7.0091,
       "step": 1603
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.254335260115608e-05,
-      "loss": 6.7413,
+      "learning_rate": 0.00024780346820809247,
+      "loss": 6.9853,
       "step": 1604
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.251445086705203e-05,
-      "loss": 6.7393,
+      "learning_rate": 0.00024771676300578034,
+      "loss": 7.0274,
       "step": 1605
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.248554913294798e-05,
-      "loss": 6.6865,
+      "learning_rate": 0.0002476300578034682,
+      "loss": 6.9714,
       "step": 1606
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.245664739884394e-05,
-      "loss": 6.6253,
+      "learning_rate": 0.00024754335260115603,
+      "loss": 6.9648,
       "step": 1607
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.242774566473989e-05,
-      "loss": 6.6008,
+      "learning_rate": 0.0002474566473988439,
+      "loss": 6.9614,
       "step": 1608
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.239884393063584e-05,
-      "loss": 6.6707,
+      "learning_rate": 0.0002473699421965318,
+      "loss": 6.9848,
       "step": 1609
     },
     {
       "epoch": 1.8,
-      "learning_rate": 8.23699421965318e-05,
-      "loss": 6.5569,
+      "learning_rate": 0.0002472832369942196,
+      "loss": 6.8763,
       "step": 1610
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.234104046242775e-05,
-      "loss": 6.7225,
+      "learning_rate": 0.0002471965317919075,
+      "loss": 7.0228,
       "step": 1611
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.23121387283237e-05,
-      "loss": 6.6101,
+      "learning_rate": 0.00024710982658959535,
+      "loss": 6.9546,
       "step": 1612
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.228323699421966e-05,
-      "loss": 6.6208,
+      "learning_rate": 0.0002470231213872832,
+      "loss": 6.9617,
       "step": 1613
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.225433526011561e-05,
-      "loss": 6.7233,
+      "learning_rate": 0.0002469364161849711,
+      "loss": 7.0605,
       "step": 1614
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.222543352601156e-05,
-      "loss": 6.6375,
+      "learning_rate": 0.0002468497109826589,
+      "loss": 6.9841,
       "step": 1615
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.219653179190752e-05,
-      "loss": 6.6025,
+      "learning_rate": 0.0002467630057803468,
+      "loss": 6.9616,
       "step": 1616
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.216763005780347e-05,
-      "loss": 6.7027,
+      "learning_rate": 0.00024667630057803467,
+      "loss": 7.0764,
       "step": 1617
     },
     {
       "epoch": 1.81,
-      "learning_rate": 8.213872832369942e-05,
-      "loss": 6.8753,
+      "learning_rate": 0.00024658959537572254,
+      "loss": 7.2082,
       "step": 1618
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.210982658959538e-05,
-      "loss": 6.5813,
+      "learning_rate": 0.00024650289017341036,
+      "loss": 6.9536,
       "step": 1619
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.208092485549133e-05,
-      "loss": 6.442,
+      "learning_rate": 0.00024641618497109823,
+      "loss": 6.837,
       "step": 1620
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.205202312138728e-05,
-      "loss": 6.6826,
+      "learning_rate": 0.0002463294797687861,
+      "loss": 7.0407,
       "step": 1621
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.202312138728324e-05,
-      "loss": 6.4822,
+      "learning_rate": 0.000246242774566474,
+      "loss": 6.9018,
       "step": 1622
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.199421965317919e-05,
-      "loss": 6.5977,
+      "learning_rate": 0.00024615606936416185,
+      "loss": 7.0039,
       "step": 1623
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.196531791907515e-05,
-      "loss": 6.4265,
+      "learning_rate": 0.0002460693641618497,
+      "loss": 6.774,
       "step": 1624
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.19364161849711e-05,
-      "loss": 6.4968,
+      "learning_rate": 0.00024598265895953755,
+      "loss": 6.9344,
       "step": 1625
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.190751445086705e-05,
-      "loss": 6.4677,
+      "learning_rate": 0.0002458959537572254,
+      "loss": 6.891,
       "step": 1626
     },
     {
       "epoch": 1.82,
-      "learning_rate": 8.187861271676302e-05,
-      "loss": 6.5294,
+      "learning_rate": 0.00024580924855491324,
+      "loss": 6.99,
       "step": 1627
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.184971098265897e-05,
-      "loss": 6.6935,
+      "learning_rate": 0.0002457225433526011,
+      "loss": 7.0999,
       "step": 1628
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.182080924855493e-05,
-      "loss": 6.5743,
+      "learning_rate": 0.000245635838150289,
+      "loss": 6.9851,
       "step": 1629
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.179190751445087e-05,
-      "loss": 6.6195,
+      "learning_rate": 0.00024554913294797686,
+      "loss": 7.0176,
       "step": 1630
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.176300578034682e-05,
-      "loss": 6.3297,
+      "learning_rate": 0.00024546242774566474,
+      "loss": 6.7989,
       "step": 1631
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.173410404624277e-05,
-      "loss": 6.5284,
+      "learning_rate": 0.00024537572254335256,
+      "loss": 6.9643,
       "step": 1632
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.170520231213873e-05,
-      "loss": 6.6593,
+      "learning_rate": 0.00024528901734104043,
+      "loss": 7.1934,
       "step": 1633
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.167630057803468e-05,
-      "loss": 6.708,
+      "learning_rate": 0.0002452023121387283,
+      "loss": 7.2043,
       "step": 1634
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.164739884393065e-05,
-      "loss": 6.4656,
+      "learning_rate": 0.0002451156069364162,
+      "loss": 6.9846,
       "step": 1635
     },
     {
       "epoch": 1.83,
-      "learning_rate": 8.16184971098266e-05,
-      "loss": 6.3241,
+      "learning_rate": 0.000245028901734104,
+      "loss": 6.8466,
       "step": 1636
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.158959537572255e-05,
-      "loss": 6.3198,
+      "learning_rate": 0.0002449421965317919,
+      "loss": 6.8145,
       "step": 1637
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.156069364161851e-05,
-      "loss": 6.2828,
+      "learning_rate": 0.00024485549132947975,
+      "loss": 6.8791,
       "step": 1638
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.153179190751446e-05,
-      "loss": 6.081,
+      "learning_rate": 0.0002447687861271676,
+      "loss": 6.8022,
       "step": 1639
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.15028901734104e-05,
-      "loss": 5.9499,
+      "learning_rate": 0.0002446820809248555,
+      "loss": 6.765,
       "step": 1640
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.147398843930635e-05,
-      "loss": 6.1443,
+      "learning_rate": 0.0002445953757225433,
+      "loss": 6.9126,
       "step": 1641
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.144508670520231e-05,
-      "loss": 5.343,
+      "learning_rate": 0.0002445086705202312,
+      "loss": 6.3914,
       "step": 1642
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.141618497109827e-05,
-      "loss": 6.7823,
+      "learning_rate": 0.00024442196531791906,
+      "loss": 6.8989,
       "step": 1643
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.138728323699423e-05,
-      "loss": 6.9006,
+      "learning_rate": 0.0002443352601156069,
+      "loss": 7.082,
       "step": 1644
     },
     {
       "epoch": 1.84,
-      "learning_rate": 8.135838150289018e-05,
-      "loss": 6.8694,
+      "learning_rate": 0.0002442485549132948,
+      "loss": 7.0282,
       "step": 1645
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.132947976878613e-05,
-      "loss": 6.8584,
+      "learning_rate": 0.00024416184971098263,
+      "loss": 6.9782,
       "step": 1646
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.130057803468209e-05,
-      "loss": 6.7826,
+      "learning_rate": 0.0002440751445086705,
+      "loss": 6.9949,
       "step": 1647
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.127167630057804e-05,
-      "loss": 6.6929,
+      "learning_rate": 0.00024398843930635838,
+      "loss": 6.9005,
       "step": 1648
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.1242774566474e-05,
-      "loss": 6.8781,
+      "learning_rate": 0.00024390173410404622,
+      "loss": 7.0589,
       "step": 1649
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.121387283236995e-05,
-      "loss": 6.8897,
+      "learning_rate": 0.00024381502890173407,
+      "loss": 7.1095,
       "step": 1650
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.11849710982659e-05,
-      "loss": 6.6799,
+      "learning_rate": 0.00024372832369942194,
+      "loss": 6.9129,
       "step": 1651
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.115606936416186e-05,
-      "loss": 6.8916,
+      "learning_rate": 0.0002436416184971098,
+      "loss": 7.1147,
       "step": 1652
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.112716763005781e-05,
-      "loss": 6.6108,
+      "learning_rate": 0.0002435549132947977,
+      "loss": 6.8738,
       "step": 1653
     },
     {
       "epoch": 1.85,
-      "learning_rate": 8.109826589595376e-05,
-      "loss": 6.6062,
+      "learning_rate": 0.00024346820809248554,
+      "loss": 6.8884,
       "step": 1654
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.106936416184972e-05,
-      "loss": 6.6517,
+      "learning_rate": 0.00024338150289017339,
+      "loss": 6.9367,
       "step": 1655
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.104046242774567e-05,
-      "loss": 6.633,
+      "learning_rate": 0.00024329479768786126,
+      "loss": 6.9414,
       "step": 1656
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.101156069364162e-05,
-      "loss": 6.7857,
+      "learning_rate": 0.0002432080924855491,
+      "loss": 7.0612,
       "step": 1657
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.098265895953758e-05,
-      "loss": 6.8228,
+      "learning_rate": 0.00024312138728323698,
+      "loss": 7.1448,
       "step": 1658
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.095375722543353e-05,
-      "loss": 6.5484,
+      "learning_rate": 0.00024303468208092483,
+      "loss": 6.9032,
       "step": 1659
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.092485549132948e-05,
-      "loss": 6.7101,
+      "learning_rate": 0.00024294797687861267,
+      "loss": 7.027,
       "step": 1660
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.089595375722544e-05,
-      "loss": 6.705,
+      "learning_rate": 0.00024286127167630057,
+      "loss": 7.009,
       "step": 1661
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.086705202312139e-05,
-      "loss": 6.6148,
+      "learning_rate": 0.00024277456647398842,
+      "loss": 6.9093,
       "step": 1662
     },
     {
       "epoch": 1.86,
-      "learning_rate": 8.083815028901734e-05,
-      "loss": 6.5899,
+      "learning_rate": 0.0002426878612716763,
+      "loss": 6.9239,
       "step": 1663
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.08092485549133e-05,
-      "loss": 6.6591,
+      "learning_rate": 0.00024260115606936414,
+      "loss": 7.0106,
       "step": 1664
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.078034682080925e-05,
-      "loss": 6.5717,
+      "learning_rate": 0.000242514450867052,
+      "loss": 6.8883,
       "step": 1665
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.07514450867052e-05,
-      "loss": 6.752,
+      "learning_rate": 0.00024242774566473986,
+      "loss": 7.075,
       "step": 1666
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.072254335260117e-05,
-      "loss": 6.4942,
+      "learning_rate": 0.0002423410404624277,
+      "loss": 6.869,
       "step": 1667
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.069364161849712e-05,
-      "loss": 6.6555,
+      "learning_rate": 0.00024225433526011558,
+      "loss": 7.0326,
       "step": 1668
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.066473988439306e-05,
-      "loss": 6.6159,
+      "learning_rate": 0.00024216763005780346,
+      "loss": 6.9928,
       "step": 1669
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.063583815028902e-05,
-      "loss": 6.5677,
+      "learning_rate": 0.00024208092485549133,
+      "loss": 6.9526,
       "step": 1670
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.060693641618497e-05,
-      "loss": 6.4999,
+      "learning_rate": 0.00024199421965317918,
+      "loss": 6.9568,
       "step": 1671
     },
     {
       "epoch": 1.87,
-      "learning_rate": 8.057803468208092e-05,
-      "loss": 6.6233,
+      "learning_rate": 0.00024190751445086702,
+      "loss": 7.0319,
       "step": 1672
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.054913294797688e-05,
-      "loss": 6.6633,
+      "learning_rate": 0.0002418208092485549,
+      "loss": 7.0589,
       "step": 1673
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.052023121387283e-05,
-      "loss": 6.6152,
+      "learning_rate": 0.00024173410404624275,
+      "loss": 6.9897,
       "step": 1674
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.049132947976878e-05,
-      "loss": 6.5208,
+      "learning_rate": 0.00024164739884393062,
+      "loss": 6.9293,
       "step": 1675
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.046242774566475e-05,
-      "loss": 6.6739,
+      "learning_rate": 0.00024156069364161847,
+      "loss": 7.0537,
       "step": 1676
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.04335260115607e-05,
-      "loss": 6.5335,
+      "learning_rate": 0.0002414739884393063,
+      "loss": 7.0128,
       "step": 1677
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.040462427745666e-05,
-      "loss": 6.3301,
+      "learning_rate": 0.0002413872832369942,
+      "loss": 6.8318,
       "step": 1678
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.037572254335261e-05,
-      "loss": 6.6887,
+      "learning_rate": 0.00024130057803468206,
+      "loss": 7.0929,
       "step": 1679
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.034682080924855e-05,
-      "loss": 6.5069,
+      "learning_rate": 0.00024121387283236993,
+      "loss": 6.9244,
       "step": 1680
     },
     {
       "epoch": 1.88,
-      "learning_rate": 8.03179190751445e-05,
-      "loss": 6.4786,
+      "learning_rate": 0.00024112716763005778,
+      "loss": 6.9887,
       "step": 1681
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.028901734104046e-05,
-      "loss": 6.3453,
+      "learning_rate": 0.00024104046242774563,
+      "loss": 6.8339,
       "step": 1682
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.026011560693641e-05,
-      "loss": 6.5969,
+      "learning_rate": 0.0002409537572254335,
+      "loss": 7.1025,
       "step": 1683
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.023121387283238e-05,
-      "loss": 6.322,
+      "learning_rate": 0.00024086705202312135,
+      "loss": 6.8479,
       "step": 1684
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.020231213872833e-05,
-      "loss": 6.2841,
+      "learning_rate": 0.00024078034682080922,
+      "loss": 6.8157,
       "step": 1685
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.017341040462429e-05,
-      "loss": 6.2402,
+      "learning_rate": 0.0002406936416184971,
+      "loss": 6.8487,
       "step": 1686
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.014450867052024e-05,
-      "loss": 6.1184,
+      "learning_rate": 0.00024060693641618497,
+      "loss": 6.7669,
       "step": 1687
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.011560693641619e-05,
-      "loss": 6.1559,
+      "learning_rate": 0.00024052023121387282,
+      "loss": 6.8225,
       "step": 1688
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.008670520231215e-05,
-      "loss": 6.1914,
+      "learning_rate": 0.00024043352601156066,
+      "loss": 6.8609,
       "step": 1689
     },
     {
       "epoch": 1.89,
-      "learning_rate": 8.005780346820809e-05,
-      "loss": 6.036,
+      "learning_rate": 0.00024034682080924854,
+      "loss": 6.9033,
       "step": 1690
     },
     {
       "epoch": 1.9,
-      "learning_rate": 8.002890173410404e-05,
-      "loss": 6.1959,
+      "learning_rate": 0.00024026011560693638,
+      "loss": 7.1005,
       "step": 1691
     },
     {
       "epoch": 1.9,
-      "learning_rate": 8e-05,
-      "loss": 5.595,
+      "learning_rate": 0.00024017341040462423,
+      "loss": 6.6242,
       "step": 1692
     },
     {
       "epoch": 1.9,
-      "learning_rate": 7.997109826589596e-05,
-      "loss": 6.8538,
+      "learning_rate": 0.0002400867052023121,
+      "loss": 7.0503,
       "step": 1693
     },
     {
       "epoch": 1.9,
-      "learning_rate": 7.994219653179191e-05,
-      "loss": 6.7909,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 7.0151,
       "step": 1694
     },
     {
       "epoch": 1.9,
-      "learning_rate": 7.991329479768787e-05,
-      "loss": 6.7677,
+      "learning_rate": 0.00023991329479768785,
+      "loss": 7.0076,
       "step": 1695
     },
     {
       "epoch": 1.9,
-      "learning_rate": 7.988439306358382e-05,
-      "loss": 6.7806,
+      "learning_rate": 0.0002398265895953757,
+      "loss": 7.0237,
       "step": 1696
     },
     {
       "epoch": 1.9,
-      "learning_rate": 7.985549132947977e-05,
-      "loss": 6.7222,
+      "learning_rate": 0.00023973988439306357,
+      "loss": 6.9528,
       "step": 1697
     },
     {
       "epoch": 1.9,
-      "learning_rate": 7.982658959537573e-05,
-      "loss": 6.7727,
+      "learning_rate": 0.00023965317919075142,
+      "loss": 7.007,
       "step": 1698
     },
     {
       "epoch": 1.9,
-      "learning_rate": 7.979768786127168e-05,
-      "loss": 6.6317,
+      "learning_rate": 0.00023956647398843927,
+      "loss": 7.0009,
       "step": 1699
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.976878612716763e-05,
-      "loss": 6.5737,
+      "learning_rate": 0.00023947976878612714,
+      "loss": 6.9077,
       "step": 1700
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.973988439306359e-05,
-      "loss": 6.5908,
+      "learning_rate": 0.000239393063583815,
+      "loss": 6.8932,
       "step": 1701
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.971098265895954e-05,
-      "loss": 6.5987,
+      "learning_rate": 0.0002393063583815029,
+      "loss": 6.9496,
       "step": 1702
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.96820809248555e-05,
-      "loss": 6.6611,
+      "learning_rate": 0.00023921965317919074,
+      "loss": 6.9613,
       "step": 1703
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.965317919075145e-05,
-      "loss": 6.6931,
+      "learning_rate": 0.0002391329479768786,
+      "loss": 7.053,
       "step": 1704
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.96242774566474e-05,
-      "loss": 6.5974,
+      "learning_rate": 0.00023904624277456646,
+      "loss": 6.9641,
       "step": 1705
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.959537572254335e-05,
-      "loss": 6.4327,
+      "learning_rate": 0.0002389595375722543,
+      "loss": 6.859,
       "step": 1706
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.956647398843931e-05,
-      "loss": 6.5949,
+      "learning_rate": 0.00023887283236994218,
+      "loss": 6.9843,
       "step": 1707
     },
     {
       "epoch": 1.91,
-      "learning_rate": 7.953757225433528e-05,
-      "loss": 6.5947,
+      "learning_rate": 0.00023878612716763002,
+      "loss": 6.967,
       "step": 1708
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.950867052023122e-05,
-      "loss": 6.6788,
+      "learning_rate": 0.00023869942196531787,
+      "loss": 7.0181,
       "step": 1709
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.947976878612717e-05,
-      "loss": 6.5677,
+      "learning_rate": 0.00023861271676300577,
+      "loss": 6.9311,
       "step": 1710
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.945086705202312e-05,
-      "loss": 6.4585,
+      "learning_rate": 0.00023852601156069362,
+      "loss": 6.8948,
       "step": 1711
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.942196531791908e-05,
-      "loss": 6.561,
+      "learning_rate": 0.0002384393063583815,
+      "loss": 6.9592,
       "step": 1712
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.939306358381503e-05,
-      "loss": 6.5203,
+      "learning_rate": 0.00023835260115606934,
+      "loss": 6.9436,
       "step": 1713
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.936416184971098e-05,
-      "loss": 6.4407,
+      "learning_rate": 0.0002382658959537572,
+      "loss": 6.8512,
       "step": 1714
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.933526011560694e-05,
-      "loss": 6.5137,
+      "learning_rate": 0.00023817919075144506,
+      "loss": 6.9112,
       "step": 1715
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.93063583815029e-05,
-      "loss": 6.6512,
+      "learning_rate": 0.0002380924855491329,
+      "loss": 7.0346,
       "step": 1716
     },
     {
       "epoch": 1.92,
-      "learning_rate": 7.927745664739886e-05,
-      "loss": 6.6595,
+      "learning_rate": 0.00023800578034682078,
+      "loss": 7.0265,
       "step": 1717
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.924855491329481e-05,
-      "loss": 6.5566,
+      "learning_rate": 0.00023791907514450865,
+      "loss": 6.9875,
       "step": 1718
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.921965317919075e-05,
-      "loss": 6.4453,
+      "learning_rate": 0.00023783236994219653,
+      "loss": 6.8476,
       "step": 1719
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.91907514450867e-05,
-      "loss": 6.4223,
+      "learning_rate": 0.00023774566473988437,
+      "loss": 6.8666,
       "step": 1720
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.916184971098266e-05,
-      "loss": 6.5889,
+      "learning_rate": 0.00023765895953757225,
+      "loss": 7.0369,
       "step": 1721
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.913294797687861e-05,
-      "loss": 6.5629,
+      "learning_rate": 0.0002375722543352601,
+      "loss": 6.9955,
       "step": 1722
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.910404624277456e-05,
-      "loss": 6.5191,
+      "learning_rate": 0.00023748554913294794,
+      "loss": 6.9253,
       "step": 1723
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.907514450867053e-05,
-      "loss": 6.4841,
+      "learning_rate": 0.00023739884393063582,
+      "loss": 6.9158,
       "step": 1724
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.904624277456648e-05,
-      "loss": 6.5121,
+      "learning_rate": 0.00023731213872832366,
+      "loss": 6.9954,
       "step": 1725
     },
     {
       "epoch": 1.93,
-      "learning_rate": 7.901734104046244e-05,
-      "loss": 6.3296,
+      "learning_rate": 0.00023722543352601156,
+      "loss": 6.823,
       "step": 1726
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.898843930635839e-05,
-      "loss": 6.5533,
+      "learning_rate": 0.0002371387283236994,
+      "loss": 6.9984,
       "step": 1727
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.895953757225434e-05,
-      "loss": 6.4972,
+      "learning_rate": 0.00023705202312138726,
+      "loss": 6.9739,
       "step": 1728
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.893063583815028e-05,
-      "loss": 6.3754,
+      "learning_rate": 0.00023696531791907513,
+      "loss": 6.8792,
       "step": 1729
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.890173410404624e-05,
-      "loss": 6.5328,
+      "learning_rate": 0.00023687861271676298,
+      "loss": 6.9951,
       "step": 1730
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.887283236994219e-05,
-      "loss": 6.3367,
+      "learning_rate": 0.00023679190751445085,
+      "loss": 6.8736,
       "step": 1731
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.884393063583816e-05,
-      "loss": 6.637,
+      "learning_rate": 0.0002367052023121387,
+      "loss": 7.134,
       "step": 1732
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.881502890173411e-05,
-      "loss": 6.2017,
+      "learning_rate": 0.00023661849710982655,
+      "loss": 6.7975,
       "step": 1733
     },
     {
       "epoch": 1.94,
-      "learning_rate": 7.878612716763006e-05,
-      "loss": 6.2981,
+      "learning_rate": 0.00023653179190751445,
+      "loss": 6.9291,
       "step": 1734
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.875722543352602e-05,
-      "loss": 6.2573,
+      "learning_rate": 0.0002364450867052023,
+      "loss": 6.8966,
       "step": 1735
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.872832369942197e-05,
-      "loss": 6.2135,
+      "learning_rate": 0.00023635838150289017,
+      "loss": 6.9407,
       "step": 1736
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.869942196531793e-05,
-      "loss": 6.1984,
+      "learning_rate": 0.00023627167630057801,
+      "loss": 6.9271,
       "step": 1737
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.867052023121388e-05,
-      "loss": 6.2189,
+      "learning_rate": 0.00023618497109826586,
+      "loss": 6.9397,
       "step": 1738
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.864161849710983e-05,
-      "loss": 6.0018,
+      "learning_rate": 0.00023609826589595373,
+      "loss": 6.9847,
       "step": 1739
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.861271676300579e-05,
-      "loss": 5.9525,
+      "learning_rate": 0.00023601156069364158,
+      "loss": 6.8968,
       "step": 1740
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.858381502890174e-05,
-      "loss": 5.7384,
+      "learning_rate": 0.00023592485549132946,
+      "loss": 6.728,
       "step": 1741
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.855491329479769e-05,
-      "loss": 5.4066,
+      "learning_rate": 0.00023583815028901733,
+      "loss": 6.5276,
       "step": 1742
     },
     {
       "epoch": 1.95,
-      "learning_rate": 7.852601156069365e-05,
-      "loss": 6.802,
+      "learning_rate": 0.0002357514450867052,
+      "loss": 6.9484,
       "step": 1743
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.84971098265896e-05,
-      "loss": 6.7225,
+      "learning_rate": 0.00023566473988439305,
+      "loss": 6.9202,
       "step": 1744
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.846820809248555e-05,
-      "loss": 6.8054,
+      "learning_rate": 0.0002355780346820809,
+      "loss": 7.0376,
       "step": 1745
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.84393063583815e-05,
-      "loss": 6.7388,
+      "learning_rate": 0.00023549132947976877,
+      "loss": 7.0005,
       "step": 1746
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.841040462427746e-05,
-      "loss": 6.6563,
+      "learning_rate": 0.00023540462427745662,
+      "loss": 6.9247,
       "step": 1747
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.838150289017341e-05,
-      "loss": 6.5882,
+      "learning_rate": 0.0002353179190751445,
+      "loss": 6.8978,
       "step": 1748
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.835260115606937e-05,
-      "loss": 6.6126,
+      "learning_rate": 0.00023523121387283234,
+      "loss": 6.9395,
       "step": 1749
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.832369942196532e-05,
-      "loss": 6.779,
+      "learning_rate": 0.00023514450867052024,
+      "loss": 7.1175,
       "step": 1750
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.829479768786127e-05,
-      "loss": 6.5665,
+      "learning_rate": 0.00023505780346820809,
+      "loss": 6.9103,
       "step": 1751
     },
     {
       "epoch": 1.96,
-      "learning_rate": 7.826589595375723e-05,
-      "loss": 6.6469,
+      "learning_rate": 0.00023497109826589593,
+      "loss": 7.0171,
       "step": 1752
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.823699421965318e-05,
-      "loss": 6.5927,
+      "learning_rate": 0.0002348843930635838,
+      "loss": 6.9861,
       "step": 1753
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.820809248554913e-05,
-      "loss": 6.5821,
+      "learning_rate": 0.00023479768786127165,
+      "loss": 6.9843,
       "step": 1754
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.817919075144509e-05,
-      "loss": 6.3997,
+      "learning_rate": 0.0002347109826589595,
+      "loss": 6.8742,
       "step": 1755
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.815028901734105e-05,
-      "loss": 6.443,
+      "learning_rate": 0.00023462427745664737,
+      "loss": 6.8891,
       "step": 1756
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.812138728323701e-05,
-      "loss": 6.5746,
+      "learning_rate": 0.00023453757225433522,
+      "loss": 6.9845,
       "step": 1757
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.809248554913296e-05,
-      "loss": 6.5475,
+      "learning_rate": 0.00023445086705202312,
+      "loss": 6.9796,
       "step": 1758
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.80635838150289e-05,
-      "loss": 6.6029,
+      "learning_rate": 0.00023436416184971097,
+      "loss": 7.0518,
       "step": 1759
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.803468208092485e-05,
-      "loss": 6.3671,
+      "learning_rate": 0.00023427745664739884,
+      "loss": 6.856,
       "step": 1760
     },
     {
       "epoch": 1.97,
-      "learning_rate": 7.800578034682081e-05,
-      "loss": 6.6155,
+      "learning_rate": 0.0002341907514450867,
+      "loss": 7.1036,
       "step": 1761
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.797687861271676e-05,
-      "loss": 6.4387,
+      "learning_rate": 0.00023410404624277454,
+      "loss": 6.8856,
       "step": 1762
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.794797687861271e-05,
-      "loss": 6.5478,
+      "learning_rate": 0.0002340173410404624,
+      "loss": 6.9882,
       "step": 1763
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.791907514450868e-05,
-      "loss": 6.478,
+      "learning_rate": 0.00023393063583815026,
+      "loss": 7.0057,
       "step": 1764
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.789017341040464e-05,
-      "loss": 6.3784,
+      "learning_rate": 0.0002338439306358381,
+      "loss": 6.8437,
       "step": 1765
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.786127167630059e-05,
-      "loss": 6.5051,
+      "learning_rate": 0.000233757225433526,
+      "loss": 6.9706,
       "step": 1766
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.783236994219654e-05,
-      "loss": 6.4688,
+      "learning_rate": 0.00023367052023121385,
+      "loss": 6.9585,
       "step": 1767
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.78034682080925e-05,
-      "loss": 6.5163,
+      "learning_rate": 0.00023358381502890173,
+      "loss": 6.9956,
       "step": 1768
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.777456647398844e-05,
-      "loss": 6.3623,
+      "learning_rate": 0.00023349710982658957,
+      "loss": 6.9121,
       "step": 1769
     },
     {
       "epoch": 1.98,
-      "learning_rate": 7.774566473988439e-05,
-      "loss": 6.5174,
+      "learning_rate": 0.00023341040462427745,
+      "loss": 7.0314,
       "step": 1770
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.771676300578034e-05,
-      "loss": 6.3664,
+      "learning_rate": 0.0002333236994219653,
+      "loss": 6.9066,
       "step": 1771
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.76878612716763e-05,
-      "loss": 6.3158,
+      "learning_rate": 0.00023323699421965314,
+      "loss": 6.9294,
       "step": 1772
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.765895953757226e-05,
-      "loss": 6.3861,
+      "learning_rate": 0.000233150289017341,
+      "loss": 6.9553,
       "step": 1773
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.763005780346822e-05,
-      "loss": 6.4911,
+      "learning_rate": 0.00023306358381502886,
+      "loss": 7.0628,
       "step": 1774
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.760115606936417e-05,
-      "loss": 6.4003,
+      "learning_rate": 0.00023297687861271676,
+      "loss": 6.9864,
       "step": 1775
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.757225433526012e-05,
-      "loss": 6.4447,
+      "learning_rate": 0.0002328901734104046,
+      "loss": 6.9753,
       "step": 1776
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.754335260115608e-05,
-      "loss": 6.4715,
+      "learning_rate": 0.00023280346820809248,
+      "loss": 7.1013,
       "step": 1777
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.751445086705203e-05,
-      "loss": 6.3629,
+      "learning_rate": 0.00023271676300578033,
+      "loss": 6.9831,
       "step": 1778
     },
     {
       "epoch": 1.99,
-      "learning_rate": 7.748554913294797e-05,
-      "loss": 6.25,
+      "learning_rate": 0.00023263005780346818,
+      "loss": 6.9113,
       "step": 1779
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.745664739884392e-05,
-      "loss": 6.0744,
+      "learning_rate": 0.00023254335260115605,
+      "loss": 6.7931,
       "step": 1780
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.742774566473989e-05,
-      "loss": 6.0143,
+      "learning_rate": 0.0002324566473988439,
+      "loss": 6.8891,
       "step": 1781
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.739884393063584e-05,
-      "loss": 6.156,
+      "learning_rate": 0.00023236994219653174,
+      "loss": 7.0188,
       "step": 1782
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.73699421965318e-05,
-      "loss": 5.9855,
+      "learning_rate": 0.00023228323699421964,
+      "loss": 6.981,
       "step": 1783
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.734104046242775e-05,
-      "loss": 5.2569,
+      "learning_rate": 0.0002321965317919075,
+      "loss": 6.5137,
       "step": 1784
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.73121387283237e-05,
-      "loss": 6.8333,
+      "learning_rate": 0.00023210982658959536,
+      "loss": 6.993,
       "step": 1785
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.728323699421966e-05,
-      "loss": 6.522,
+      "learning_rate": 0.0002320231213872832,
+      "loss": 6.7281,
       "step": 1786
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.725433526011561e-05,
-      "loss": 6.5269,
+      "learning_rate": 0.00023193641618497109,
+      "loss": 6.8715,
       "step": 1787
     },
     {
       "epoch": 2.0,
-      "learning_rate": 7.722543352601156e-05,
-      "loss": 6.6433,
+      "learning_rate": 0.00023184971098265893,
+      "loss": 6.9592,
       "step": 1788
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.719653179190752e-05,
-      "loss": 6.613,
+      "learning_rate": 0.00023176300578034678,
+      "loss": 6.9045,
       "step": 1789
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.716763005780347e-05,
-      "loss": 6.526,
+      "learning_rate": 0.00023167630057803465,
+      "loss": 6.9014,
       "step": 1790
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.713872832369942e-05,
-      "loss": 6.6175,
+      "learning_rate": 0.00023158959537572253,
+      "loss": 7.0274,
       "step": 1791
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.710982658959538e-05,
-      "loss": 6.5773,
+      "learning_rate": 0.0002315028901734104,
+      "loss": 7.0005,
       "step": 1792
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.708092485549133e-05,
-      "loss": 6.536,
+      "learning_rate": 0.00023141618497109825,
+      "loss": 6.8884,
       "step": 1793
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.705202312138729e-05,
-      "loss": 6.5446,
+      "learning_rate": 0.00023132947976878612,
+      "loss": 6.9422,
       "step": 1794
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.702312138728324e-05,
-      "loss": 6.6292,
+      "learning_rate": 0.00023124277456647397,
+      "loss": 7.0111,
       "step": 1795
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.699421965317919e-05,
-      "loss": 6.5763,
+      "learning_rate": 0.00023115606936416181,
+      "loss": 7.0015,
       "step": 1796
     },
     {
       "epoch": 2.01,
-      "learning_rate": 7.696531791907516e-05,
-      "loss": 6.4727,
+      "learning_rate": 0.0002310693641618497,
+      "loss": 6.935,
       "step": 1797
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.69364161849711e-05,
-      "loss": 6.4519,
+      "learning_rate": 0.00023098265895953754,
+      "loss": 6.9014,
       "step": 1798
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.690751445086705e-05,
-      "loss": 6.4768,
+      "learning_rate": 0.00023089595375722544,
+      "loss": 6.9654,
       "step": 1799
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.6878612716763e-05,
-      "loss": 6.4311,
+      "learning_rate": 0.00023080924855491328,
+      "loss": 6.9591,
       "step": 1800
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.684971098265896e-05,
-      "loss": 6.6298,
+      "learning_rate": 0.00023072254335260113,
+      "loss": 7.0417,
       "step": 1801
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.682080924855491e-05,
-      "loss": 6.6922,
+      "learning_rate": 0.000230635838150289,
+      "loss": 7.2036,
       "step": 1802
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.679190751445087e-05,
-      "loss": 6.4724,
+      "learning_rate": 0.00023054913294797685,
+      "loss": 6.9441,
       "step": 1803
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.676300578034682e-05,
-      "loss": 6.3926,
+      "learning_rate": 0.00023046242774566472,
+      "loss": 6.9382,
       "step": 1804
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.673410404624279e-05,
-      "loss": 6.3586,
+      "learning_rate": 0.00023037572254335257,
+      "loss": 6.8796,
       "step": 1805
     },
     {
       "epoch": 2.02,
-      "learning_rate": 7.670520231213874e-05,
-      "loss": 6.5056,
+      "learning_rate": 0.00023028901734104042,
+      "loss": 6.97,
       "step": 1806
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.66763005780347e-05,
-      "loss": 6.4388,
+      "learning_rate": 0.00023020231213872832,
+      "loss": 6.9688,
       "step": 1807
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.664739884393063e-05,
-      "loss": 6.3845,
+      "learning_rate": 0.00023011560693641617,
+      "loss": 6.947,
       "step": 1808
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.661849710982659e-05,
-      "loss": 6.3525,
+      "learning_rate": 0.00023002890173410404,
+      "loss": 6.8628,
       "step": 1809
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.658959537572254e-05,
-      "loss": 6.3826,
+      "learning_rate": 0.0002299421965317919,
+      "loss": 6.9312,
       "step": 1810
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.65606936416185e-05,
-      "loss": 6.3008,
+      "learning_rate": 0.00022985549132947973,
+      "loss": 6.8596,
       "step": 1811
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.653179190751445e-05,
-      "loss": 6.3169,
+      "learning_rate": 0.0002297687861271676,
+      "loss": 6.9214,
       "step": 1812
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.650289017341041e-05,
-      "loss": 6.4131,
+      "learning_rate": 0.00022968208092485545,
+      "loss": 6.9912,
       "step": 1813
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.647398843930637e-05,
-      "loss": 6.43,
+      "learning_rate": 0.00022959537572254333,
+      "loss": 6.9941,
       "step": 1814
     },
     {
       "epoch": 2.03,
-      "learning_rate": 7.644508670520232e-05,
-      "loss": 6.4502,
+      "learning_rate": 0.0002295086705202312,
+      "loss": 7.0595,
       "step": 1815
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.641618497109827e-05,
-      "loss": 6.3812,
+      "learning_rate": 0.00022942196531791908,
+      "loss": 7.0104,
       "step": 1816
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.638728323699423e-05,
-      "loss": 6.3319,
+      "learning_rate": 0.00022933526011560692,
+      "loss": 6.9441,
       "step": 1817
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.635838150289018e-05,
-      "loss": 6.275,
+      "learning_rate": 0.00022924855491329477,
+      "loss": 6.9223,
       "step": 1818
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.632947976878612e-05,
-      "loss": 6.3185,
+      "learning_rate": 0.00022916184971098264,
+      "loss": 6.9249,
       "step": 1819
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.630057803468207e-05,
-      "loss": 6.1381,
+      "learning_rate": 0.0002290751445086705,
+      "loss": 6.771,
       "step": 1820
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.627167630057804e-05,
-      "loss": 6.361,
+      "learning_rate": 0.00022898843930635836,
+      "loss": 6.938,
       "step": 1821
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.6242774566474e-05,
-      "loss": 6.3287,
+      "learning_rate": 0.0002289017341040462,
+      "loss": 6.9295,
       "step": 1822
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.621387283236995e-05,
-      "loss": 6.1816,
+      "learning_rate": 0.0002288150289017341,
+      "loss": 6.8929,
       "step": 1823
     },
     {
       "epoch": 2.04,
-      "learning_rate": 7.61849710982659e-05,
-      "loss": 6.3576,
+      "learning_rate": 0.00022872832369942196,
+      "loss": 6.959,
       "step": 1824
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.615606936416186e-05,
-      "loss": 6.2422,
+      "learning_rate": 0.0002286416184971098,
+      "loss": 6.9323,
       "step": 1825
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.612716763005781e-05,
-      "loss": 6.4353,
+      "learning_rate": 0.00022855491329479768,
+      "loss": 7.0665,
       "step": 1826
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.609826589595376e-05,
-      "loss": 6.0329,
+      "learning_rate": 0.00022846820809248553,
+      "loss": 6.8507,
       "step": 1827
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.606936416184972e-05,
-      "loss": 6.1442,
+      "learning_rate": 0.00022838150289017337,
+      "loss": 7.0056,
       "step": 1828
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.604046242774567e-05,
-      "loss": 5.969,
+      "learning_rate": 0.00022829479768786125,
+      "loss": 6.8784,
       "step": 1829
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.601156069364162e-05,
-      "loss": 6.0069,
+      "learning_rate": 0.0002282080924855491,
+      "loss": 6.9211,
       "step": 1830
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.598265895953758e-05,
-      "loss": 5.7755,
+      "learning_rate": 0.000228121387283237,
+      "loss": 6.8468,
       "step": 1831
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.595375722543353e-05,
-      "loss": 5.7838,
+      "learning_rate": 0.00022803468208092484,
+      "loss": 6.821,
       "step": 1832
     },
     {
       "epoch": 2.05,
-      "learning_rate": 7.592485549132948e-05,
-      "loss": 5.5654,
+      "learning_rate": 0.00022794797687861272,
+      "loss": 6.7606,
       "step": 1833
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.589595375722544e-05,
-      "loss": 5.2373,
+      "learning_rate": 0.00022786127167630056,
+      "loss": 6.5116,
       "step": 1834
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.586705202312139e-05,
-      "loss": 6.911,
+      "learning_rate": 0.0002277745664739884,
+      "loss": 7.0452,
       "step": 1835
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.583815028901734e-05,
-      "loss": 6.9098,
+      "learning_rate": 0.00022768786127167628,
+      "loss": 6.9866,
       "step": 1836
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.58092485549133e-05,
-      "loss": 6.7005,
+      "learning_rate": 0.00022760115606936413,
+      "loss": 6.8646,
       "step": 1837
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.578034682080925e-05,
-      "loss": 6.7287,
+      "learning_rate": 0.00022751445086705198,
+      "loss": 6.9724,
       "step": 1838
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.57514450867052e-05,
-      "loss": 6.7345,
+      "learning_rate": 0.00022742774566473988,
+      "loss": 7.0192,
       "step": 1839
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.572254335260116e-05,
-      "loss": 6.5152,
+      "learning_rate": 0.00022734104046242772,
+      "loss": 6.8173,
       "step": 1840
     },
     {
       "epoch": 2.06,
-      "learning_rate": 7.569364161849711e-05,
-      "loss": 6.7042,
+      "learning_rate": 0.0002272543352601156,
+      "loss": 7.0313,
       "step": 1841
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.566473988439306e-05,
-      "loss": 6.5204,
+      "learning_rate": 0.00022716763005780344,
+      "loss": 6.902,
       "step": 1842
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.563583815028902e-05,
-      "loss": 6.383,
+      "learning_rate": 0.00022708092485549132,
+      "loss": 6.8537,
       "step": 1843
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.560693641618497e-05,
-      "loss": 6.6022,
+      "learning_rate": 0.00022699421965317917,
+      "loss": 6.9835,
       "step": 1844
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.557803468208094e-05,
-      "loss": 6.5682,
+      "learning_rate": 0.000226907514450867,
+      "loss": 7.0436,
       "step": 1845
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.554913294797689e-05,
-      "loss": 6.4928,
+      "learning_rate": 0.00022682080924855489,
+      "loss": 6.9327,
       "step": 1846
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.552023121387284e-05,
-      "loss": 6.44,
+      "learning_rate": 0.00022673410404624276,
+      "loss": 6.877,
       "step": 1847
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.549132947976878e-05,
-      "loss": 6.3496,
+      "learning_rate": 0.00022664739884393063,
+      "loss": 6.8539,
       "step": 1848
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.546242774566474e-05,
-      "loss": 6.5562,
+      "learning_rate": 0.00022656069364161848,
+      "loss": 7.0094,
       "step": 1849
     },
     {
       "epoch": 2.07,
-      "learning_rate": 7.543352601156069e-05,
-      "loss": 6.361,
+      "learning_rate": 0.00022647398843930635,
+      "loss": 6.8663,
       "step": 1850
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.540462427745665e-05,
-      "loss": 6.6142,
+      "learning_rate": 0.0002263872832369942,
+      "loss": 7.0956,
       "step": 1851
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.53757225433526e-05,
-      "loss": 6.3927,
+      "learning_rate": 0.00022630057803468205,
+      "loss": 6.8403,
       "step": 1852
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.534682080924857e-05,
-      "loss": 6.5854,
+      "learning_rate": 0.00022621387283236992,
+      "loss": 7.0661,
       "step": 1853
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.531791907514452e-05,
-      "loss": 6.5257,
+      "learning_rate": 0.00022612716763005777,
+      "loss": 6.9704,
       "step": 1854
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.528901734104047e-05,
-      "loss": 6.6462,
+      "learning_rate": 0.00022604046242774567,
+      "loss": 7.0911,
       "step": 1855
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.526011560693643e-05,
-      "loss": 6.5538,
+      "learning_rate": 0.00022595375722543352,
+      "loss": 7.0103,
       "step": 1856
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.523121387283238e-05,
-      "loss": 6.4782,
+      "learning_rate": 0.00022586705202312136,
+      "loss": 6.9775,
       "step": 1857
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.520231213872832e-05,
-      "loss": 6.305,
+      "learning_rate": 0.00022578034682080924,
+      "loss": 6.8844,
       "step": 1858
     },
     {
       "epoch": 2.08,
-      "learning_rate": 7.517341040462427e-05,
-      "loss": 6.5978,
+      "learning_rate": 0.00022569364161849708,
+      "loss": 7.1201,
       "step": 1859
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.514450867052023e-05,
-      "loss": 6.4158,
+      "learning_rate": 0.00022560693641618496,
+      "loss": 6.9914,
       "step": 1860
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.511560693641619e-05,
-      "loss": 6.5264,
+      "learning_rate": 0.0002255202312138728,
+      "loss": 7.0532,
       "step": 1861
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.508670520231215e-05,
-      "loss": 6.4108,
+      "learning_rate": 0.00022543352601156065,
+      "loss": 6.9403,
       "step": 1862
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.50578034682081e-05,
-      "loss": 6.2327,
+      "learning_rate": 0.00022534682080924855,
+      "loss": 6.8468,
       "step": 1863
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.502890173410405e-05,
-      "loss": 6.3773,
+      "learning_rate": 0.0002252601156069364,
+      "loss": 6.9925,
       "step": 1864
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.500000000000001e-05,
-      "loss": 6.432,
+      "learning_rate": 0.00022517341040462427,
+      "loss": 6.969,
       "step": 1865
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.497109826589596e-05,
-      "loss": 6.2696,
+      "learning_rate": 0.00022508670520231212,
+      "loss": 6.9127,
       "step": 1866
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.494219653179191e-05,
-      "loss": 6.4304,
+      "learning_rate": 0.000225,
+      "loss": 7.0342,
       "step": 1867
     },
     {
       "epoch": 2.09,
-      "learning_rate": 7.491329479768787e-05,
-      "loss": 6.3428,
+      "learning_rate": 0.00022491329479768784,
+      "loss": 7.0122,
       "step": 1868
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.488439306358381e-05,
-      "loss": 6.3769,
+      "learning_rate": 0.0002248265895953757,
+      "loss": 6.9824,
       "step": 1869
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.485549132947977e-05,
-      "loss": 6.1863,
+      "learning_rate": 0.00022473988439306356,
+      "loss": 6.9216,
       "step": 1870
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.482658959537573e-05,
-      "loss": 6.434,
+      "learning_rate": 0.0002246531791907514,
+      "loss": 7.0445,
       "step": 1871
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.479768786127168e-05,
-      "loss": 6.2222,
+      "learning_rate": 0.0002245664739884393,
+      "loss": 6.8895,
       "step": 1872
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.476878612716763e-05,
-      "loss": 6.372,
+      "learning_rate": 0.00022447976878612716,
+      "loss": 7.0285,
       "step": 1873
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.473988439306359e-05,
-      "loss": 6.3455,
+      "learning_rate": 0.000224393063583815,
+      "loss": 7.0373,
       "step": 1874
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.471098265895954e-05,
-      "loss": 6.2814,
+      "learning_rate": 0.00022430635838150288,
+      "loss": 7.028,
       "step": 1875
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.46820809248555e-05,
-      "loss": 6.3219,
+      "learning_rate": 0.00022421965317919072,
+      "loss": 7.0466,
       "step": 1876
     },
     {
       "epoch": 2.1,
-      "learning_rate": 7.465317919075145e-05,
-      "loss": 6.1849,
+      "learning_rate": 0.0002241329479768786,
+      "loss": 6.9411,
       "step": 1877
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.46242774566474e-05,
-      "loss": 5.9726,
+      "learning_rate": 0.00022404624277456644,
+      "loss": 6.7512,
       "step": 1878
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.459537572254336e-05,
-      "loss": 6.2458,
+      "learning_rate": 0.0002239595375722543,
+      "loss": 7.0301,
       "step": 1879
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.456647398843931e-05,
-      "loss": 5.9725,
+      "learning_rate": 0.0002238728323699422,
+      "loss": 6.8877,
       "step": 1880
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.453757225433526e-05,
-      "loss": 5.9462,
+      "learning_rate": 0.00022378612716763004,
+      "loss": 6.8896,
       "step": 1881
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.450867052023122e-05,
-      "loss": 5.9272,
+      "learning_rate": 0.0002236994219653179,
+      "loss": 6.9464,
       "step": 1882
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.447976878612717e-05,
-      "loss": 5.34,
+      "learning_rate": 0.00022361271676300576,
+      "loss": 6.6682,
       "step": 1883
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.445086705202312e-05,
-      "loss": 5.1677,
+      "learning_rate": 0.0002235260115606936,
+      "loss": 6.4918,
       "step": 1884
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.442196531791908e-05,
-      "loss": 6.8453,
+      "learning_rate": 0.00022343930635838148,
+      "loss": 7.0318,
       "step": 1885
     },
     {
       "epoch": 2.11,
-      "learning_rate": 7.439306358381504e-05,
-      "loss": 6.7216,
+      "learning_rate": 0.00022335260115606933,
+      "loss": 6.9892,
       "step": 1886
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.436416184971098e-05,
-      "loss": 6.8261,
+      "learning_rate": 0.0002232658959537572,
+      "loss": 7.0593,
       "step": 1887
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.433526011560694e-05,
-      "loss": 6.6394,
+      "learning_rate": 0.00022317919075144507,
+      "loss": 6.946,
       "step": 1888
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.430635838150289e-05,
-      "loss": 6.6184,
+      "learning_rate": 0.00022309248554913295,
+      "loss": 6.9782,
       "step": 1889
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.427745664739884e-05,
-      "loss": 6.6614,
+      "learning_rate": 0.0002230057803468208,
+      "loss": 7.0332,
       "step": 1890
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.42485549132948e-05,
-      "loss": 6.5641,
+      "learning_rate": 0.00022291907514450864,
+      "loss": 6.8972,
       "step": 1891
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.421965317919075e-05,
-      "loss": 6.5653,
+      "learning_rate": 0.00022283236994219652,
+      "loss": 6.9552,
       "step": 1892
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.41907514450867e-05,
-      "loss": 6.7867,
+      "learning_rate": 0.00022274566473988436,
+      "loss": 7.1819,
       "step": 1893
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.416184971098267e-05,
-      "loss": 6.5897,
+      "learning_rate": 0.00022265895953757224,
+      "loss": 7.0481,
       "step": 1894
     },
     {
       "epoch": 2.12,
-      "learning_rate": 7.413294797687862e-05,
-      "loss": 6.3322,
+      "learning_rate": 0.00022257225433526008,
+      "loss": 6.8457,
       "step": 1895
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.410404624277458e-05,
-      "loss": 6.5521,
+      "learning_rate": 0.00022248554913294798,
+      "loss": 7.0342,
       "step": 1896
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.407514450867053e-05,
-      "loss": 6.3439,
+      "learning_rate": 0.00022239884393063583,
+      "loss": 6.8742,
       "step": 1897
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.404624277456647e-05,
-      "loss": 6.3986,
+      "learning_rate": 0.00022231213872832368,
+      "loss": 6.9142,
       "step": 1898
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.401734104046242e-05,
-      "loss": 6.333,
+      "learning_rate": 0.00022222543352601155,
+      "loss": 6.9059,
       "step": 1899
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.398843930635838e-05,
-      "loss": 6.2886,
+      "learning_rate": 0.0002221387283236994,
+      "loss": 6.8394,
       "step": 1900
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.395953757225433e-05,
-      "loss": 6.4489,
+      "learning_rate": 0.00022205202312138725,
+      "loss": 6.9607,
       "step": 1901
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.39306358381503e-05,
-      "loss": 6.5433,
+      "learning_rate": 0.00022196531791907512,
+      "loss": 7.023,
       "step": 1902
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.390173410404625e-05,
-      "loss": 6.5201,
+      "learning_rate": 0.00022187861271676297,
+      "loss": 7.0328,
       "step": 1903
     },
     {
       "epoch": 2.13,
-      "learning_rate": 7.38728323699422e-05,
-      "loss": 6.4127,
+      "learning_rate": 0.00022179190751445087,
+      "loss": 6.9699,
       "step": 1904
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.384393063583816e-05,
-      "loss": 6.3835,
+      "learning_rate": 0.00022170520231213871,
+      "loss": 6.9482,
       "step": 1905
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.381502890173411e-05,
-      "loss": 6.4887,
+      "learning_rate": 0.0002216184971098266,
+      "loss": 7.0225,
       "step": 1906
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.378612716763007e-05,
-      "loss": 6.3742,
+      "learning_rate": 0.00022153179190751443,
+      "loss": 6.9578,
       "step": 1907
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.3757225433526e-05,
-      "loss": 6.48,
+      "learning_rate": 0.00022144508670520228,
+      "loss": 7.0362,
       "step": 1908
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.372832369942196e-05,
-      "loss": 6.3472,
+      "learning_rate": 0.00022135838150289016,
+      "loss": 6.9372,
       "step": 1909
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.369942196531793e-05,
-      "loss": 6.3956,
+      "learning_rate": 0.000221271676300578,
+      "loss": 6.9914,
       "step": 1910
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.367052023121388e-05,
-      "loss": 6.4029,
+      "learning_rate": 0.00022118497109826585,
+      "loss": 6.9883,
       "step": 1911
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.364161849710983e-05,
-      "loss": 6.5008,
+      "learning_rate": 0.00022109826589595375,
+      "loss": 7.0744,
       "step": 1912
     },
     {
       "epoch": 2.14,
-      "learning_rate": 7.361271676300579e-05,
-      "loss": 6.4262,
+      "learning_rate": 0.0002210115606936416,
+      "loss": 7.0366,
       "step": 1913
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.358381502890174e-05,
-      "loss": 6.2449,
+      "learning_rate": 0.00022092485549132947,
+      "loss": 6.9272,
       "step": 1914
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.355491329479769e-05,
-      "loss": 6.2643,
+      "learning_rate": 0.00022083815028901732,
+      "loss": 6.85,
       "step": 1915
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.352601156069365e-05,
-      "loss": 6.3233,
+      "learning_rate": 0.0002207514450867052,
+      "loss": 6.9743,
       "step": 1916
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.34971098265896e-05,
-      "loss": 6.2594,
+      "learning_rate": 0.00022066473988439304,
+      "loss": 6.862,
       "step": 1917
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.346820809248555e-05,
-      "loss": 6.2239,
+      "learning_rate": 0.00022057803468208088,
+      "loss": 6.913,
       "step": 1918
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.34393063583815e-05,
-      "loss": 6.3418,
+      "learning_rate": 0.00022049132947976876,
+      "loss": 7.0248,
       "step": 1919
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.341040462427746e-05,
-      "loss": 6.4499,
+      "learning_rate": 0.00022040462427745663,
+      "loss": 7.1144,
       "step": 1920
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.338150289017341e-05,
-      "loss": 6.1532,
+      "learning_rate": 0.0002203179190751445,
+      "loss": 6.8356,
       "step": 1921
     },
     {
       "epoch": 2.15,
-      "learning_rate": 7.335260115606937e-05,
-      "loss": 6.2167,
+      "learning_rate": 0.00022023121387283235,
+      "loss": 6.9588,
       "step": 1922
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.332369942196532e-05,
-      "loss": 6.1816,
+      "learning_rate": 0.00022014450867052023,
+      "loss": 6.9329,
       "step": 1923
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.329479768786127e-05,
-      "loss": 6.0652,
+      "learning_rate": 0.00022005780346820807,
+      "loss": 6.8594,
       "step": 1924
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.326589595375723e-05,
-      "loss": 6.3583,
+      "learning_rate": 0.00021997109826589592,
+      "loss": 7.1572,
       "step": 1925
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.32369942196532e-05,
-      "loss": 6.063,
+      "learning_rate": 0.0002198843930635838,
+      "loss": 6.8971,
       "step": 1926
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.320809248554913e-05,
-      "loss": 5.9025,
+      "learning_rate": 0.00021979768786127164,
+      "loss": 6.7865,
       "step": 1927
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.317919075144509e-05,
-      "loss": 5.8001,
+      "learning_rate": 0.00021971098265895954,
+      "loss": 6.8031,
       "step": 1928
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.315028901734104e-05,
-      "loss": 5.7516,
+      "learning_rate": 0.0002196242774566474,
+      "loss": 6.7694,
       "step": 1929
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.3121387283237e-05,
-      "loss": 5.8168,
+      "learning_rate": 0.00021953757225433524,
+      "loss": 6.8757,
       "step": 1930
     },
     {
       "epoch": 2.16,
-      "learning_rate": 7.309248554913295e-05,
-      "loss": 5.5822,
+      "learning_rate": 0.0002194508670520231,
+      "loss": 6.7752,
       "step": 1931
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.30635838150289e-05,
-      "loss": 5.3009,
+      "learning_rate": 0.00021936416184971096,
+      "loss": 6.5619,
       "step": 1932
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.303468208092485e-05,
-      "loss": 5.5696,
+      "learning_rate": 0.00021927745664739883,
+      "loss": 6.9041,
       "step": 1933
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.300578034682082e-05,
-      "loss": 5.4071,
+      "learning_rate": 0.00021919075144508668,
+      "loss": 6.8843,
       "step": 1934
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.297687861271678e-05,
-      "loss": 6.9424,
+      "learning_rate": 0.00021910404624277452,
+      "loss": 7.0508,
       "step": 1935
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.294797687861273e-05,
-      "loss": 6.8129,
+      "learning_rate": 0.00021901734104046243,
+      "loss": 6.9832,
       "step": 1936
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.291907514450867e-05,
-      "loss": 6.6447,
+      "learning_rate": 0.00021893063583815027,
+      "loss": 6.9409,
       "step": 1937
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.289017341040462e-05,
-      "loss": 6.4993,
+      "learning_rate": 0.00021884393063583815,
+      "loss": 6.8815,
       "step": 1938
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.286127167630058e-05,
-      "loss": 6.6199,
+      "learning_rate": 0.000218757225433526,
+      "loss": 6.9649,
       "step": 1939
     },
     {
       "epoch": 2.17,
-      "learning_rate": 7.283236994219653e-05,
-      "loss": 6.779,
+      "learning_rate": 0.00021867052023121384,
+      "loss": 7.1136,
       "step": 1940
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.280346820809248e-05,
-      "loss": 6.5994,
+      "learning_rate": 0.0002185838150289017,
+      "loss": 6.9927,
       "step": 1941
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.277456647398845e-05,
-      "loss": 6.4797,
+      "learning_rate": 0.00021849710982658956,
+      "loss": 6.916,
       "step": 1942
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.27456647398844e-05,
-      "loss": 6.4113,
+      "learning_rate": 0.00021841040462427743,
+      "loss": 6.9027,
       "step": 1943
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.271676300578036e-05,
-      "loss": 6.2949,
+      "learning_rate": 0.0002183236994219653,
+      "loss": 6.9012,
       "step": 1944
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.268786127167631e-05,
-      "loss": 6.5589,
+      "learning_rate": 0.00021823699421965318,
+      "loss": 7.1131,
       "step": 1945
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.265895953757226e-05,
-      "loss": 6.402,
+      "learning_rate": 0.00021815028901734103,
+      "loss": 6.8669,
       "step": 1946
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.263005780346822e-05,
-      "loss": 6.3614,
+      "learning_rate": 0.00021806358381502888,
+      "loss": 6.906,
       "step": 1947
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.260115606936416e-05,
-      "loss": 6.2891,
+      "learning_rate": 0.00021797687861271675,
+      "loss": 6.8536,
       "step": 1948
     },
     {
       "epoch": 2.18,
-      "learning_rate": 7.257225433526011e-05,
-      "loss": 6.6374,
+      "learning_rate": 0.0002178901734104046,
+      "loss": 7.1827,
       "step": 1949
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.254335260115608e-05,
-      "loss": 6.3932,
+      "learning_rate": 0.00021780346820809247,
+      "loss": 6.9994,
       "step": 1950
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.251445086705203e-05,
-      "loss": 6.2689,
+      "learning_rate": 0.00021771676300578032,
+      "loss": 6.8975,
       "step": 1951
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.248554913294798e-05,
-      "loss": 6.2324,
+      "learning_rate": 0.00021763005780346822,
+      "loss": 6.8962,
       "step": 1952
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.245664739884394e-05,
-      "loss": 6.3081,
+      "learning_rate": 0.00021754335260115606,
+      "loss": 6.8902,
       "step": 1953
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.242774566473989e-05,
-      "loss": 6.4842,
+      "learning_rate": 0.0002174566473988439,
+      "loss": 7.0856,
       "step": 1954
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.239884393063584e-05,
-      "loss": 6.3772,
+      "learning_rate": 0.00021736994219653179,
+      "loss": 7.0322,
       "step": 1955
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.23699421965318e-05,
-      "loss": 6.401,
+      "learning_rate": 0.00021728323699421963,
+      "loss": 6.995,
       "step": 1956
     },
     {
       "epoch": 2.19,
-      "learning_rate": 7.234104046242775e-05,
-      "loss": 6.3543,
+      "learning_rate": 0.00021719653179190748,
+      "loss": 6.9974,
       "step": 1957
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.23121387283237e-05,
-      "loss": 6.2597,
+      "learning_rate": 0.00021710982658959535,
+      "loss": 6.895,
       "step": 1958
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.228323699421966e-05,
-      "loss": 6.1583,
+      "learning_rate": 0.0002170231213872832,
+      "loss": 6.8384,
       "step": 1959
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.225433526011561e-05,
-      "loss": 6.4219,
+      "learning_rate": 0.0002169364161849711,
+      "loss": 7.0707,
       "step": 1960
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.222543352601156e-05,
-      "loss": 6.2022,
+      "learning_rate": 0.00021684971098265895,
+      "loss": 6.9165,
       "step": 1961
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.219653179190752e-05,
-      "loss": 6.2095,
+      "learning_rate": 0.00021676300578034682,
+      "loss": 6.9072,
       "step": 1962
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.216763005780347e-05,
-      "loss": 6.2845,
+      "learning_rate": 0.00021667630057803467,
+      "loss": 6.9945,
       "step": 1963
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.213872832369943e-05,
-      "loss": 6.1732,
+      "learning_rate": 0.00021658959537572251,
+      "loss": 6.9131,
       "step": 1964
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.210982658959538e-05,
-      "loss": 6.1975,
+      "learning_rate": 0.0002165028901734104,
+      "loss": 6.9226,
       "step": 1965
     },
     {
       "epoch": 2.2,
-      "learning_rate": 7.208092485549133e-05,
-      "loss": 6.2277,
+      "learning_rate": 0.00021641618497109824,
+      "loss": 6.941,
       "step": 1966
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.205202312138729e-05,
-      "loss": 6.1466,
+      "learning_rate": 0.0002163294797687861,
+      "loss": 6.8925,
       "step": 1967
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.202312138728324e-05,
-      "loss": 6.2038,
+      "learning_rate": 0.00021624277456647396,
+      "loss": 6.9777,
       "step": 1968
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.199421965317919e-05,
-      "loss": 6.3159,
+      "learning_rate": 0.00021615606936416186,
+      "loss": 7.0716,
       "step": 1969
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.196531791907515e-05,
-      "loss": 6.0993,
+      "learning_rate": 0.0002160693641618497,
+      "loss": 6.8889,
       "step": 1970
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.19364161849711e-05,
-      "loss": 6.1749,
+      "learning_rate": 0.00021598265895953755,
+      "loss": 6.9567,
       "step": 1971
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.190751445086705e-05,
-      "loss": 6.0814,
+      "learning_rate": 0.00021589595375722542,
+      "loss": 6.9446,
       "step": 1972
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.1878612716763e-05,
-      "loss": 5.9925,
+      "learning_rate": 0.00021580924855491327,
+      "loss": 6.8444,
       "step": 1973
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.184971098265896e-05,
-      "loss": 6.1261,
+      "learning_rate": 0.00021572254335260112,
+      "loss": 6.9444,
       "step": 1974
     },
     {
       "epoch": 2.21,
-      "learning_rate": 7.182080924855493e-05,
-      "loss": 6.3055,
+      "learning_rate": 0.000215635838150289,
+      "loss": 7.0171,
       "step": 1975
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.179190751445088e-05,
-      "loss": 6.2305,
+      "learning_rate": 0.00021554913294797684,
+      "loss": 6.9187,
       "step": 1976
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.176300578034682e-05,
-      "loss": 5.8749,
+      "learning_rate": 0.00021546242774566474,
+      "loss": 6.8104,
       "step": 1977
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.173410404624277e-05,
-      "loss": 6.1942,
+      "learning_rate": 0.00021537572254335259,
+      "loss": 6.9814,
       "step": 1978
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.170520231213873e-05,
-      "loss": 5.8981,
+      "learning_rate": 0.00021528901734104046,
+      "loss": 6.6869,
       "step": 1979
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.167630057803468e-05,
-      "loss": 5.7754,
+      "learning_rate": 0.0002152023121387283,
+      "loss": 6.8565,
       "step": 1980
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.164739884393063e-05,
-      "loss": 5.8689,
+      "learning_rate": 0.00021511560693641615,
+      "loss": 6.8523,
       "step": 1981
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.161849710982659e-05,
-      "loss": 6.2245,
+      "learning_rate": 0.00021502890173410403,
+      "loss": 6.8579,
       "step": 1982
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.158959537572255e-05,
-      "loss": 6.0732,
+      "learning_rate": 0.00021494219653179187,
+      "loss": 6.7726,
       "step": 1983
     },
     {
       "epoch": 2.22,
-      "learning_rate": 7.156069364161851e-05,
-      "loss": 5.228,
+      "learning_rate": 0.00021485549132947972,
+      "loss": 6.52,
       "step": 1984
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.153179190751446e-05,
-      "loss": 6.951,
+      "learning_rate": 0.00021476878612716762,
+      "loss": 7.0321,
       "step": 1985
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.150289017341041e-05,
-      "loss": 6.7966,
+      "learning_rate": 0.00021468208092485547,
+      "loss": 7.0204,
       "step": 1986
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.147398843930635e-05,
-      "loss": 6.8235,
+      "learning_rate": 0.00021459537572254334,
+      "loss": 7.0101,
       "step": 1987
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.144508670520231e-05,
-      "loss": 6.6203,
+      "learning_rate": 0.0002145086705202312,
+      "loss": 6.8474,
       "step": 1988
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.141618497109826e-05,
-      "loss": 6.7247,
+      "learning_rate": 0.00021442196531791906,
+      "loss": 6.9374,
       "step": 1989
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.138728323699421e-05,
-      "loss": 6.7491,
+      "learning_rate": 0.0002143352601156069,
+      "loss": 7.0387,
       "step": 1990
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.135838150289018e-05,
-      "loss": 6.7114,
+      "learning_rate": 0.00021424855491329476,
+      "loss": 6.989,
       "step": 1991
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.132947976878614e-05,
-      "loss": 6.7461,
+      "learning_rate": 0.00021416184971098263,
+      "loss": 7.0508,
       "step": 1992
     },
     {
       "epoch": 2.23,
-      "learning_rate": 7.130057803468209e-05,
-      "loss": 6.5973,
+      "learning_rate": 0.0002140751445086705,
+      "loss": 6.8804,
       "step": 1993
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.127167630057804e-05,
-      "loss": 6.5879,
+      "learning_rate": 0.00021398843930635838,
+      "loss": 6.9509,
       "step": 1994
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.1242774566474e-05,
-      "loss": 6.6758,
+      "learning_rate": 0.00021390173410404623,
+      "loss": 7.0645,
       "step": 1995
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.121387283236995e-05,
-      "loss": 6.4846,
+      "learning_rate": 0.0002138150289017341,
+      "loss": 6.8757,
       "step": 1996
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.118497109826589e-05,
-      "loss": 6.4285,
+      "learning_rate": 0.00021372832369942195,
+      "loss": 6.9151,
       "step": 1997
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.115606936416184e-05,
-      "loss": 6.3705,
+      "learning_rate": 0.0002136416184971098,
+      "loss": 6.868,
       "step": 1998
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.112716763005781e-05,
-      "loss": 6.5073,
+      "learning_rate": 0.00021355491329479767,
+      "loss": 7.0282,
       "step": 1999
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.109826589595376e-05,
-      "loss": 6.3103,
+      "learning_rate": 0.00021346820809248551,
+      "loss": 6.8091,
       "step": 2000
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.106936416184972e-05,
-      "loss": 6.3266,
+      "learning_rate": 0.00021338150289017341,
+      "loss": 6.8638,
       "step": 2001
     },
     {
       "epoch": 2.24,
-      "learning_rate": 7.104046242774567e-05,
-      "loss": 6.303,
+      "learning_rate": 0.00021329479768786126,
+      "loss": 6.8018,
       "step": 2002
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.101156069364162e-05,
-      "loss": 6.3856,
+      "learning_rate": 0.0002132080924855491,
+      "loss": 6.9031,
       "step": 2003
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.098265895953758e-05,
-      "loss": 6.4132,
+      "learning_rate": 0.00021312138728323698,
+      "loss": 7.0185,
       "step": 2004
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.095375722543353e-05,
-      "loss": 6.3648,
+      "learning_rate": 0.00021303468208092483,
+      "loss": 6.9997,
       "step": 2005
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.092485549132948e-05,
-      "loss": 6.4897,
+      "learning_rate": 0.0002129479768786127,
+      "loss": 7.0221,
       "step": 2006
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.089595375722544e-05,
-      "loss": 6.3063,
+      "learning_rate": 0.00021286127167630055,
+      "loss": 6.8832,
       "step": 2007
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.086705202312139e-05,
-      "loss": 6.3709,
+      "learning_rate": 0.0002127745664739884,
+      "loss": 6.9698,
       "step": 2008
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.083815028901734e-05,
-      "loss": 6.3941,
+      "learning_rate": 0.0002126878612716763,
+      "loss": 6.9599,
       "step": 2009
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.08092485549133e-05,
-      "loss": 6.3307,
+      "learning_rate": 0.00021260115606936414,
+      "loss": 6.9653,
       "step": 2010
     },
     {
       "epoch": 2.25,
-      "learning_rate": 7.078034682080925e-05,
-      "loss": 6.3309,
+      "learning_rate": 0.00021251445086705202,
+      "loss": 6.9907,
       "step": 2011
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.07514450867052e-05,
-      "loss": 6.2503,
+      "learning_rate": 0.00021242774566473987,
+      "loss": 6.9604,
       "step": 2012
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.072254335260116e-05,
-      "loss": 6.2348,
+      "learning_rate": 0.0002123410404624277,
+      "loss": 6.9587,
       "step": 2013
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.069364161849711e-05,
-      "loss": 6.3271,
+      "learning_rate": 0.00021225433526011559,
+      "loss": 7.0054,
       "step": 2014
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.066473988439308e-05,
-      "loss": 6.3151,
+      "learning_rate": 0.00021216763005780343,
+      "loss": 6.9405,
       "step": 2015
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.063583815028902e-05,
-      "loss": 6.2918,
+      "learning_rate": 0.0002120809248554913,
+      "loss": 6.9759,
       "step": 2016
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.060693641618497e-05,
-      "loss": 6.2393,
+      "learning_rate": 0.00021199421965317918,
+      "loss": 6.9942,
       "step": 2017
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.057803468208092e-05,
-      "loss": 6.1759,
+      "learning_rate": 0.00021190751445086705,
+      "loss": 6.8951,
       "step": 2018
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.054913294797688e-05,
-      "loss": 6.3239,
+      "learning_rate": 0.0002118208092485549,
+      "loss": 7.0844,
       "step": 2019
     },
     {
       "epoch": 2.26,
-      "learning_rate": 7.052023121387283e-05,
-      "loss": 6.3608,
+      "learning_rate": 0.00021173410404624275,
+      "loss": 7.0763,
       "step": 2020
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.049132947976879e-05,
-      "loss": 6.2099,
+      "learning_rate": 0.00021164739884393062,
+      "loss": 6.8841,
       "step": 2021
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.046242774566474e-05,
-      "loss": 6.2054,
+      "learning_rate": 0.00021156069364161847,
+      "loss": 7.0038,
       "step": 2022
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.04335260115607e-05,
-      "loss": 6.1373,
+      "learning_rate": 0.00021147398843930634,
+      "loss": 6.9805,
       "step": 2023
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.040462427745666e-05,
-      "loss": 6.0486,
+      "learning_rate": 0.0002113872832369942,
+      "loss": 6.9433,
       "step": 2024
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.037572254335261e-05,
-      "loss": 5.9753,
+      "learning_rate": 0.0002113005780346821,
+      "loss": 6.8886,
       "step": 2025
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.034682080924855e-05,
-      "loss": 5.9848,
+      "learning_rate": 0.00021121387283236994,
+      "loss": 6.9897,
       "step": 2026
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.03179190751445e-05,
-      "loss": 5.9763,
+      "learning_rate": 0.00021112716763005778,
+      "loss": 6.9658,
       "step": 2027
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.028901734104046e-05,
-      "loss": 5.7583,
+      "learning_rate": 0.00021104046242774566,
+      "loss": 6.8846,
       "step": 2028
     },
     {
       "epoch": 2.27,
-      "learning_rate": 7.026011560693641e-05,
-      "loss": 5.7898,
+      "learning_rate": 0.0002109537572254335,
+      "loss": 6.7809,
       "step": 2029
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.023121387283237e-05,
-      "loss": 5.8298,
+      "learning_rate": 0.00021086705202312135,
+      "loss": 7.0632,
       "step": 2030
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.020231213872833e-05,
-      "loss": 5.4671,
+      "learning_rate": 0.00021078034682080922,
+      "loss": 6.6858,
       "step": 2031
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.017341040462429e-05,
-      "loss": 5.8264,
+      "learning_rate": 0.00021069364161849707,
+      "loss": 6.914,
       "step": 2032
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.014450867052024e-05,
-      "loss": 5.0707,
+      "learning_rate": 0.00021060693641618497,
+      "loss": 6.5817,
       "step": 2033
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.01156069364162e-05,
-      "loss": 5.3941,
+      "learning_rate": 0.00021052023121387282,
+      "loss": 6.7844,
       "step": 2034
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.008670520231215e-05,
-      "loss": 6.8356,
+      "learning_rate": 0.0002104335260115607,
+      "loss": 7.0612,
       "step": 2035
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.00578034682081e-05,
-      "loss": 6.7286,
+      "learning_rate": 0.00021034682080924854,
+      "loss": 7.0231,
       "step": 2036
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7.002890173410404e-05,
-      "loss": 6.6488,
+      "learning_rate": 0.0002102601156069364,
+      "loss": 6.9822,
       "step": 2037
     },
     {
       "epoch": 2.28,
-      "learning_rate": 7e-05,
-      "loss": 6.6285,
+      "learning_rate": 0.00021017341040462426,
+      "loss": 6.9962,
       "step": 2038
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.997109826589596e-05,
-      "loss": 6.5331,
+      "learning_rate": 0.0002100867052023121,
+      "loss": 6.906,
       "step": 2039
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.994219653179191e-05,
-      "loss": 6.7313,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 7.1105,
       "step": 2040
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.991329479768787e-05,
-      "loss": 6.497,
+      "learning_rate": 0.00020991329479768786,
+      "loss": 6.9435,
       "step": 2041
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.988439306358382e-05,
-      "loss": 6.6354,
+      "learning_rate": 0.00020982658959537573,
+      "loss": 7.0628,
       "step": 2042
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.985549132947977e-05,
-      "loss": 6.4892,
+      "learning_rate": 0.00020973988439306358,
+      "loss": 7.0151,
       "step": 2043
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.982658959537573e-05,
-      "loss": 6.2512,
+      "learning_rate": 0.00020965317919075142,
+      "loss": 6.8504,
       "step": 2044
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.979768786127168e-05,
-      "loss": 6.4761,
+      "learning_rate": 0.0002095664739884393,
+      "loss": 6.9819,
       "step": 2045
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.976878612716763e-05,
-      "loss": 6.3497,
+      "learning_rate": 0.00020947976878612714,
+      "loss": 6.9592,
       "step": 2046
     },
     {
       "epoch": 2.29,
-      "learning_rate": 6.973988439306359e-05,
-      "loss": 6.4583,
+      "learning_rate": 0.000209393063583815,
+      "loss": 7.0924,
       "step": 2047
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.971098265895954e-05,
-      "loss": 6.2062,
+      "learning_rate": 0.00020930635838150286,
+      "loss": 6.848,
       "step": 2048
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.96820809248555e-05,
-      "loss": 6.568,
+      "learning_rate": 0.00020921965317919074,
+      "loss": 7.1495,
       "step": 2049
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.965317919075145e-05,
-      "loss": 6.2684,
+      "learning_rate": 0.0002091329479768786,
+      "loss": 6.9253,
       "step": 2050
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.96242774566474e-05,
-      "loss": 6.4081,
+      "learning_rate": 0.00020904624277456646,
+      "loss": 7.0552,
       "step": 2051
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.959537572254336e-05,
-      "loss": 6.2467,
+      "learning_rate": 0.00020895953757225433,
+      "loss": 6.9263,
       "step": 2052
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.956647398843931e-05,
-      "loss": 6.3423,
+      "learning_rate": 0.00020887283236994218,
+      "loss": 6.9737,
       "step": 2053
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.953757225433526e-05,
-      "loss": 6.323,
+      "learning_rate": 0.00020878612716763003,
+      "loss": 6.991,
       "step": 2054
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.950867052023122e-05,
-      "loss": 6.4066,
+      "learning_rate": 0.0002086994219653179,
+      "loss": 7.0506,
       "step": 2055
     },
     {
       "epoch": 2.3,
-      "learning_rate": 6.947976878612717e-05,
-      "loss": 6.1305,
+      "learning_rate": 0.00020861271676300575,
+      "loss": 6.8358,
       "step": 2056
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.945086705202312e-05,
-      "loss": 6.2875,
+      "learning_rate": 0.00020852601156069365,
+      "loss": 6.9742,
       "step": 2057
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.942196531791908e-05,
-      "loss": 6.1222,
+      "learning_rate": 0.0002084393063583815,
+      "loss": 6.8166,
       "step": 2058
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.939306358381503e-05,
-      "loss": 6.3224,
+      "learning_rate": 0.00020835260115606934,
+      "loss": 7.079,
       "step": 2059
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.936416184971098e-05,
-      "loss": 6.242,
+      "learning_rate": 0.00020826589595375722,
+      "loss": 6.9433,
       "step": 2060
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.933526011560694e-05,
-      "loss": 6.4748,
+      "learning_rate": 0.00020817919075144506,
+      "loss": 7.1704,
       "step": 2061
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.930635838150289e-05,
-      "loss": 6.3409,
+      "learning_rate": 0.00020809248554913294,
+      "loss": 7.0347,
       "step": 2062
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.927745664739884e-05,
-      "loss": 6.1038,
+      "learning_rate": 0.00020800578034682078,
+      "loss": 6.86,
       "step": 2063
     },
     {
       "epoch": 2.31,
-      "learning_rate": 6.924855491329481e-05,
-      "loss": 6.3558,
+      "learning_rate": 0.00020791907514450863,
+      "loss": 7.1047,
       "step": 2064
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.921965317919076e-05,
-      "loss": 6.2441,
+      "learning_rate": 0.0002078323699421965,
+      "loss": 6.9783,
       "step": 2065
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.91907514450867e-05,
-      "loss": 6.0611,
+      "learning_rate": 0.00020774566473988438,
+      "loss": 6.8379,
       "step": 2066
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.916184971098266e-05,
-      "loss": 6.1587,
+      "learning_rate": 0.00020765895953757225,
+      "loss": 6.9094,
       "step": 2067
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.913294797687861e-05,
-      "loss": 6.1319,
+      "learning_rate": 0.0002075722543352601,
+      "loss": 6.9238,
       "step": 2068
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.910404624277456e-05,
-      "loss": 6.2098,
+      "learning_rate": 0.00020748554913294797,
+      "loss": 7.0123,
       "step": 2069
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.907514450867052e-05,
-      "loss": 6.1203,
+      "learning_rate": 0.00020739884393063582,
+      "loss": 6.9896,
       "step": 2070
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.904624277456647e-05,
-      "loss": 6.0794,
+      "learning_rate": 0.00020731213872832367,
+      "loss": 6.8936,
       "step": 2071
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.901734104046244e-05,
-      "loss": 6.1498,
+      "learning_rate": 0.00020722543352601154,
+      "loss": 6.9538,
       "step": 2072
     },
     {
       "epoch": 2.32,
-      "learning_rate": 6.898843930635839e-05,
-      "loss": 5.8597,
+      "learning_rate": 0.00020713872832369939,
+      "loss": 6.7828,
       "step": 2073
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.895953757225435e-05,
-      "loss": 6.0519,
+      "learning_rate": 0.0002070520231213873,
+      "loss": 7.0055,
       "step": 2074
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.89306358381503e-05,
-      "loss": 6.0283,
+      "learning_rate": 0.00020696531791907513,
+      "loss": 6.963,
       "step": 2075
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.890173410404624e-05,
-      "loss": 5.957,
+      "learning_rate": 0.00020687861271676298,
+      "loss": 6.8708,
       "step": 2076
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.887283236994219e-05,
-      "loss": 5.9698,
+      "learning_rate": 0.00020679190751445085,
+      "loss": 6.9526,
       "step": 2077
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.884393063583815e-05,
-      "loss": 6.0001,
+      "learning_rate": 0.0002067052023121387,
+      "loss": 6.9581,
       "step": 2078
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.88150289017341e-05,
-      "loss": 5.7036,
+      "learning_rate": 0.00020661849710982658,
+      "loss": 6.7262,
       "step": 2079
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.878612716763007e-05,
-      "loss": 5.8363,
+      "learning_rate": 0.00020653179190751442,
+      "loss": 6.9372,
       "step": 2080
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.875722543352602e-05,
-      "loss": 5.6631,
+      "learning_rate": 0.00020644508670520227,
+      "loss": 6.6874,
       "step": 2081
     },
     {
       "epoch": 2.33,
-      "learning_rate": 6.872832369942197e-05,
-      "loss": 5.842,
+      "learning_rate": 0.00020635838150289017,
+      "loss": 6.6799,
       "step": 2082
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.869942196531793e-05,
-      "loss": 5.5189,
+      "learning_rate": 0.00020627167630057802,
+      "loss": 6.5473,
       "step": 2083
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.867052023121388e-05,
-      "loss": 4.9468,
+      "learning_rate": 0.0002061849710982659,
+      "loss": 6.5257,
       "step": 2084
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.864161849710983e-05,
-      "loss": 6.8171,
+      "learning_rate": 0.00020609826589595374,
+      "loss": 6.9541,
       "step": 2085
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.861271676300579e-05,
-      "loss": 6.8341,
+      "learning_rate": 0.00020601156069364158,
+      "loss": 7.0308,
       "step": 2086
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.858381502890173e-05,
-      "loss": 6.8958,
+      "learning_rate": 0.00020592485549132946,
+      "loss": 7.0212,
       "step": 2087
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.85549132947977e-05,
-      "loss": 6.7736,
+      "learning_rate": 0.0002058381502890173,
+      "loss": 6.9367,
       "step": 2088
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.852601156069365e-05,
-      "loss": 6.6023,
+      "learning_rate": 0.00020575144508670518,
+      "loss": 6.9288,
       "step": 2089
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.84971098265896e-05,
-      "loss": 6.718,
+      "learning_rate": 0.00020566473988439305,
+      "loss": 6.9238,
       "step": 2090
     },
     {
       "epoch": 2.34,
-      "learning_rate": 6.846820809248555e-05,
-      "loss": 6.6374,
+      "learning_rate": 0.00020557803468208093,
+      "loss": 6.975,
       "step": 2091
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.843930635838151e-05,
-      "loss": 6.5906,
+      "learning_rate": 0.00020549132947976877,
+      "loss": 6.9068,
       "step": 2092
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.841040462427746e-05,
-      "loss": 6.5541,
+      "learning_rate": 0.00020540462427745662,
+      "loss": 6.971,
       "step": 2093
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.838150289017341e-05,
-      "loss": 6.666,
+      "learning_rate": 0.0002053179190751445,
+      "loss": 7.0822,
       "step": 2094
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.835260115606937e-05,
-      "loss": 6.4336,
+      "learning_rate": 0.00020523121387283234,
+      "loss": 6.9237,
       "step": 2095
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.832369942196532e-05,
-      "loss": 6.618,
+      "learning_rate": 0.00020514450867052021,
+      "loss": 7.056,
       "step": 2096
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.829479768786127e-05,
-      "loss": 6.5561,
+      "learning_rate": 0.00020505780346820806,
+      "loss": 7.1426,
       "step": 2097
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.826589595375723e-05,
-      "loss": 6.6008,
+      "learning_rate": 0.00020497109826589596,
+      "loss": 7.1512,
       "step": 2098
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.823699421965318e-05,
-      "loss": 6.3025,
+      "learning_rate": 0.0002048843930635838,
+      "loss": 6.9327,
       "step": 2099
     },
     {
       "epoch": 2.35,
-      "learning_rate": 6.820809248554913e-05,
-      "loss": 6.4648,
+      "learning_rate": 0.00020479768786127166,
+      "loss": 6.9617,
       "step": 2100
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.817919075144509e-05,
-      "loss": 6.2608,
+      "learning_rate": 0.00020471098265895953,
+      "loss": 6.7966,
       "step": 2101
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.815028901734104e-05,
-      "loss": 6.4653,
+      "learning_rate": 0.00020462427745664738,
+      "loss": 7.0401,
       "step": 2102
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.8121387283237e-05,
-      "loss": 6.4377,
+      "learning_rate": 0.00020453757225433522,
+      "loss": 7.0574,
       "step": 2103
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.809248554913296e-05,
-      "loss": 6.2604,
+      "learning_rate": 0.0002044508670520231,
+      "loss": 6.9434,
       "step": 2104
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.80635838150289e-05,
-      "loss": 6.3783,
+      "learning_rate": 0.00020436416184971094,
+      "loss": 7.0533,
       "step": 2105
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.803468208092486e-05,
-      "loss": 6.3499,
+      "learning_rate": 0.00020427745664739885,
+      "loss": 7.0053,
       "step": 2106
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.800578034682081e-05,
-      "loss": 6.157,
+      "learning_rate": 0.0002041907514450867,
+      "loss": 6.8389,
       "step": 2107
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.797687861271676e-05,
-      "loss": 6.3067,
+      "learning_rate": 0.00020410404624277457,
+      "loss": 6.9768,
       "step": 2108
     },
     {
       "epoch": 2.36,
-      "learning_rate": 6.794797687861272e-05,
-      "loss": 6.2274,
+      "learning_rate": 0.0002040173410404624,
+      "loss": 6.943,
       "step": 2109
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.791907514450867e-05,
-      "loss": 6.2712,
+      "learning_rate": 0.00020393063583815026,
+      "loss": 6.9191,
       "step": 2110
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.789017341040462e-05,
-      "loss": 6.2983,
+      "learning_rate": 0.00020384393063583813,
+      "loss": 7.0167,
       "step": 2111
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.786127167630059e-05,
-      "loss": 6.1493,
+      "learning_rate": 0.00020375722543352598,
+      "loss": 6.9218,
       "step": 2112
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.783236994219654e-05,
-      "loss": 6.2077,
+      "learning_rate": 0.00020367052023121385,
+      "loss": 6.9042,
       "step": 2113
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.78034682080925e-05,
-      "loss": 6.3103,
+      "learning_rate": 0.00020358381502890173,
+      "loss": 7.017,
       "step": 2114
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.777456647398845e-05,
-      "loss": 6.1463,
+      "learning_rate": 0.00020349710982658957,
+      "loss": 6.8421,
       "step": 2115
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.774566473988439e-05,
-      "loss": 6.1529,
+      "learning_rate": 0.00020341040462427745,
+      "loss": 6.9375,
       "step": 2116
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.771676300578034e-05,
-      "loss": 6.0041,
+      "learning_rate": 0.0002033236994219653,
+      "loss": 6.8177,
       "step": 2117
     },
     {
       "epoch": 2.37,
-      "learning_rate": 6.76878612716763e-05,
-      "loss": 6.1095,
+      "learning_rate": 0.00020323699421965317,
+      "loss": 6.9083,
       "step": 2118
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.765895953757225e-05,
-      "loss": 6.2278,
+      "learning_rate": 0.00020315028901734102,
+      "loss": 6.9888,
       "step": 2119
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.763005780346822e-05,
-      "loss": 6.2832,
+      "learning_rate": 0.00020306358381502886,
+      "loss": 7.0204,
       "step": 2120
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.760115606936417e-05,
-      "loss": 6.1321,
+      "learning_rate": 0.00020297687861271674,
+      "loss": 6.8961,
       "step": 2121
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.757225433526012e-05,
-      "loss": 6.2306,
+      "learning_rate": 0.0002028901734104046,
+      "loss": 7.0429,
       "step": 2122
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.754335260115608e-05,
-      "loss": 6.0556,
+      "learning_rate": 0.00020280346820809248,
+      "loss": 6.9339,
       "step": 2123
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.751445086705203e-05,
-      "loss": 6.0592,
+      "learning_rate": 0.00020271676300578033,
+      "loss": 6.9862,
       "step": 2124
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.748554913294798e-05,
-      "loss": 6.0594,
+      "learning_rate": 0.0002026300578034682,
+      "loss": 6.9925,
       "step": 2125
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.745664739884392e-05,
-      "loss": 6.0343,
+      "learning_rate": 0.00020254335260115605,
+      "loss": 6.9628,
       "step": 2126
     },
     {
       "epoch": 2.38,
-      "learning_rate": 6.742774566473988e-05,
-      "loss": 5.9209,
+      "learning_rate": 0.0002024566473988439,
+      "loss": 6.98,
       "step": 2127
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.739884393063584e-05,
-      "loss": 5.8043,
+      "learning_rate": 0.00020236994219653177,
+      "loss": 6.8147,
       "step": 2128
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.73699421965318e-05,
-      "loss": 5.9634,
+      "learning_rate": 0.00020228323699421962,
+      "loss": 6.9135,
       "step": 2129
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.734104046242775e-05,
-      "loss": 5.6916,
+      "learning_rate": 0.00020219653179190752,
+      "loss": 6.8121,
       "step": 2130
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.73121387283237e-05,
-      "loss": 5.5224,
+      "learning_rate": 0.00020210982658959537,
+      "loss": 6.7514,
       "step": 2131
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.728323699421966e-05,
-      "loss": 5.6769,
+      "learning_rate": 0.00020202312138728321,
+      "loss": 6.8974,
       "step": 2132
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.725433526011561e-05,
-      "loss": 5.5512,
+      "learning_rate": 0.0002019364161849711,
+      "loss": 6.8414,
       "step": 2133
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.722543352601157e-05,
-      "loss": 5.0503,
+      "learning_rate": 0.00020184971098265893,
+      "loss": 6.5512,
       "step": 2134
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.719653179190752e-05,
-      "loss": 6.67,
+      "learning_rate": 0.0002017630057803468,
+      "loss": 7.005,
       "step": 2135
     },
     {
       "epoch": 2.39,
-      "learning_rate": 6.716763005780347e-05,
-      "loss": 6.7549,
+      "learning_rate": 0.00020167630057803466,
+      "loss": 7.0229,
       "step": 2136
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.713872832369943e-05,
-      "loss": 6.5025,
+      "learning_rate": 0.0002015895953757225,
+      "loss": 6.8793,
       "step": 2137
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.710982658959538e-05,
-      "loss": 6.5191,
+      "learning_rate": 0.0002015028901734104,
+      "loss": 6.9326,
       "step": 2138
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.708092485549133e-05,
-      "loss": 6.5237,
+      "learning_rate": 0.00020141618497109825,
+      "loss": 7.0075,
       "step": 2139
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.705202312138729e-05,
-      "loss": 6.512,
+      "learning_rate": 0.00020132947976878612,
+      "loss": 6.9151,
       "step": 2140
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.702312138728324e-05,
-      "loss": 6.3417,
+      "learning_rate": 0.00020124277456647397,
+      "loss": 6.8443,
       "step": 2141
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.699421965317919e-05,
-      "loss": 6.6021,
+      "learning_rate": 0.00020115606936416184,
+      "loss": 7.1443,
       "step": 2142
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.696531791907515e-05,
-      "loss": 6.4843,
+      "learning_rate": 0.0002010693641618497,
+      "loss": 7.0756,
       "step": 2143
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.693641618497111e-05,
-      "loss": 6.149,
+      "learning_rate": 0.00020098265895953754,
+      "loss": 6.7966,
       "step": 2144
     },
     {
       "epoch": 2.4,
-      "learning_rate": 6.690751445086705e-05,
-      "loss": 6.2054,
+      "learning_rate": 0.0002008959537572254,
+      "loss": 6.8244,
       "step": 2145
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.6878612716763e-05,
-      "loss": 6.3557,
+      "learning_rate": 0.00020080924855491329,
+      "loss": 6.9603,
       "step": 2146
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.684971098265896e-05,
-      "loss": 6.4951,
+      "learning_rate": 0.00020072254335260116,
+      "loss": 7.0392,
       "step": 2147
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.682080924855491e-05,
-      "loss": 6.3176,
+      "learning_rate": 0.000200635838150289,
+      "loss": 6.955,
       "step": 2148
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.679190751445087e-05,
-      "loss": 6.3189,
+      "learning_rate": 0.00020054913294797685,
+      "loss": 6.9865,
       "step": 2149
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.676300578034682e-05,
-      "loss": 6.319,
+      "learning_rate": 0.00020046242774566473,
+      "loss": 7.0218,
       "step": 2150
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.673410404624277e-05,
-      "loss": 6.3084,
+      "learning_rate": 0.00020037572254335257,
+      "loss": 7.0424,
       "step": 2151
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.670520231213874e-05,
-      "loss": 6.3691,
+      "learning_rate": 0.00020028901734104045,
+      "loss": 6.9913,
       "step": 2152
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.66763005780347e-05,
-      "loss": 6.1701,
+      "learning_rate": 0.0002002023121387283,
+      "loss": 6.9176,
       "step": 2153
     },
     {
       "epoch": 2.41,
-      "learning_rate": 6.664739884393065e-05,
-      "loss": 6.3376,
+      "learning_rate": 0.0002001156069364162,
+      "loss": 7.008,
       "step": 2154
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.661849710982659e-05,
-      "loss": 6.4101,
+      "learning_rate": 0.00020002890173410404,
+      "loss": 7.041,
       "step": 2155
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.658959537572254e-05,
-      "loss": 6.1331,
+      "learning_rate": 0.0001999421965317919,
+      "loss": 6.9456,
       "step": 2156
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.65606936416185e-05,
-      "loss": 6.1555,
+      "learning_rate": 0.00019985549132947976,
+      "loss": 6.8233,
       "step": 2157
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.653179190751445e-05,
-      "loss": 6.1745,
+      "learning_rate": 0.0001997687861271676,
+      "loss": 6.8739,
       "step": 2158
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.65028901734104e-05,
-      "loss": 6.0891,
+      "learning_rate": 0.00019968208092485546,
+      "loss": 6.9215,
       "step": 2159
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.647398843930635e-05,
-      "loss": 6.3492,
+      "learning_rate": 0.00019959537572254333,
+      "loss": 7.1107,
       "step": 2160
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.644508670520232e-05,
-      "loss": 6.2678,
+      "learning_rate": 0.00019950867052023118,
+      "loss": 6.9844,
       "step": 2161
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.641618497109828e-05,
-      "loss": 6.0272,
+      "learning_rate": 0.00019942196531791905,
+      "loss": 6.8184,
       "step": 2162
     },
     {
       "epoch": 2.42,
-      "learning_rate": 6.638728323699423e-05,
-      "loss": 6.1791,
+      "learning_rate": 0.00019933526011560693,
+      "loss": 6.9835,
       "step": 2163
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.635838150289018e-05,
-      "loss": 6.1499,
+      "learning_rate": 0.0001992485549132948,
+      "loss": 6.9782,
       "step": 2164
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.632947976878614e-05,
-      "loss": 6.1974,
+      "learning_rate": 0.00019916184971098265,
+      "loss": 7.0813,
       "step": 2165
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.630057803468208e-05,
-      "loss": 6.0845,
+      "learning_rate": 0.0001990751445086705,
+      "loss": 6.8786,
       "step": 2166
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.627167630057803e-05,
-      "loss": 6.2293,
+      "learning_rate": 0.00019898843930635837,
+      "loss": 7.0171,
       "step": 2167
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.624277456647398e-05,
-      "loss": 6.1176,
+      "learning_rate": 0.0001989017341040462,
+      "loss": 6.9252,
       "step": 2168
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.621387283236995e-05,
-      "loss": 6.0309,
+      "learning_rate": 0.0001988150289017341,
+      "loss": 6.949,
       "step": 2169
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.61849710982659e-05,
-      "loss": 6.1427,
+      "learning_rate": 0.00019872832369942193,
+      "loss": 6.9741,
       "step": 2170
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.615606936416186e-05,
-      "loss": 6.259,
+      "learning_rate": 0.00019864161849710984,
+      "loss": 7.0893,
       "step": 2171
     },
     {
       "epoch": 2.43,
-      "learning_rate": 6.612716763005781e-05,
-      "loss": 6.1359,
+      "learning_rate": 0.00019855491329479768,
+      "loss": 6.9957,
       "step": 2172
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.609826589595376e-05,
-      "loss": 5.9935,
+      "learning_rate": 0.00019846820809248553,
+      "loss": 6.9421,
       "step": 2173
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.606936416184972e-05,
-      "loss": 5.9072,
+      "learning_rate": 0.0001983815028901734,
+      "loss": 6.8792,
       "step": 2174
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.604046242774567e-05,
-      "loss": 6.1121,
+      "learning_rate": 0.00019829479768786125,
+      "loss": 6.9371,
       "step": 2175
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.601156069364161e-05,
-      "loss": 6.1095,
+      "learning_rate": 0.0001982080924855491,
+      "loss": 6.8966,
       "step": 2176
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.598265895953758e-05,
-      "loss": 6.2343,
+      "learning_rate": 0.00019812138728323697,
+      "loss": 7.1452,
       "step": 2177
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.595375722543353e-05,
-      "loss": 6.1025,
+      "learning_rate": 0.00019803468208092482,
+      "loss": 7.1342,
       "step": 2178
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.592485549132948e-05,
-      "loss": 5.8195,
+      "learning_rate": 0.00019794797687861272,
+      "loss": 6.8888,
       "step": 2179
     },
     {
       "epoch": 2.44,
-      "learning_rate": 6.589595375722544e-05,
-      "loss": 5.7557,
+      "learning_rate": 0.00019786127167630056,
+      "loss": 6.9203,
       "step": 2180
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.586705202312139e-05,
-      "loss": 5.4694,
+      "learning_rate": 0.00019777456647398844,
+      "loss": 6.7944,
       "step": 2181
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.583815028901734e-05,
-      "loss": 5.6351,
+      "learning_rate": 0.00019768786127167629,
+      "loss": 7.0215,
       "step": 2182
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.58092485549133e-05,
-      "loss": 5.1786,
+      "learning_rate": 0.00019760115606936413,
+      "loss": 6.6014,
       "step": 2183
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.578034682080925e-05,
-      "loss": 5.2347,
+      "learning_rate": 0.000197514450867052,
+      "loss": 6.7394,
       "step": 2184
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.57514450867052e-05,
-      "loss": 6.7054,
+      "learning_rate": 0.00019742774566473985,
+      "loss": 6.9852,
       "step": 2185
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.572254335260116e-05,
-      "loss": 6.6934,
+      "learning_rate": 0.0001973410404624277,
+      "loss": 6.9349,
       "step": 2186
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.569364161849711e-05,
-      "loss": 6.6614,
+      "learning_rate": 0.0001972543352601156,
+      "loss": 7.0507,
       "step": 2187
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.566473988439307e-05,
-      "loss": 6.581,
+      "learning_rate": 0.00019716763005780345,
+      "loss": 6.9362,
       "step": 2188
     },
     {
       "epoch": 2.45,
-      "learning_rate": 6.563583815028902e-05,
-      "loss": 6.4651,
+      "learning_rate": 0.00019708092485549132,
+      "loss": 6.8944,
       "step": 2189
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.560693641618497e-05,
-      "loss": 6.3655,
+      "learning_rate": 0.00019699421965317917,
+      "loss": 6.8528,
       "step": 2190
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.557803468208093e-05,
-      "loss": 6.5188,
+      "learning_rate": 0.00019690751445086704,
+      "loss": 6.9726,
       "step": 2191
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.554913294797688e-05,
-      "loss": 6.466,
+      "learning_rate": 0.0001968208092485549,
+      "loss": 7.0471,
       "step": 2192
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.552023121387285e-05,
-      "loss": 6.3972,
+      "learning_rate": 0.00019673410404624274,
+      "loss": 6.9905,
       "step": 2193
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.54913294797688e-05,
-      "loss": 6.3644,
+      "learning_rate": 0.0001966473988439306,
+      "loss": 6.9873,
       "step": 2194
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.546242774566474e-05,
-      "loss": 6.2896,
+      "learning_rate": 0.00019656069364161848,
+      "loss": 7.0071,
       "step": 2195
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.543352601156069e-05,
-      "loss": 6.18,
+      "learning_rate": 0.00019647398843930636,
+      "loss": 6.9064,
       "step": 2196
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.540462427745665e-05,
-      "loss": 6.2135,
+      "learning_rate": 0.0001963872832369942,
+      "loss": 6.8852,
       "step": 2197
     },
     {
       "epoch": 2.46,
-      "learning_rate": 6.53757225433526e-05,
-      "loss": 6.1674,
+      "learning_rate": 0.00019630057803468208,
+      "loss": 6.8639,
       "step": 2198
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.534682080924855e-05,
-      "loss": 6.2877,
+      "learning_rate": 0.00019621387283236992,
+      "loss": 7.0239,
       "step": 2199
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.53179190751445e-05,
-      "loss": 6.2896,
+      "learning_rate": 0.00019612716763005777,
+      "loss": 6.9944,
       "step": 2200
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.528901734104047e-05,
-      "loss": 6.2601,
+      "learning_rate": 0.00019604046242774565,
+      "loss": 6.9736,
       "step": 2201
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.526011560693643e-05,
-      "loss": 6.2079,
+      "learning_rate": 0.0001959537572254335,
+      "loss": 6.932,
       "step": 2202
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.523121387283238e-05,
-      "loss": 6.1356,
+      "learning_rate": 0.0001958670520231214,
+      "loss": 6.822,
       "step": 2203
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.520231213872833e-05,
-      "loss": 6.3199,
+      "learning_rate": 0.00019578034682080924,
+      "loss": 7.0522,
       "step": 2204
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.517341040462427e-05,
-      "loss": 6.2054,
+      "learning_rate": 0.0001956936416184971,
+      "loss": 6.9664,
       "step": 2205
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.514450867052023e-05,
-      "loss": 6.1832,
+      "learning_rate": 0.00019560693641618496,
+      "loss": 6.9162,
       "step": 2206
     },
     {
       "epoch": 2.47,
-      "learning_rate": 6.511560693641618e-05,
-      "loss": 6.1811,
+      "learning_rate": 0.0001955202312138728,
+      "loss": 6.8922,
       "step": 2207
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.508670520231213e-05,
-      "loss": 6.2345,
+      "learning_rate": 0.00019543352601156068,
+      "loss": 7.0159,
       "step": 2208
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.50578034682081e-05,
-      "loss": 6.2337,
+      "learning_rate": 0.00019534682080924853,
+      "loss": 6.9669,
       "step": 2209
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.502890173410405e-05,
-      "loss": 6.0245,
+      "learning_rate": 0.00019526011560693637,
+      "loss": 6.8704,
       "step": 2210
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.500000000000001e-05,
-      "loss": 6.3077,
+      "learning_rate": 0.00019517341040462428,
+      "loss": 7.0689,
       "step": 2211
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.497109826589596e-05,
-      "loss": 6.1958,
+      "learning_rate": 0.00019508670520231212,
+      "loss": 7.0136,
       "step": 2212
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.494219653179191e-05,
-      "loss": 6.2759,
+      "learning_rate": 0.000195,
+      "loss": 7.0995,
       "step": 2213
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.491329479768787e-05,
-      "loss": 6.2227,
+      "learning_rate": 0.00019491329479768784,
+      "loss": 7.0854,
       "step": 2214
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.488439306358381e-05,
-      "loss": 6.3899,
+      "learning_rate": 0.00019482658959537572,
+      "loss": 7.1113,
       "step": 2215
     },
     {
       "epoch": 2.48,
-      "learning_rate": 6.485549132947976e-05,
-      "loss": 6.0662,
+      "learning_rate": 0.00019473988439306356,
+      "loss": 6.9315,
       "step": 2216
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.482658959537573e-05,
-      "loss": 6.0687,
+      "learning_rate": 0.0001946531791907514,
+      "loss": 6.919,
       "step": 2217
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.479768786127168e-05,
-      "loss": 5.9777,
+      "learning_rate": 0.00019456647398843928,
+      "loss": 6.9292,
       "step": 2218
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.476878612716764e-05,
-      "loss": 5.9636,
+      "learning_rate": 0.00019447976878612716,
+      "loss": 6.8709,
       "step": 2219
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.473988439306359e-05,
-      "loss": 6.1898,
+      "learning_rate": 0.00019439306358381503,
+      "loss": 7.041,
       "step": 2220
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.471098265895954e-05,
-      "loss": 6.1315,
+      "learning_rate": 0.00019430635838150288,
+      "loss": 6.9684,
       "step": 2221
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.46820809248555e-05,
-      "loss": 6.028,
+      "learning_rate": 0.00019421965317919073,
+      "loss": 6.9574,
       "step": 2222
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.465317919075145e-05,
-      "loss": 6.1695,
+      "learning_rate": 0.0001941329479768786,
+      "loss": 7.1876,
       "step": 2223
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.46242774566474e-05,
-      "loss": 6.0204,
+      "learning_rate": 0.00019404624277456645,
+      "loss": 6.9982,
       "step": 2224
     },
     {
       "epoch": 2.49,
-      "learning_rate": 6.459537572254336e-05,
-      "loss": 5.8724,
+      "learning_rate": 0.00019395953757225432,
+      "loss": 6.9613,
       "step": 2225
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.456647398843931e-05,
-      "loss": 5.8861,
+      "learning_rate": 0.00019387283236994217,
+      "loss": 6.9482,
       "step": 2226
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.453757225433526e-05,
-      "loss": 5.8431,
+      "learning_rate": 0.00019378612716763007,
+      "loss": 6.9348,
       "step": 2227
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.450867052023122e-05,
-      "loss": 5.4304,
+      "learning_rate": 0.00019369942196531792,
+      "loss": 6.6954,
       "step": 2228
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.447976878612717e-05,
-      "loss": 5.6235,
+      "learning_rate": 0.00019361271676300576,
+      "loss": 6.8187,
       "step": 2229
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.445086705202312e-05,
-      "loss": 5.4772,
+      "learning_rate": 0.00019352601156069364,
+      "loss": 6.7839,
       "step": 2230
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.442196531791908e-05,
-      "loss": 5.392,
+      "learning_rate": 0.00019343930635838148,
+      "loss": 6.7878,
       "step": 2231
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.439306358381503e-05,
-      "loss": 5.4413,
+      "learning_rate": 0.00019335260115606933,
+      "loss": 6.8933,
       "step": 2232
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.4364161849711e-05,
-      "loss": 5.6125,
+      "learning_rate": 0.0001932658959537572,
+      "loss": 6.9667,
       "step": 2233
     },
     {
       "epoch": 2.5,
-      "learning_rate": 6.433526011560694e-05,
-      "loss": 5.0809,
+      "learning_rate": 0.00019317919075144505,
+      "loss": 6.7212,
       "step": 2234
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.430635838150289e-05,
-      "loss": 6.7567,
+      "learning_rate": 0.00019309248554913295,
+      "loss": 7.0233,
       "step": 2235
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.427745664739884e-05,
-      "loss": 6.6717,
+      "learning_rate": 0.0001930057803468208,
+      "loss": 6.985,
       "step": 2236
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.42485549132948e-05,
-      "loss": 6.7137,
+      "learning_rate": 0.00019291907514450867,
+      "loss": 7.0858,
       "step": 2237
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.421965317919075e-05,
-      "loss": 6.6518,
+      "learning_rate": 0.00019283236994219652,
+      "loss": 7.1361,
       "step": 2238
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.41907514450867e-05,
-      "loss": 6.4456,
+      "learning_rate": 0.00019274566473988437,
+      "loss": 6.9472,
       "step": 2239
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.416184971098266e-05,
-      "loss": 6.4023,
+      "learning_rate": 0.00019265895953757224,
+      "loss": 6.9308,
       "step": 2240
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.413294797687862e-05,
-      "loss": 6.3774,
+      "learning_rate": 0.00019257225433526009,
+      "loss": 6.988,
       "step": 2241
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.410404624277458e-05,
-      "loss": 6.4152,
+      "learning_rate": 0.00019248554913294796,
+      "loss": 7.0087,
       "step": 2242
     },
     {
       "epoch": 2.51,
-      "learning_rate": 6.407514450867053e-05,
-      "loss": 6.4344,
+      "learning_rate": 0.00019239884393063583,
+      "loss": 7.012,
       "step": 2243
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.404624277456649e-05,
-      "loss": 6.2455,
+      "learning_rate": 0.0001923121387283237,
+      "loss": 6.9964,
       "step": 2244
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.401734104046243e-05,
-      "loss": 6.0957,
+      "learning_rate": 0.00019222543352601155,
+      "loss": 6.8609,
       "step": 2245
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.398843930635838e-05,
-      "loss": 6.3064,
+      "learning_rate": 0.0001921387283236994,
+      "loss": 6.9199,
       "step": 2246
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.395953757225433e-05,
-      "loss": 6.2631,
+      "learning_rate": 0.00019205202312138728,
+      "loss": 6.9128,
       "step": 2247
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.393063583815029e-05,
-      "loss": 6.3901,
+      "learning_rate": 0.00019196531791907512,
+      "loss": 7.1092,
       "step": 2248
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.390173410404625e-05,
-      "loss": 6.1374,
+      "learning_rate": 0.00019187861271676297,
+      "loss": 6.9251,
       "step": 2249
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.38728323699422e-05,
-      "loss": 6.0308,
+      "learning_rate": 0.00019179190751445084,
+      "loss": 6.8644,
       "step": 2250
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.384393063583816e-05,
-      "loss": 5.9995,
+      "learning_rate": 0.00019170520231213872,
+      "loss": 6.8241,
       "step": 2251
     },
     {
       "epoch": 2.52,
-      "learning_rate": 6.381502890173411e-05,
-      "loss": 6.2649,
+      "learning_rate": 0.0001916184971098266,
+      "loss": 7.0763,
       "step": 2252
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.378612716763007e-05,
-      "loss": 6.0801,
+      "learning_rate": 0.00019153179190751444,
+      "loss": 6.926,
       "step": 2253
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.375722543352602e-05,
-      "loss": 6.2368,
+      "learning_rate": 0.0001914450867052023,
+      "loss": 7.0258,
       "step": 2254
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.372832369942196e-05,
-      "loss": 6.2122,
+      "learning_rate": 0.00019135838150289016,
+      "loss": 6.9978,
       "step": 2255
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.369942196531791e-05,
-      "loss": 6.0118,
+      "learning_rate": 0.000191271676300578,
+      "loss": 6.9176,
       "step": 2256
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.367052023121387e-05,
-      "loss": 6.0468,
+      "learning_rate": 0.00019118497109826588,
+      "loss": 6.8828,
       "step": 2257
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.364161849710983e-05,
-      "loss": 6.049,
+      "learning_rate": 0.00019109826589595373,
+      "loss": 6.9542,
       "step": 2258
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.361271676300579e-05,
-      "loss": 6.1772,
+      "learning_rate": 0.00019101156069364157,
+      "loss": 6.9879,
       "step": 2259
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.358381502890174e-05,
-      "loss": 6.1366,
+      "learning_rate": 0.00019092485549132947,
+      "loss": 7.0014,
       "step": 2260
     },
     {
       "epoch": 2.53,
-      "learning_rate": 6.35549132947977e-05,
-      "loss": 5.9881,
+      "learning_rate": 0.00019083815028901732,
+      "loss": 6.887,
       "step": 2261
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.352601156069365e-05,
-      "loss": 6.0627,
+      "learning_rate": 0.0001907514450867052,
+      "loss": 6.9745,
       "step": 2262
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.34971098265896e-05,
-      "loss": 6.0591,
+      "learning_rate": 0.00019066473988439304,
+      "loss": 6.9631,
       "step": 2263
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.346820809248555e-05,
-      "loss": 6.1337,
+      "learning_rate": 0.00019057803468208091,
+      "loss": 7.0327,
       "step": 2264
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.34393063583815e-05,
-      "loss": 6.0749,
+      "learning_rate": 0.00019049132947976876,
+      "loss": 6.9537,
       "step": 2265
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.341040462427746e-05,
-      "loss": 5.9845,
+      "learning_rate": 0.0001904046242774566,
+      "loss": 6.9047,
       "step": 2266
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.338150289017341e-05,
-      "loss": 6.077,
+      "learning_rate": 0.00019031791907514448,
+      "loss": 7.0491,
       "step": 2267
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.335260115606937e-05,
-      "loss": 6.1031,
+      "learning_rate": 0.00019023121387283236,
+      "loss": 7.0433,
       "step": 2268
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.332369942196532e-05,
-      "loss": 6.0507,
+      "learning_rate": 0.00019014450867052023,
+      "loss": 7.0794,
       "step": 2269
     },
     {
       "epoch": 2.54,
-      "learning_rate": 6.329479768786127e-05,
-      "loss": 6.0939,
+      "learning_rate": 0.00019005780346820808,
+      "loss": 7.0486,
       "step": 2270
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.326589595375723e-05,
-      "loss": 5.9524,
+      "learning_rate": 0.00018997109826589595,
+      "loss": 6.9532,
       "step": 2271
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.323699421965318e-05,
-      "loss": 5.8621,
+      "learning_rate": 0.0001898843930635838,
+      "loss": 6.8391,
       "step": 2272
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.320809248554914e-05,
-      "loss": 5.9774,
+      "learning_rate": 0.00018979768786127164,
+      "loss": 7.0214,
       "step": 2273
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.317919075144509e-05,
-      "loss": 5.9155,
+      "learning_rate": 0.00018971098265895952,
+      "loss": 6.9798,
       "step": 2274
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.315028901734104e-05,
-      "loss": 5.908,
+      "learning_rate": 0.00018962427745664736,
+      "loss": 6.979,
       "step": 2275
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.3121387283237e-05,
-      "loss": 5.8693,
+      "learning_rate": 0.00018953757225433527,
+      "loss": 6.9793,
       "step": 2276
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.309248554913295e-05,
-      "loss": 5.7897,
+      "learning_rate": 0.0001894508670520231,
+      "loss": 6.9258,
       "step": 2277
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.30635838150289e-05,
-      "loss": 5.7296,
+      "learning_rate": 0.00018936416184971096,
+      "loss": 6.9259,
       "step": 2278
     },
     {
       "epoch": 2.55,
-      "learning_rate": 6.303468208092486e-05,
-      "loss": 5.9523,
+      "learning_rate": 0.00018927745664739883,
+      "loss": 7.1213,
       "step": 2279
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.300578034682081e-05,
-      "loss": 5.5181,
+      "learning_rate": 0.00018919075144508668,
+      "loss": 6.9068,
       "step": 2280
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.297687861271676e-05,
-      "loss": 5.501,
+      "learning_rate": 0.00018910404624277455,
+      "loss": 6.8597,
       "step": 2281
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.294797687861273e-05,
-      "loss": 5.5238,
+      "learning_rate": 0.0001890173410404624,
+      "loss": 6.9347,
       "step": 2282
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.291907514450868e-05,
-      "loss": 4.9858,
+      "learning_rate": 0.00018893063583815025,
+      "loss": 6.7116,
       "step": 2283
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.289017341040462e-05,
-      "loss": 5.088,
+      "learning_rate": 0.00018884393063583815,
+      "loss": 6.429,
       "step": 2284
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.286127167630058e-05,
-      "loss": 6.9055,
+      "learning_rate": 0.000188757225433526,
+      "loss": 7.072,
       "step": 2285
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.283236994219653e-05,
-      "loss": 6.6274,
+      "learning_rate": 0.00018867052023121387,
+      "loss": 6.8576,
       "step": 2286
     },
     {
       "epoch": 2.56,
-      "learning_rate": 6.280346820809248e-05,
-      "loss": 6.5455,
+      "learning_rate": 0.00018858381502890172,
+      "loss": 6.9037,
       "step": 2287
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.277456647398844e-05,
-      "loss": 6.6525,
+      "learning_rate": 0.0001884971098265896,
+      "loss": 6.981,
       "step": 2288
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.274566473988439e-05,
-      "loss": 6.5386,
+      "learning_rate": 0.00018841040462427744,
+      "loss": 6.9589,
       "step": 2289
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.271676300578036e-05,
-      "loss": 6.5105,
+      "learning_rate": 0.00018832369942196528,
+      "loss": 6.966,
       "step": 2290
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.268786127167631e-05,
-      "loss": 6.4635,
+      "learning_rate": 0.00018823699421965316,
+      "loss": 7.0038,
       "step": 2291
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.265895953757226e-05,
-      "loss": 6.4163,
+      "learning_rate": 0.00018815028901734103,
+      "loss": 6.9298,
       "step": 2292
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.263005780346822e-05,
-      "loss": 6.4379,
+      "learning_rate": 0.0001880635838150289,
+      "loss": 6.9665,
       "step": 2293
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.260115606936416e-05,
-      "loss": 6.3781,
+      "learning_rate": 0.00018797687861271675,
+      "loss": 6.9886,
       "step": 2294
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.257225433526011e-05,
-      "loss": 6.3129,
+      "learning_rate": 0.0001878901734104046,
+      "loss": 6.9647,
       "step": 2295
     },
     {
       "epoch": 2.57,
-      "learning_rate": 6.254335260115606e-05,
-      "loss": 6.3981,
+      "learning_rate": 0.00018780346820809247,
+      "loss": 7.0777,
       "step": 2296
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.251445086705202e-05,
-      "loss": 6.3841,
+      "learning_rate": 0.00018771676300578032,
+      "loss": 7.0866,
       "step": 2297
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.248554913294798e-05,
-      "loss": 6.2883,
+      "learning_rate": 0.0001876300578034682,
+      "loss": 6.9904,
       "step": 2298
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.245664739884394e-05,
-      "loss": 6.3198,
+      "learning_rate": 0.00018754335260115604,
+      "loss": 7.0259,
       "step": 2299
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.242774566473989e-05,
-      "loss": 6.2368,
+      "learning_rate": 0.00018745664739884394,
+      "loss": 6.8667,
       "step": 2300
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.239884393063585e-05,
-      "loss": 6.2307,
+      "learning_rate": 0.0001873699421965318,
+      "loss": 6.9883,
       "step": 2301
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.23699421965318e-05,
-      "loss": 6.392,
+      "learning_rate": 0.00018728323699421963,
+      "loss": 6.9481,
       "step": 2302
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.234104046242775e-05,
-      "loss": 6.1472,
+      "learning_rate": 0.0001871965317919075,
+      "loss": 6.9905,
       "step": 2303
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.23121387283237e-05,
-      "loss": 6.4138,
+      "learning_rate": 0.00018710982658959536,
+      "loss": 7.0005,
       "step": 2304
     },
     {
       "epoch": 2.58,
-      "learning_rate": 6.228323699421965e-05,
-      "loss": 6.1814,
+      "learning_rate": 0.0001870231213872832,
+      "loss": 6.7942,
       "step": 2305
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.225433526011561e-05,
-      "loss": 6.1574,
+      "learning_rate": 0.00018693641618497108,
+      "loss": 6.9081,
       "step": 2306
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.222543352601157e-05,
-      "loss": 6.2799,
+      "learning_rate": 0.00018684971098265892,
+      "loss": 6.9546,
       "step": 2307
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.219653179190752e-05,
-      "loss": 6.3701,
+      "learning_rate": 0.00018676300578034682,
+      "loss": 7.0335,
       "step": 2308
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.216763005780347e-05,
-      "loss": 6.3817,
+      "learning_rate": 0.00018667630057803467,
+      "loss": 7.0768,
       "step": 2309
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.213872832369943e-05,
-      "loss": 6.1994,
+      "learning_rate": 0.00018658959537572254,
+      "loss": 7.0509,
       "step": 2310
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.210982658959538e-05,
-      "loss": 6.1122,
+      "learning_rate": 0.0001865028901734104,
+      "loss": 6.8954,
       "step": 2311
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.208092485549133e-05,
-      "loss": 6.1179,
+      "learning_rate": 0.00018641618497109824,
+      "loss": 6.9434,
       "step": 2312
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.205202312138729e-05,
-      "loss": 6.1906,
+      "learning_rate": 0.0001863294797687861,
+      "loss": 6.9879,
       "step": 2313
     },
     {
       "epoch": 2.59,
-      "learning_rate": 6.202312138728324e-05,
-      "loss": 6.2915,
+      "learning_rate": 0.00018624277456647396,
+      "loss": 7.0399,
       "step": 2314
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.19942196531792e-05,
-      "loss": 6.179,
+      "learning_rate": 0.00018615606936416183,
+      "loss": 6.8442,
       "step": 2315
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.196531791907515e-05,
-      "loss": 6.1488,
+      "learning_rate": 0.0001860693641618497,
+      "loss": 6.8866,
       "step": 2316
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.19364161849711e-05,
-      "loss": 6.0831,
+      "learning_rate": 0.00018598265895953758,
+      "loss": 6.894,
       "step": 2317
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.190751445086705e-05,
-      "loss": 6.2188,
+      "learning_rate": 0.00018589595375722543,
+      "loss": 6.9655,
       "step": 2318
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.187861271676301e-05,
-      "loss": 6.2583,
+      "learning_rate": 0.00018580924855491327,
+      "loss": 6.9558,
       "step": 2319
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.184971098265896e-05,
-      "loss": 6.3879,
+      "learning_rate": 0.00018572254335260115,
+      "loss": 7.0881,
       "step": 2320
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.182080924855491e-05,
-      "loss": 6.1374,
+      "learning_rate": 0.000185635838150289,
+      "loss": 7.0132,
       "step": 2321
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.179190751445088e-05,
-      "loss": 6.0829,
+      "learning_rate": 0.00018554913294797684,
+      "loss": 7.0161,
       "step": 2322
     },
     {
       "epoch": 2.6,
-      "learning_rate": 6.176300578034682e-05,
-      "loss": 6.0485,
+      "learning_rate": 0.00018546242774566472,
+      "loss": 6.9863,
       "step": 2323
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.173410404624277e-05,
-      "loss": 5.9228,
+      "learning_rate": 0.0001853757225433526,
+      "loss": 6.8654,
       "step": 2324
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.170520231213873e-05,
-      "loss": 6.0523,
+      "learning_rate": 0.00018528901734104046,
+      "loss": 7.0654,
       "step": 2325
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.167630057803468e-05,
-      "loss": 5.9059,
+      "learning_rate": 0.0001852023121387283,
+      "loss": 6.9601,
       "step": 2326
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.164739884393063e-05,
-      "loss": 5.8903,
+      "learning_rate": 0.00018511560693641618,
+      "loss": 7.0114,
       "step": 2327
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.161849710982659e-05,
-      "loss": 5.6098,
+      "learning_rate": 0.00018502890173410403,
+      "loss": 6.7972,
       "step": 2328
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.158959537572254e-05,
-      "loss": 5.7852,
+      "learning_rate": 0.00018494219653179188,
+      "loss": 6.914,
       "step": 2329
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.156069364161851e-05,
-      "loss": 5.5043,
+      "learning_rate": 0.00018485549132947975,
+      "loss": 6.7499,
       "step": 2330
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.153179190751446e-05,
-      "loss": 5.4523,
+      "learning_rate": 0.0001847687861271676,
+      "loss": 6.64,
       "step": 2331
     },
     {
       "epoch": 2.61,
-      "learning_rate": 6.150289017341042e-05,
-      "loss": 5.1479,
+      "learning_rate": 0.0001846820809248555,
+      "loss": 6.6831,
       "step": 2332
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.147398843930637e-05,
-      "loss": 5.3916,
+      "learning_rate": 0.00018459537572254335,
+      "loss": 6.9852,
       "step": 2333
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.144508670520231e-05,
-      "loss": 4.6376,
+      "learning_rate": 0.0001845086705202312,
+      "loss": 6.5128,
       "step": 2334
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.141618497109826e-05,
-      "loss": 6.7546,
+      "learning_rate": 0.00018442196531791907,
+      "loss": 7.0646,
       "step": 2335
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.138728323699422e-05,
-      "loss": 6.6541,
+      "learning_rate": 0.0001843352601156069,
+      "loss": 6.9523,
       "step": 2336
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.135838150289017e-05,
-      "loss": 6.5374,
+      "learning_rate": 0.0001842485549132948,
+      "loss": 6.945,
       "step": 2337
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.132947976878614e-05,
-      "loss": 6.4809,
+      "learning_rate": 0.00018416184971098263,
+      "loss": 6.8861,
       "step": 2338
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.130057803468209e-05,
-      "loss": 6.4437,
+      "learning_rate": 0.00018407514450867048,
+      "loss": 6.9097,
       "step": 2339
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.127167630057804e-05,
-      "loss": 6.5729,
+      "learning_rate": 0.00018398843930635838,
+      "loss": 7.0461,
       "step": 2340
     },
     {
       "epoch": 2.62,
-      "learning_rate": 6.1242774566474e-05,
-      "loss": 6.4862,
+      "learning_rate": 0.00018390173410404623,
+      "loss": 7.0079,
       "step": 2341
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.121387283236995e-05,
-      "loss": 6.4213,
+      "learning_rate": 0.0001838150289017341,
+      "loss": 6.9814,
       "step": 2342
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.11849710982659e-05,
-      "loss": 6.3845,
+      "learning_rate": 0.00018372832369942195,
+      "loss": 6.9795,
       "step": 2343
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.115606936416184e-05,
-      "loss": 6.2977,
+      "learning_rate": 0.00018364161849710982,
+      "loss": 6.9272,
       "step": 2344
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.11271676300578e-05,
-      "loss": 6.2359,
+      "learning_rate": 0.00018355491329479767,
+      "loss": 6.9004,
       "step": 2345
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.109826589595375e-05,
-      "loss": 6.2633,
+      "learning_rate": 0.00018346820809248552,
+      "loss": 6.9035,
       "step": 2346
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.106936416184972e-05,
-      "loss": 6.3008,
+      "learning_rate": 0.0001833815028901734,
+      "loss": 6.9743,
       "step": 2347
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.104046242774567e-05,
-      "loss": 6.3586,
+      "learning_rate": 0.00018329479768786124,
+      "loss": 7.0236,
       "step": 2348
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.1011560693641624e-05,
-      "loss": 6.0994,
+      "learning_rate": 0.00018320809248554914,
+      "loss": 6.8364,
       "step": 2349
     },
     {
       "epoch": 2.63,
-      "learning_rate": 6.098265895953758e-05,
-      "loss": 6.3414,
+      "learning_rate": 0.00018312138728323698,
+      "loss": 7.0918,
       "step": 2350
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.0953757225433524e-05,
-      "loss": 6.1794,
+      "learning_rate": 0.00018303468208092483,
+      "loss": 6.888,
       "step": 2351
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.092485549132948e-05,
-      "loss": 6.1551,
+      "learning_rate": 0.0001829479768786127,
+      "loss": 6.8959,
       "step": 2352
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.089595375722543e-05,
-      "loss": 6.1745,
+      "learning_rate": 0.00018286127167630055,
+      "loss": 6.9124,
       "step": 2353
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.0867052023121385e-05,
-      "loss": 6.0899,
+      "learning_rate": 0.00018277456647398843,
+      "loss": 6.9408,
       "step": 2354
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.083815028901735e-05,
-      "loss": 6.1592,
+      "learning_rate": 0.00018268786127167627,
+      "loss": 6.946,
       "step": 2355
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.08092485549133e-05,
-      "loss": 6.1599,
+      "learning_rate": 0.00018260115606936412,
+      "loss": 7.0443,
       "step": 2356
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.078034682080925e-05,
-      "loss": 6.0662,
+      "learning_rate": 0.00018251445086705202,
+      "loss": 6.883,
       "step": 2357
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.0751445086705205e-05,
-      "loss": 6.1209,
+      "learning_rate": 0.00018242774566473987,
+      "loss": 6.9202,
       "step": 2358
     },
     {
       "epoch": 2.64,
-      "learning_rate": 6.072254335260116e-05,
-      "loss": 6.3439,
+      "learning_rate": 0.00018234104046242774,
+      "loss": 7.0363,
       "step": 2359
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.069364161849711e-05,
-      "loss": 6.1354,
+      "learning_rate": 0.0001822543352601156,
+      "loss": 6.8729,
       "step": 2360
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.0664739884393065e-05,
-      "loss": 6.2353,
+      "learning_rate": 0.00018216763005780346,
+      "loss": 7.0392,
       "step": 2361
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.063583815028901e-05,
-      "loss": 6.1437,
+      "learning_rate": 0.0001820809248554913,
+      "loss": 7.0296,
       "step": 2362
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.060693641618498e-05,
-      "loss": 6.1512,
+      "learning_rate": 0.00018199421965317916,
+      "loss": 7.0782,
       "step": 2363
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.057803468208093e-05,
-      "loss": 6.194,
+      "learning_rate": 0.00018190751445086703,
+      "loss": 7.0179,
       "step": 2364
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.0549132947976886e-05,
-      "loss": 6.0366,
+      "learning_rate": 0.0001818208092485549,
+      "loss": 6.9276,
       "step": 2365
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.052023121387283e-05,
-      "loss": 6.1093,
+      "learning_rate": 0.00018173410404624278,
+      "loss": 7.0438,
       "step": 2366
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.0491329479768786e-05,
-      "loss": 6.1993,
+      "learning_rate": 0.00018164739884393062,
+      "loss": 7.0175,
       "step": 2367
     },
     {
       "epoch": 2.65,
-      "learning_rate": 6.046242774566474e-05,
-      "loss": 5.8911,
+      "learning_rate": 0.00018156069364161847,
+      "loss": 6.8788,
       "step": 2368
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.043352601156069e-05,
-      "loss": 6.0275,
+      "learning_rate": 0.00018147398843930634,
+      "loss": 7.0079,
       "step": 2369
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.0404624277456646e-05,
-      "loss": 5.9444,
+      "learning_rate": 0.0001813872832369942,
+      "loss": 6.9039,
       "step": 2370
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.037572254335261e-05,
-      "loss": 5.9611,
+      "learning_rate": 0.00018130057803468207,
+      "loss": 7.0085,
       "step": 2371
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.034682080924856e-05,
-      "loss": 6.1062,
+      "learning_rate": 0.0001812138728323699,
+      "loss": 7.1188,
       "step": 2372
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.0317919075144514e-05,
-      "loss": 5.8355,
+      "learning_rate": 0.0001811271676300578,
+      "loss": 6.8821,
       "step": 2373
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.028901734104047e-05,
-      "loss": 5.8489,
+      "learning_rate": 0.00018104046242774566,
+      "loss": 6.8823,
       "step": 2374
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.026011560693642e-05,
-      "loss": 6.0035,
+      "learning_rate": 0.0001809537572254335,
+      "loss": 7.0898,
       "step": 2375
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.023121387283237e-05,
-      "loss": 5.8405,
+      "learning_rate": 0.00018086705202312138,
+      "loss": 6.9548,
       "step": 2376
     },
     {
       "epoch": 2.66,
-      "learning_rate": 6.020231213872832e-05,
-      "loss": 5.7128,
+      "learning_rate": 0.00018078034682080923,
+      "loss": 6.974,
       "step": 2377
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.0173410404624274e-05,
-      "loss": 5.7835,
+      "learning_rate": 0.00018069364161849707,
+      "loss": 6.8937,
       "step": 2378
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.014450867052024e-05,
-      "loss": 5.5911,
+      "learning_rate": 0.00018060693641618495,
+      "loss": 6.848,
       "step": 2379
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.0115606936416195e-05,
-      "loss": 5.4949,
+      "learning_rate": 0.0001805202312138728,
+      "loss": 6.9255,
       "step": 2380
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.008670520231214e-05,
-      "loss": 5.5674,
+      "learning_rate": 0.0001804335260115607,
+      "loss": 6.9246,
       "step": 2381
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.0057803468208095e-05,
-      "loss": 5.1638,
+      "learning_rate": 0.00018034682080924854,
+      "loss": 6.7807,
       "step": 2382
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6.002890173410405e-05,
-      "loss": 5.3396,
+      "learning_rate": 0.00018026011560693642,
+      "loss": 6.9638,
       "step": 2383
     },
     {
       "epoch": 2.67,
-      "learning_rate": 6e-05,
-      "loss": 4.657,
+      "learning_rate": 0.00018017341040462426,
+      "loss": 6.6612,
       "step": 2384
     },
     {
       "epoch": 2.67,
-      "learning_rate": 5.9971098265895955e-05,
-      "loss": 6.9542,
+      "learning_rate": 0.0001800867052023121,
+      "loss": 7.0488,
       "step": 2385
     },
     {
       "epoch": 2.67,
-      "learning_rate": 5.994219653179191e-05,
-      "loss": 6.9672,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 7.0164,
       "step": 2386
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.991329479768787e-05,
-      "loss": 6.7439,
+      "learning_rate": 0.00017991329479768783,
+      "loss": 6.9272,
       "step": 2387
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.988439306358382e-05,
-      "loss": 6.7137,
+      "learning_rate": 0.0001798265895953757,
+      "loss": 6.947,
       "step": 2388
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.9855491329479776e-05,
-      "loss": 6.6445,
+      "learning_rate": 0.00017973988439306358,
+      "loss": 6.9072,
       "step": 2389
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.982658959537573e-05,
-      "loss": 6.636,
+      "learning_rate": 0.00017965317919075145,
+      "loss": 6.9694,
       "step": 2390
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.9797687861271676e-05,
-      "loss": 6.5534,
+      "learning_rate": 0.0001795664739884393,
+      "loss": 6.9192,
       "step": 2391
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.976878612716763e-05,
-      "loss": 6.6281,
+      "learning_rate": 0.00017947976878612715,
+      "loss": 6.9926,
       "step": 2392
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.973988439306358e-05,
-      "loss": 6.5368,
+      "learning_rate": 0.00017939306358381502,
+      "loss": 6.9783,
       "step": 2393
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.9710982658959536e-05,
-      "loss": 6.2929,
+      "learning_rate": 0.00017930635838150287,
+      "loss": 6.7875,
       "step": 2394
     },
     {
       "epoch": 2.68,
-      "learning_rate": 5.9682080924855496e-05,
-      "loss": 6.3187,
+      "learning_rate": 0.00017921965317919071,
+      "loss": 6.8181,
       "step": 2395
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.965317919075145e-05,
-      "loss": 6.4957,
+      "learning_rate": 0.0001791329479768786,
+      "loss": 7.0024,
       "step": 2396
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.96242774566474e-05,
-      "loss": 6.2053,
+      "learning_rate": 0.00017904624277456646,
+      "loss": 6.8538,
       "step": 2397
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.9595375722543357e-05,
-      "loss": 6.1801,
+      "learning_rate": 0.00017895953757225434,
+      "loss": 6.8502,
       "step": 2398
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.956647398843931e-05,
-      "loss": 6.1226,
+      "learning_rate": 0.00017887283236994218,
+      "loss": 6.8298,
       "step": 2399
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.9537572254335263e-05,
-      "loss": 6.1231,
+      "learning_rate": 0.00017878612716763006,
+      "loss": 6.875,
       "step": 2400
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.950867052023121e-05,
-      "loss": 6.3284,
+      "learning_rate": 0.0001786994219653179,
+      "loss": 6.9769,
       "step": 2401
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.9479768786127164e-05,
-      "loss": 6.2302,
+      "learning_rate": 0.00017861271676300575,
+      "loss": 6.8847,
       "step": 2402
     },
     {
       "epoch": 2.69,
-      "learning_rate": 5.945086705202313e-05,
-      "loss": 6.254,
+      "learning_rate": 0.00017852601156069362,
+      "loss": 6.917,
       "step": 2403
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.9421965317919084e-05,
-      "loss": 6.29,
+      "learning_rate": 0.00017843930635838147,
+      "loss": 6.9771,
       "step": 2404
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.939306358381503e-05,
-      "loss": 6.1823,
+      "learning_rate": 0.00017835260115606937,
+      "loss": 6.9121,
       "step": 2405
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.9364161849710984e-05,
-      "loss": 6.4198,
+      "learning_rate": 0.00017826589595375722,
+      "loss": 7.1342,
       "step": 2406
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.933526011560694e-05,
-      "loss": 6.2248,
+      "learning_rate": 0.00017817919075144506,
+      "loss": 7.0019,
       "step": 2407
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.930635838150289e-05,
-      "loss": 6.1113,
+      "learning_rate": 0.00017809248554913294,
+      "loss": 6.9613,
       "step": 2408
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.9277456647398844e-05,
-      "loss": 6.0467,
+      "learning_rate": 0.00017800578034682079,
+      "loss": 6.8714,
       "step": 2409
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.92485549132948e-05,
-      "loss": 6.0394,
+      "learning_rate": 0.00017791907514450866,
+      "loss": 6.8766,
       "step": 2410
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.921965317919076e-05,
-      "loss": 6.08,
+      "learning_rate": 0.0001778323699421965,
+      "loss": 6.9889,
       "step": 2411
     },
     {
       "epoch": 2.7,
-      "learning_rate": 5.919075144508671e-05,
-      "loss": 6.1144,
+      "learning_rate": 0.00017774566473988435,
+      "loss": 6.9258,
       "step": 2412
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.9161849710982665e-05,
-      "loss": 6.1284,
+      "learning_rate": 0.00017765895953757225,
+      "loss": 7.0195,
       "step": 2413
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.913294797687862e-05,
-      "loss": 6.0414,
+      "learning_rate": 0.0001775722543352601,
+      "loss": 6.9506,
       "step": 2414
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.910404624277457e-05,
-      "loss": 6.0091,
+      "learning_rate": 0.00017748554913294797,
+      "loss": 6.925,
       "step": 2415
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.907514450867052e-05,
-      "loss": 6.1747,
+      "learning_rate": 0.00017739884393063582,
+      "loss": 7.0695,
       "step": 2416
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.904624277456647e-05,
-      "loss": 5.9245,
+      "learning_rate": 0.0001773121387283237,
+      "loss": 6.8615,
       "step": 2417
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.9017341040462425e-05,
-      "loss": 6.1485,
+      "learning_rate": 0.00017722543352601154,
+      "loss": 7.0563,
       "step": 2418
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.898843930635839e-05,
-      "loss": 5.8241,
+      "learning_rate": 0.0001771387283236994,
+      "loss": 6.8383,
       "step": 2419
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.895953757225434e-05,
-      "loss": 6.0084,
+      "learning_rate": 0.00017705202312138726,
+      "loss": 6.9552,
       "step": 2420
     },
     {
       "epoch": 2.71,
-      "learning_rate": 5.893063583815029e-05,
-      "loss": 5.9521,
+      "learning_rate": 0.00017696531791907514,
+      "loss": 7.0038,
       "step": 2421
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.8901734104046246e-05,
-      "loss": 5.945,
+      "learning_rate": 0.000176878612716763,
+      "loss": 7.0174,
       "step": 2422
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.88728323699422e-05,
-      "loss": 5.7001,
+      "learning_rate": 0.00017679190751445086,
+      "loss": 6.7948,
       "step": 2423
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.884393063583815e-05,
-      "loss": 5.7121,
+      "learning_rate": 0.0001767052023121387,
+      "loss": 6.8706,
       "step": 2424
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.8815028901734106e-05,
-      "loss": 6.015,
+      "learning_rate": 0.00017661849710982658,
+      "loss": 7.0729,
       "step": 2425
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.878612716763005e-05,
-      "loss": 5.6923,
+      "learning_rate": 0.00017653179190751442,
+      "loss": 6.8393,
       "step": 2426
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.875722543352602e-05,
-      "loss": 5.6962,
+      "learning_rate": 0.0001764450867052023,
+      "loss": 6.9299,
       "step": 2427
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.8728323699421973e-05,
-      "loss": 5.7361,
+      "learning_rate": 0.00017635838150289015,
+      "loss": 6.9795,
       "step": 2428
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.869942196531793e-05,
-      "loss": 5.6363,
+      "learning_rate": 0.00017627167630057805,
+      "loss": 6.9804,
       "step": 2429
     },
     {
       "epoch": 2.72,
-      "learning_rate": 5.8670520231213874e-05,
-      "loss": 5.3284,
+      "learning_rate": 0.0001761849710982659,
+      "loss": 6.8311,
       "step": 2430
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.864161849710983e-05,
-      "loss": 5.4862,
+      "learning_rate": 0.00017609826589595374,
+      "loss": 6.8772,
       "step": 2431
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.861271676300578e-05,
-      "loss": 5.4113,
+      "learning_rate": 0.00017601156069364161,
+      "loss": 6.8287,
       "step": 2432
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.8583815028901734e-05,
-      "loss": 5.1162,
+      "learning_rate": 0.00017592485549132946,
+      "loss": 6.8072,
       "step": 2433
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.855491329479769e-05,
-      "loss": 4.2607,
+      "learning_rate": 0.0001758381502890173,
+      "loss": 6.39,
       "step": 2434
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.852601156069365e-05,
-      "loss": 6.6957,
+      "learning_rate": 0.00017575144508670518,
+      "loss": 7.1329,
       "step": 2435
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.84971098265896e-05,
-      "loss": 6.5161,
+      "learning_rate": 0.00017566473988439303,
+      "loss": 6.932,
       "step": 2436
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.8468208092485555e-05,
-      "loss": 6.5731,
+      "learning_rate": 0.00017557803468208093,
+      "loss": 7.0592,
       "step": 2437
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.843930635838151e-05,
-      "loss": 6.5048,
+      "learning_rate": 0.00017549132947976878,
+      "loss": 7.0125,
       "step": 2438
     },
     {
       "epoch": 2.73,
-      "learning_rate": 5.841040462427746e-05,
-      "loss": 6.3509,
+      "learning_rate": 0.00017540462427745665,
+      "loss": 6.9536,
       "step": 2439
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.8381502890173415e-05,
-      "loss": 6.5262,
+      "learning_rate": 0.0001753179190751445,
+      "loss": 7.038,
       "step": 2440
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.835260115606936e-05,
-      "loss": 6.4566,
+      "learning_rate": 0.00017523121387283234,
+      "loss": 7.0297,
       "step": 2441
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.8323699421965315e-05,
-      "loss": 6.2889,
+      "learning_rate": 0.00017514450867052022,
+      "loss": 6.9818,
       "step": 2442
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.829479768786127e-05,
-      "loss": 6.2159,
+      "learning_rate": 0.00017505780346820806,
+      "loss": 6.9385,
       "step": 2443
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.8265895953757235e-05,
-      "loss": 6.1635,
+      "learning_rate": 0.00017497109826589594,
+      "loss": 6.9522,
       "step": 2444
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.823699421965318e-05,
-      "loss": 6.0665,
+      "learning_rate": 0.00017488439306358378,
+      "loss": 6.9319,
       "step": 2445
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.8208092485549136e-05,
-      "loss": 6.1667,
+      "learning_rate": 0.00017479768786127169,
+      "loss": 6.903,
       "step": 2446
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.817919075144509e-05,
-      "loss": 6.1322,
+      "learning_rate": 0.00017471098265895953,
+      "loss": 6.9252,
       "step": 2447
     },
     {
       "epoch": 2.74,
-      "learning_rate": 5.815028901734104e-05,
-      "loss": 6.3434,
+      "learning_rate": 0.00017462427745664738,
+      "loss": 7.1435,
       "step": 2448
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.8121387283236996e-05,
-      "loss": 6.2608,
+      "learning_rate": 0.00017453757225433525,
+      "loss": 7.0086,
       "step": 2449
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.809248554913295e-05,
-      "loss": 6.2628,
+      "learning_rate": 0.0001744508670520231,
+      "loss": 7.077,
       "step": 2450
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.8063583815028896e-05,
-      "loss": 6.2231,
+      "learning_rate": 0.00017436416184971095,
+      "loss": 7.0884,
       "step": 2451
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.803468208092486e-05,
-      "loss": 6.1289,
+      "learning_rate": 0.00017427745664739882,
+      "loss": 7.0037,
       "step": 2452
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.8005780346820816e-05,
-      "loss": 6.0147,
+      "learning_rate": 0.00017419075144508667,
+      "loss": 6.8476,
       "step": 2453
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.797687861271677e-05,
-      "loss": 6.2329,
+      "learning_rate": 0.00017410404624277457,
+      "loss": 7.0687,
       "step": 2454
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.7947976878612717e-05,
-      "loss": 5.9366,
+      "learning_rate": 0.00017401734104046242,
+      "loss": 6.9118,
       "step": 2455
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.791907514450867e-05,
-      "loss": 6.1205,
+      "learning_rate": 0.0001739306358381503,
+      "loss": 7.0461,
       "step": 2456
     },
     {
       "epoch": 2.75,
-      "learning_rate": 5.7890173410404623e-05,
-      "loss": 6.047,
+      "learning_rate": 0.00017384393063583814,
+      "loss": 6.9577,
       "step": 2457
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.786127167630058e-05,
-      "loss": 5.9701,
+      "learning_rate": 0.00017375722543352598,
+      "loss": 6.9937,
       "step": 2458
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.783236994219653e-05,
-      "loss": 5.9948,
+      "learning_rate": 0.00017367052023121386,
+      "loss": 6.9027,
       "step": 2459
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.780346820809249e-05,
-      "loss": 5.9934,
+      "learning_rate": 0.0001735838150289017,
+      "loss": 6.9809,
       "step": 2460
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.7774566473988444e-05,
-      "loss": 6.0904,
+      "learning_rate": 0.00017349710982658958,
+      "loss": 7.0103,
       "step": 2461
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.77456647398844e-05,
-      "loss": 5.9399,
+      "learning_rate": 0.00017341040462427745,
+      "loss": 6.9173,
       "step": 2462
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.771676300578035e-05,
-      "loss": 6.0151,
+      "learning_rate": 0.00017332369942196533,
+      "loss": 7.0391,
       "step": 2463
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.7687861271676304e-05,
-      "loss": 6.0288,
+      "learning_rate": 0.00017323699421965317,
+      "loss": 6.9831,
       "step": 2464
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.765895953757226e-05,
-      "loss": 5.9908,
+      "learning_rate": 0.00017315028901734102,
+      "loss": 6.9748,
       "step": 2465
     },
     {
       "epoch": 2.76,
-      "learning_rate": 5.7630057803468204e-05,
-      "loss": 5.8206,
+      "learning_rate": 0.0001730635838150289,
+      "loss": 6.9136,
       "step": 2466
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.760115606936416e-05,
-      "loss": 5.9951,
+      "learning_rate": 0.00017297687861271674,
+      "loss": 6.9688,
       "step": 2467
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.7572254335260125e-05,
-      "loss": 6.0179,
+      "learning_rate": 0.00017289017341040459,
+      "loss": 7.0048,
       "step": 2468
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.754335260115608e-05,
-      "loss": 5.8911,
+      "learning_rate": 0.00017280346820809246,
+      "loss": 6.9616,
       "step": 2469
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.7514450867052025e-05,
-      "loss": 5.8477,
+      "learning_rate": 0.00017271676300578033,
+      "loss": 6.9987,
       "step": 2470
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.748554913294798e-05,
-      "loss": 5.7944,
+      "learning_rate": 0.0001726300578034682,
+      "loss": 6.8915,
       "step": 2471
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.745664739884393e-05,
-      "loss": 5.9823,
+      "learning_rate": 0.00017254335260115605,
+      "loss": 6.9757,
       "step": 2472
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.7427745664739885e-05,
-      "loss": 5.9991,
+      "learning_rate": 0.00017245664739884393,
+      "loss": 6.9635,
       "step": 2473
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.739884393063584e-05,
-      "loss": 5.8527,
+      "learning_rate": 0.00017236994219653178,
+      "loss": 6.8496,
       "step": 2474
     },
     {
       "epoch": 2.77,
-      "learning_rate": 5.736994219653179e-05,
-      "loss": 5.8254,
+      "learning_rate": 0.00017228323699421962,
+      "loss": 7.0491,
       "step": 2475
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.734104046242775e-05,
-      "loss": 5.9536,
+      "learning_rate": 0.0001721965317919075,
+      "loss": 7.0934,
       "step": 2476
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.7312138728323706e-05,
-      "loss": 5.6554,
+      "learning_rate": 0.00017210982658959534,
+      "loss": 6.8463,
       "step": 2477
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.728323699421966e-05,
-      "loss": 5.6078,
+      "learning_rate": 0.00017202312138728324,
+      "loss": 6.9276,
       "step": 2478
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.725433526011561e-05,
-      "loss": 5.5089,
+      "learning_rate": 0.0001719364161849711,
+      "loss": 6.9239,
       "step": 2479
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.722543352601156e-05,
-      "loss": 5.3939,
+      "learning_rate": 0.00017184971098265894,
+      "loss": 6.8117,
       "step": 2480
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.719653179190751e-05,
-      "loss": 5.2429,
+      "learning_rate": 0.0001717630057803468,
+      "loss": 6.7961,
       "step": 2481
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.7167630057803466e-05,
-      "loss": 5.2406,
+      "learning_rate": 0.00017167630057803466,
+      "loss": 6.852,
       "step": 2482
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.713872832369942e-05,
-      "loss": 5.0514,
+      "learning_rate": 0.00017158959537572253,
+      "loss": 6.7669,
       "step": 2483
     },
     {
       "epoch": 2.78,
-      "learning_rate": 5.710982658959538e-05,
-      "loss": 5.015,
+      "learning_rate": 0.00017150289017341038,
+      "loss": 6.7423,
       "step": 2484
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.7080924855491333e-05,
-      "loss": 6.6695,
+      "learning_rate": 0.00017141618497109823,
+      "loss": 6.9142,
       "step": 2485
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.705202312138729e-05,
-      "loss": 6.7054,
+      "learning_rate": 0.00017132947976878613,
+      "loss": 6.9564,
       "step": 2486
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.702312138728324e-05,
-      "loss": 6.6182,
+      "learning_rate": 0.00017124277456647397,
+      "loss": 6.9523,
       "step": 2487
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.6994219653179194e-05,
-      "loss": 6.637,
+      "learning_rate": 0.00017115606936416185,
+      "loss": 7.0529,
       "step": 2488
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.696531791907515e-05,
-      "loss": 6.5276,
+      "learning_rate": 0.0001710693641618497,
+      "loss": 7.0312,
       "step": 2489
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.69364161849711e-05,
-      "loss": 6.3182,
+      "learning_rate": 0.00017098265895953757,
+      "loss": 6.9655,
       "step": 2490
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.690751445086705e-05,
-      "loss": 6.2394,
+      "learning_rate": 0.00017089595375722541,
+      "loss": 6.9094,
       "step": 2491
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.6878612716763014e-05,
-      "loss": 6.5477,
+      "learning_rate": 0.00017080924855491326,
+      "loss": 7.0808,
       "step": 2492
     },
     {
       "epoch": 2.79,
-      "learning_rate": 5.684971098265897e-05,
-      "loss": 6.3886,
+      "learning_rate": 0.00017072254335260114,
+      "loss": 6.919,
       "step": 2493
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.682080924855492e-05,
-      "loss": 6.3841,
+      "learning_rate": 0.000170635838150289,
+      "loss": 7.0591,
       "step": 2494
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.679190751445087e-05,
-      "loss": 5.9835,
+      "learning_rate": 0.00017054913294797688,
+      "loss": 6.8762,
       "step": 2495
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.676300578034682e-05,
-      "loss": 6.1689,
+      "learning_rate": 0.00017046242774566473,
+      "loss": 7.0097,
       "step": 2496
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.6734104046242775e-05,
-      "loss": 6.1597,
+      "learning_rate": 0.00017037572254335258,
+      "loss": 6.9099,
       "step": 2497
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.670520231213873e-05,
-      "loss": 5.9398,
+      "learning_rate": 0.00017028901734104045,
+      "loss": 6.7964,
       "step": 2498
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.667630057803468e-05,
-      "loss": 6.052,
+      "learning_rate": 0.0001702023121387283,
+      "loss": 6.8615,
       "step": 2499
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.664739884393064e-05,
-      "loss": 6.0075,
+      "learning_rate": 0.00017011560693641617,
+      "loss": 6.8527,
       "step": 2500
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.6618497109826595e-05,
-      "loss": 6.09,
+      "learning_rate": 0.00017002890173410402,
+      "loss": 6.9898,
       "step": 2501
     },
     {
       "epoch": 2.8,
-      "learning_rate": 5.658959537572255e-05,
-      "loss": 5.9364,
+      "learning_rate": 0.00016994219653179192,
+      "loss": 6.8627,
       "step": 2502
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.65606936416185e-05,
-      "loss": 5.7632,
+      "learning_rate": 0.00016985549132947977,
+      "loss": 6.7458,
       "step": 2503
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.6531791907514456e-05,
-      "loss": 5.9333,
+      "learning_rate": 0.0001697687861271676,
+      "loss": 6.9081,
       "step": 2504
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.65028901734104e-05,
-      "loss": 6.0012,
+      "learning_rate": 0.0001696820809248555,
+      "loss": 6.99,
       "step": 2505
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.6473988439306356e-05,
-      "loss": 5.9533,
+      "learning_rate": 0.00016959537572254333,
+      "loss": 6.9402,
       "step": 2506
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.644508670520231e-05,
-      "loss": 5.9672,
+      "learning_rate": 0.00016950867052023118,
+      "loss": 6.9278,
       "step": 2507
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.6416184971098276e-05,
-      "loss": 5.9034,
+      "learning_rate": 0.00016942196531791905,
+      "loss": 6.9155,
       "step": 2508
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.638728323699422e-05,
-      "loss": 5.9494,
+      "learning_rate": 0.0001693352601156069,
+      "loss": 6.9888,
       "step": 2509
     },
     {
       "epoch": 2.81,
-      "learning_rate": 5.6358381502890176e-05,
-      "loss": 5.9452,
+      "learning_rate": 0.0001692485549132948,
+      "loss": 6.9721,
       "step": 2510
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.632947976878613e-05,
-      "loss": 6.0606,
+      "learning_rate": 0.00016916184971098265,
+      "loss": 7.1012,
       "step": 2511
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.630057803468208e-05,
-      "loss": 5.8216,
+      "learning_rate": 0.00016907514450867052,
+      "loss": 6.8799,
       "step": 2512
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.627167630057804e-05,
-      "loss": 5.8857,
+      "learning_rate": 0.00016898843930635837,
+      "loss": 7.0338,
       "step": 2513
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.624277456647399e-05,
-      "loss": 5.8474,
+      "learning_rate": 0.00016890173410404622,
+      "loss": 6.9665,
       "step": 2514
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.6213872832369944e-05,
-      "loss": 6.1423,
+      "learning_rate": 0.0001688150289017341,
+      "loss": 7.0518,
       "step": 2515
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.6184971098265904e-05,
-      "loss": 5.7968,
+      "learning_rate": 0.00016872832369942194,
+      "loss": 6.9334,
       "step": 2516
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.615606936416186e-05,
-      "loss": 5.8297,
+      "learning_rate": 0.0001686416184971098,
+      "loss": 6.9031,
       "step": 2517
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.612716763005781e-05,
-      "loss": 6.0192,
+      "learning_rate": 0.00016855491329479768,
+      "loss": 7.0161,
       "step": 2518
     },
     {
       "epoch": 2.82,
-      "learning_rate": 5.6098265895953764e-05,
-      "loss": 5.8885,
+      "learning_rate": 0.00016846820809248556,
+      "loss": 6.9369,
       "step": 2519
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.606936416184971e-05,
-      "loss": 5.7829,
+      "learning_rate": 0.0001683815028901734,
+      "loss": 6.9052,
       "step": 2520
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.6040462427745664e-05,
-      "loss": 6.0151,
+      "learning_rate": 0.00016829479768786125,
+      "loss": 7.0315,
       "step": 2521
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.601156069364162e-05,
-      "loss": 5.9118,
+      "learning_rate": 0.00016820809248554913,
+      "loss": 7.0277,
       "step": 2522
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.598265895953757e-05,
-      "loss": 5.8672,
+      "learning_rate": 0.00016812138728323697,
+      "loss": 7.0478,
       "step": 2523
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.595375722543353e-05,
-      "loss": 5.6602,
+      "learning_rate": 0.00016803468208092482,
+      "loss": 6.9377,
       "step": 2524
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.5924855491329485e-05,
-      "loss": 5.7943,
+      "learning_rate": 0.0001679479768786127,
+      "loss": 6.996,
       "step": 2525
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.589595375722544e-05,
-      "loss": 5.9269,
+      "learning_rate": 0.00016786127167630057,
+      "loss": 7.0468,
       "step": 2526
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.586705202312139e-05,
-      "loss": 5.8437,
+      "learning_rate": 0.00016777456647398844,
+      "loss": 7.1163,
       "step": 2527
     },
     {
       "epoch": 2.83,
-      "learning_rate": 5.5838150289017345e-05,
-      "loss": 5.7986,
+      "learning_rate": 0.0001676878612716763,
+      "loss": 6.9643,
       "step": 2528
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.58092485549133e-05,
-      "loss": 5.6465,
+      "learning_rate": 0.00016760115606936416,
+      "loss": 6.83,
       "step": 2529
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.5780346820809245e-05,
-      "loss": 5.6277,
+      "learning_rate": 0.000167514450867052,
+      "loss": 6.9303,
       "step": 2530
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.57514450867052e-05,
-      "loss": 5.2436,
+      "learning_rate": 0.00016742774566473986,
+      "loss": 6.8768,
       "step": 2531
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.5722543352601166e-05,
-      "loss": 5.2752,
+      "learning_rate": 0.00016734104046242773,
+      "loss": 6.9109,
       "step": 2532
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.569364161849712e-05,
-      "loss": 5.0661,
+      "learning_rate": 0.00016725433526011558,
+      "loss": 6.7552,
       "step": 2533
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.5664739884393066e-05,
-      "loss": 4.6427,
+      "learning_rate": 0.00016716763005780348,
+      "loss": 6.5349,
       "step": 2534
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.563583815028902e-05,
-      "loss": 6.6538,
+      "learning_rate": 0.00016708092485549132,
+      "loss": 7.0458,
       "step": 2535
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.560693641618497e-05,
-      "loss": 6.5505,
+      "learning_rate": 0.0001669942196531792,
+      "loss": 6.9765,
       "step": 2536
     },
     {
       "epoch": 2.84,
-      "learning_rate": 5.5578034682080926e-05,
-      "loss": 6.4498,
+      "learning_rate": 0.00016690751445086704,
+      "loss": 6.8965,
       "step": 2537
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.554913294797688e-05,
-      "loss": 6.5135,
+      "learning_rate": 0.0001668208092485549,
+      "loss": 7.0277,
       "step": 2538
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.552023121387283e-05,
-      "loss": 6.2822,
+      "learning_rate": 0.00016673410404624277,
+      "loss": 6.8712,
       "step": 2539
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.5491329479768787e-05,
-      "loss": 6.3325,
+      "learning_rate": 0.0001666473988439306,
+      "loss": 6.9982,
       "step": 2540
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.546242774566475e-05,
-      "loss": 6.3239,
+      "learning_rate": 0.00016656069364161846,
+      "loss": 7.0451,
       "step": 2541
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.54335260115607e-05,
-      "loss": 6.2712,
+      "learning_rate": 0.00016647398843930633,
+      "loss": 7.0646,
       "step": 2542
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.5404624277456654e-05,
-      "loss": 6.1821,
+      "learning_rate": 0.0001663872832369942,
+      "loss": 6.9874,
       "step": 2543
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.537572254335261e-05,
-      "loss": 6.1806,
+      "learning_rate": 0.00016630057803468208,
+      "loss": 7.0034,
       "step": 2544
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.5346820809248554e-05,
-      "loss": 6.1423,
+      "learning_rate": 0.00016621387283236993,
+      "loss": 6.9434,
       "step": 2545
     },
     {
       "epoch": 2.85,
-      "learning_rate": 5.531791907514451e-05,
-      "loss": 6.1479,
+      "learning_rate": 0.0001661271676300578,
+      "loss": 6.9678,
       "step": 2546
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.528901734104046e-05,
-      "loss": 6.1203,
+      "learning_rate": 0.00016604046242774565,
+      "loss": 6.9459,
       "step": 2547
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.5260115606936414e-05,
-      "loss": 5.9355,
+      "learning_rate": 0.0001659537572254335,
+      "loss": 6.864,
       "step": 2548
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.5231213872832374e-05,
-      "loss": 6.0391,
+      "learning_rate": 0.00016586705202312137,
+      "loss": 7.0096,
       "step": 2549
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.520231213872833e-05,
-      "loss": 5.9825,
+      "learning_rate": 0.00016578034682080922,
+      "loss": 6.9368,
       "step": 2550
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.517341040462428e-05,
-      "loss": 5.993,
+      "learning_rate": 0.00016569364161849712,
+      "loss": 6.9355,
       "step": 2551
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.5144508670520235e-05,
-      "loss": 5.9679,
+      "learning_rate": 0.00016560693641618496,
+      "loss": 6.941,
       "step": 2552
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.511560693641619e-05,
-      "loss": 5.8285,
+      "learning_rate": 0.0001655202312138728,
+      "loss": 6.8743,
       "step": 2553
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.508670520231214e-05,
-      "loss": 5.9016,
+      "learning_rate": 0.00016543352601156068,
+      "loss": 6.9427,
       "step": 2554
     },
     {
       "epoch": 2.86,
-      "learning_rate": 5.505780346820809e-05,
-      "loss": 5.7136,
+      "learning_rate": 0.00016534682080924853,
+      "loss": 6.8001,
       "step": 2555
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.502890173410404e-05,
-      "loss": 5.8885,
+      "learning_rate": 0.0001652601156069364,
+      "loss": 6.9106,
       "step": 2556
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.500000000000001e-05,
-      "loss": 5.9373,
+      "learning_rate": 0.00016517341040462425,
+      "loss": 6.954,
       "step": 2557
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.497109826589596e-05,
-      "loss": 5.7741,
+      "learning_rate": 0.0001650867052023121,
+      "loss": 6.8515,
       "step": 2558
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.494219653179191e-05,
-      "loss": 6.0578,
+      "learning_rate": 0.000165,
+      "loss": 7.0369,
       "step": 2559
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.491329479768786e-05,
-      "loss": 5.8247,
+      "learning_rate": 0.00016491329479768785,
+      "loss": 6.8513,
       "step": 2560
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.4884393063583816e-05,
-      "loss": 5.824,
+      "learning_rate": 0.00016482658959537572,
+      "loss": 6.852,
       "step": 2561
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.485549132947977e-05,
-      "loss": 5.7774,
+      "learning_rate": 0.00016473988439306357,
+      "loss": 6.8641,
       "step": 2562
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.482658959537572e-05,
-      "loss": 5.8703,
+      "learning_rate": 0.00016465317919075144,
+      "loss": 6.8602,
       "step": 2563
     },
     {
       "epoch": 2.87,
-      "learning_rate": 5.4797687861271676e-05,
-      "loss": 6.0401,
+      "learning_rate": 0.0001645664739884393,
+      "loss": 6.9844,
       "step": 2564
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.4768786127167636e-05,
-      "loss": 5.8111,
+      "learning_rate": 0.00016447976878612713,
+      "loss": 6.8149,
       "step": 2565
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.473988439306359e-05,
-      "loss": 5.7252,
+      "learning_rate": 0.000164393063583815,
+      "loss": 6.8529,
       "step": 2566
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.471098265895954e-05,
-      "loss": 5.7807,
+      "learning_rate": 0.00016430635838150288,
+      "loss": 6.7932,
       "step": 2567
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.46820809248555e-05,
-      "loss": 5.8583,
+      "learning_rate": 0.00016421965317919076,
+      "loss": 6.9129,
       "step": 2568
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.465317919075145e-05,
-      "loss": 5.9531,
+      "learning_rate": 0.0001641329479768786,
+      "loss": 7.0296,
       "step": 2569
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.46242774566474e-05,
-      "loss": 5.9381,
+      "learning_rate": 0.00016404624277456645,
+      "loss": 6.8953,
       "step": 2570
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.459537572254335e-05,
-      "loss": 5.9115,
+      "learning_rate": 0.00016395953757225432,
+      "loss": 7.0389,
       "step": 2571
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.4566473988439304e-05,
-      "loss": 5.7482,
+      "learning_rate": 0.00016387283236994217,
+      "loss": 6.9892,
       "step": 2572
     },
     {
       "epoch": 2.88,
-      "learning_rate": 5.453757225433527e-05,
-      "loss": 5.7683,
+      "learning_rate": 0.00016378612716763004,
+      "loss": 6.8557,
       "step": 2573
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.450867052023122e-05,
-      "loss": 5.6568,
+      "learning_rate": 0.0001636994219653179,
+      "loss": 6.9692,
       "step": 2574
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.447976878612717e-05,
-      "loss": 5.7387,
+      "learning_rate": 0.0001636127167630058,
+      "loss": 7.0066,
       "step": 2575
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.4450867052023124e-05,
-      "loss": 5.6568,
+      "learning_rate": 0.00016352601156069364,
+      "loss": 6.9226,
       "step": 2576
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.442196531791908e-05,
-      "loss": 5.4416,
+      "learning_rate": 0.00016343930635838149,
+      "loss": 6.9051,
       "step": 2577
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.439306358381503e-05,
-      "loss": 5.4158,
+      "learning_rate": 0.00016335260115606936,
+      "loss": 6.9289,
       "step": 2578
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.4364161849710985e-05,
-      "loss": 5.4182,
+      "learning_rate": 0.0001632658959537572,
+      "loss": 6.8415,
       "step": 2579
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.433526011560693e-05,
-      "loss": 5.3856,
+      "learning_rate": 0.00016317919075144505,
+      "loss": 6.8868,
       "step": 2580
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.43063583815029e-05,
-      "loss": 5.3328,
+      "learning_rate": 0.00016309248554913293,
+      "loss": 6.8468,
       "step": 2581
     },
     {
       "epoch": 2.89,
-      "learning_rate": 5.427745664739885e-05,
-      "loss": 5.1147,
+      "learning_rate": 0.00016300578034682077,
+      "loss": 6.9028,
       "step": 2582
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.4248554913294805e-05,
-      "loss": 4.8674,
+      "learning_rate": 0.00016291907514450867,
+      "loss": 6.7563,
       "step": 2583
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.421965317919075e-05,
-      "loss": 4.653,
+      "learning_rate": 0.00016283236994219652,
+      "loss": 6.4673,
       "step": 2584
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.4190751445086705e-05,
-      "loss": 6.8815,
+      "learning_rate": 0.0001627456647398844,
+      "loss": 7.0659,
       "step": 2585
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.416184971098266e-05,
-      "loss": 6.6353,
+      "learning_rate": 0.00016265895953757224,
+      "loss": 6.8638,
       "step": 2586
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.413294797687861e-05,
-      "loss": 6.546,
+      "learning_rate": 0.0001625722543352601,
+      "loss": 6.9273,
       "step": 2587
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.4104046242774566e-05,
-      "loss": 6.5388,
+      "learning_rate": 0.00016248554913294796,
+      "loss": 7.0073,
       "step": 2588
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.4075144508670526e-05,
-      "loss": 6.4214,
+      "learning_rate": 0.0001623988439306358,
+      "loss": 6.9986,
       "step": 2589
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.404624277456648e-05,
-      "loss": 6.5851,
+      "learning_rate": 0.00016231213872832368,
+      "loss": 7.1122,
       "step": 2590
     },
     {
       "epoch": 2.9,
-      "learning_rate": 5.401734104046243e-05,
-      "loss": 6.3229,
+      "learning_rate": 0.00016222543352601156,
+      "loss": 6.9257,
       "step": 2591
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.3988439306358386e-05,
-      "loss": 6.3243,
+      "learning_rate": 0.00016213872832369943,
+      "loss": 7.0365,
       "step": 2592
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.395953757225434e-05,
-      "loss": 6.0833,
+      "learning_rate": 0.00016205202312138728,
+      "loss": 6.9272,
       "step": 2593
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.3930635838150286e-05,
-      "loss": 6.0838,
+      "learning_rate": 0.00016196531791907512,
+      "loss": 6.8887,
       "step": 2594
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.390173410404624e-05,
-      "loss": 6.0599,
+      "learning_rate": 0.000161878612716763,
+      "loss": 6.9246,
       "step": 2595
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.387283236994219e-05,
-      "loss": 6.0459,
+      "learning_rate": 0.00016179190751445085,
+      "loss": 6.8835,
       "step": 2596
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.384393063583816e-05,
-      "loss": 6.2026,
+      "learning_rate": 0.0001617052023121387,
+      "loss": 6.9188,
       "step": 2597
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.3815028901734114e-05,
-      "loss": 6.0522,
+      "learning_rate": 0.00016161849710982657,
+      "loss": 6.9088,
       "step": 2598
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.378612716763006e-05,
-      "loss": 6.0036,
+      "learning_rate": 0.00016153179190751444,
+      "loss": 6.9205,
       "step": 2599
     },
     {
       "epoch": 2.91,
-      "learning_rate": 5.3757225433526014e-05,
-      "loss": 6.077,
+      "learning_rate": 0.00016144508670520231,
+      "loss": 6.9721,
       "step": 2600
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.372832369942197e-05,
-      "loss": 6.0345,
+      "learning_rate": 0.00016135838150289016,
+      "loss": 6.9858,
       "step": 2601
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.369942196531792e-05,
-      "loss": 5.9289,
+      "learning_rate": 0.00016127167630057803,
+      "loss": 6.9345,
       "step": 2602
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.3670520231213874e-05,
-      "loss": 5.9492,
+      "learning_rate": 0.00016118497109826588,
+      "loss": 6.9512,
       "step": 2603
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.364161849710983e-05,
-      "loss": 6.1015,
+      "learning_rate": 0.00016109826589595373,
+      "loss": 7.0434,
       "step": 2604
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.361271676300579e-05,
-      "loss": 5.8207,
+      "learning_rate": 0.0001610115606936416,
+      "loss": 6.9133,
       "step": 2605
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.358381502890174e-05,
-      "loss": 5.7709,
+      "learning_rate": 0.00016092485549132945,
+      "loss": 6.8621,
       "step": 2606
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.3554913294797695e-05,
-      "loss": 5.8232,
+      "learning_rate": 0.00016083815028901735,
+      "loss": 6.914,
       "step": 2607
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.352601156069365e-05,
-      "loss": 5.7237,
+      "learning_rate": 0.0001607514450867052,
+      "loss": 6.7804,
       "step": 2608
     },
     {
       "epoch": 2.92,
-      "learning_rate": 5.3497109826589595e-05,
-      "loss": 5.7578,
+      "learning_rate": 0.00016066473988439304,
+      "loss": 6.8874,
       "step": 2609
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.346820809248555e-05,
-      "loss": 5.8691,
+      "learning_rate": 0.00016057803468208092,
+      "loss": 6.8911,
       "step": 2610
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.34393063583815e-05,
-      "loss": 5.8066,
+      "learning_rate": 0.00016049132947976876,
+      "loss": 6.9373,
       "step": 2611
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.3410404624277455e-05,
-      "loss": 5.7801,
+      "learning_rate": 0.00016040462427745664,
+      "loss": 6.8217,
       "step": 2612
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.3381502890173415e-05,
-      "loss": 6.0056,
+      "learning_rate": 0.00016031791907514448,
+      "loss": 7.0281,
       "step": 2613
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.335260115606937e-05,
-      "loss": 5.8437,
+      "learning_rate": 0.00016023121387283233,
+      "loss": 6.9675,
       "step": 2614
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.332369942196532e-05,
-      "loss": 5.9056,
+      "learning_rate": 0.00016014450867052023,
+      "loss": 7.0156,
       "step": 2615
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.3294797687861276e-05,
-      "loss": 5.8413,
+      "learning_rate": 0.00016005780346820808,
+      "loss": 7.0103,
       "step": 2616
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.326589595375723e-05,
-      "loss": 5.8326,
+      "learning_rate": 0.00015997109826589595,
+      "loss": 6.9716,
       "step": 2617
     },
     {
       "epoch": 2.93,
-      "learning_rate": 5.323699421965318e-05,
-      "loss": 5.6881,
+      "learning_rate": 0.0001598843930635838,
+      "loss": 6.8046,
       "step": 2618
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.320809248554913e-05,
-      "loss": 5.7689,
+      "learning_rate": 0.00015979768786127167,
+      "loss": 6.9888,
       "step": 2619
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.317919075144508e-05,
-      "loss": 5.7071,
+      "learning_rate": 0.00015971098265895952,
+      "loss": 6.8018,
       "step": 2620
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.315028901734105e-05,
-      "loss": 5.6013,
+      "learning_rate": 0.00015962427745664737,
+      "loss": 6.7999,
       "step": 2621
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.3121387283237e-05,
-      "loss": 5.6888,
+      "learning_rate": 0.00015953757225433524,
+      "loss": 6.8867,
       "step": 2622
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.3092485549132957e-05,
-      "loss": 5.7929,
+      "learning_rate": 0.00015945086705202312,
+      "loss": 6.9545,
       "step": 2623
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.30635838150289e-05,
-      "loss": 5.945,
+      "learning_rate": 0.000159364161849711,
+      "loss": 7.1406,
       "step": 2624
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.303468208092486e-05,
-      "loss": 5.7009,
+      "learning_rate": 0.00015927745664739884,
+      "loss": 6.7864,
       "step": 2625
     },
     {
       "epoch": 2.94,
-      "learning_rate": 5.300578034682081e-05,
-      "loss": 5.5919,
+      "learning_rate": 0.00015919075144508668,
+      "loss": 6.8791,
       "step": 2626
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.2976878612716764e-05,
-      "loss": 5.3174,
+      "learning_rate": 0.00015910404624277456,
+      "loss": 6.7152,
       "step": 2627
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.294797687861272e-05,
-      "loss": 5.3759,
+      "learning_rate": 0.0001590173410404624,
+      "loss": 6.7868,
       "step": 2628
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.291907514450868e-05,
-      "loss": 5.3742,
+      "learning_rate": 0.00015893063583815028,
+      "loss": 6.9441,
       "step": 2629
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.289017341040463e-05,
-      "loss": 5.5109,
+      "learning_rate": 0.00015884393063583812,
+      "loss": 7.0525,
       "step": 2630
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.2861271676300584e-05,
-      "loss": 4.9926,
+      "learning_rate": 0.00015875722543352602,
+      "loss": 6.6617,
       "step": 2631
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.283236994219654e-05,
-      "loss": 4.8586,
+      "learning_rate": 0.00015867052023121387,
+      "loss": 6.6296,
       "step": 2632
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.280346820809249e-05,
-      "loss": 4.6979,
+      "learning_rate": 0.00015858381502890172,
+      "loss": 6.6638,
       "step": 2633
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.277456647398844e-05,
-      "loss": 4.4922,
+      "learning_rate": 0.0001584971098265896,
+      "loss": 6.5228,
       "step": 2634
     },
     {
       "epoch": 2.95,
-      "learning_rate": 5.274566473988439e-05,
-      "loss": 6.9104,
+      "learning_rate": 0.00015841040462427744,
+      "loss": 7.0448,
       "step": 2635
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.2716763005780345e-05,
-      "loss": 6.848,
+      "learning_rate": 0.0001583236994219653,
+      "loss": 6.9681,
       "step": 2636
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.26878612716763e-05,
-      "loss": 6.925,
+      "learning_rate": 0.00015823699421965316,
+      "loss": 7.1269,
       "step": 2637
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.265895953757226e-05,
-      "loss": 6.5791,
+      "learning_rate": 0.000158150289017341,
+      "loss": 6.8651,
       "step": 2638
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.263005780346821e-05,
-      "loss": 6.5502,
+      "learning_rate": 0.00015806358381502888,
+      "loss": 6.895,
       "step": 2639
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.2601156069364165e-05,
-      "loss": 6.4375,
+      "learning_rate": 0.00015797687861271675,
+      "loss": 6.8139,
       "step": 2640
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.257225433526012e-05,
-      "loss": 6.475,
+      "learning_rate": 0.00015789017341040463,
+      "loss": 6.9095,
       "step": 2641
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.254335260115607e-05,
-      "loss": 6.4316,
+      "learning_rate": 0.00015780346820809248,
+      "loss": 6.9313,
       "step": 2642
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.2514450867052025e-05,
-      "loss": 6.4559,
+      "learning_rate": 0.00015771676300578032,
+      "loss": 6.9563,
       "step": 2643
     },
     {
       "epoch": 2.96,
-      "learning_rate": 5.248554913294797e-05,
-      "loss": 6.104,
+      "learning_rate": 0.0001576300578034682,
+      "loss": 6.8161,
       "step": 2644
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.2456647398843926e-05,
-      "loss": 6.136,
+      "learning_rate": 0.00015754335260115604,
+      "loss": 6.8745,
       "step": 2645
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.242774566473989e-05,
-      "loss": 6.2753,
+      "learning_rate": 0.00015745664739884392,
+      "loss": 6.9626,
       "step": 2646
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.2398843930635846e-05,
-      "loss": 6.2575,
+      "learning_rate": 0.00015736994219653176,
+      "loss": 7.0282,
       "step": 2647
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.236994219653179e-05,
-      "loss": 6.1208,
+      "learning_rate": 0.00015728323699421966,
+      "loss": 6.9532,
       "step": 2648
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.2341040462427746e-05,
-      "loss": 6.303,
+      "learning_rate": 0.0001571965317919075,
+      "loss": 7.0777,
       "step": 2649
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.23121387283237e-05,
-      "loss": 5.9872,
+      "learning_rate": 0.00015710982658959536,
+      "loss": 6.8689,
       "step": 2650
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.228323699421965e-05,
-      "loss": 6.1837,
+      "learning_rate": 0.00015702312138728323,
+      "loss": 7.0158,
       "step": 2651
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.2254335260115606e-05,
-      "loss": 6.1593,
+      "learning_rate": 0.00015693641618497108,
+      "loss": 6.9528,
       "step": 2652
     },
     {
       "epoch": 2.97,
-      "learning_rate": 5.222543352601156e-05,
-      "loss": 6.0761,
+      "learning_rate": 0.00015684971098265893,
+      "loss": 6.9979,
       "step": 2653
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.219653179190752e-05,
-      "loss": 5.9138,
+      "learning_rate": 0.0001567630057803468,
+      "loss": 6.8765,
       "step": 2654
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.2167630057803474e-05,
-      "loss": 5.928,
+      "learning_rate": 0.00015667630057803465,
+      "loss": 6.9419,
       "step": 2655
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.213872832369943e-05,
-      "loss": 5.9865,
+      "learning_rate": 0.00015658959537572255,
+      "loss": 6.9878,
       "step": 2656
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.210982658959538e-05,
-      "loss": 5.9927,
+      "learning_rate": 0.0001565028901734104,
+      "loss": 6.984,
       "step": 2657
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.2080924855491334e-05,
-      "loss": 6.0613,
+      "learning_rate": 0.00015641618497109827,
+      "loss": 6.9977,
       "step": 2658
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.205202312138728e-05,
-      "loss": 5.9722,
+      "learning_rate": 0.00015632947976878611,
+      "loss": 6.8829,
       "step": 2659
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.2023121387283234e-05,
-      "loss": 5.8359,
+      "learning_rate": 0.00015624277456647396,
+      "loss": 6.8103,
       "step": 2660
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.199421965317919e-05,
-      "loss": 5.9861,
+      "learning_rate": 0.00015615606936416184,
+      "loss": 6.9925,
       "step": 2661
     },
     {
       "epoch": 2.98,
-      "learning_rate": 5.1965317919075154e-05,
-      "loss": 6.0891,
+      "learning_rate": 0.00015606936416184968,
+      "loss": 7.1382,
       "step": 2662
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.19364161849711e-05,
-      "loss": 5.8088,
+      "learning_rate": 0.00015598265895953756,
+      "loss": 6.8887,
       "step": 2663
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.1907514450867055e-05,
-      "loss": 5.9291,
+      "learning_rate": 0.00015589595375722543,
+      "loss": 7.0603,
       "step": 2664
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.187861271676301e-05,
-      "loss": 5.5748,
+      "learning_rate": 0.0001558092485549133,
+      "loss": 6.6928,
       "step": 2665
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.184971098265896e-05,
-      "loss": 5.7538,
+      "learning_rate": 0.00015572254335260115,
+      "loss": 6.9123,
       "step": 2666
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.1820809248554915e-05,
-      "loss": 5.9201,
+      "learning_rate": 0.000155635838150289,
+      "loss": 7.0303,
       "step": 2667
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.179190751445087e-05,
-      "loss": 5.8395,
+      "learning_rate": 0.00015554913294797687,
+      "loss": 7.0019,
       "step": 2668
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.1763005780346815e-05,
-      "loss": 5.62,
+      "learning_rate": 0.00015546242774566472,
+      "loss": 6.8906,
       "step": 2669
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.173410404624278e-05,
-      "loss": 5.647,
+      "learning_rate": 0.00015537572254335256,
+      "loss": 6.9562,
       "step": 2670
     },
     {
       "epoch": 2.99,
-      "learning_rate": 5.1705202312138736e-05,
-      "loss": 5.4842,
+      "learning_rate": 0.00015528901734104044,
+      "loss": 6.8661,
       "step": 2671
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.167630057803469e-05,
-      "loss": 5.3492,
+      "learning_rate": 0.0001552023121387283,
+      "loss": 6.77,
       "step": 2672
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.1647398843930636e-05,
-      "loss": 5.3923,
+      "learning_rate": 0.00015511560693641619,
+      "loss": 6.9108,
       "step": 2673
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.161849710982659e-05,
-      "loss": 4.9948,
+      "learning_rate": 0.00015502890173410403,
+      "loss": 6.7264,
       "step": 2674
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.158959537572254e-05,
-      "loss": 4.9453,
+      "learning_rate": 0.0001549421965317919,
+      "loss": 6.8082,
       "step": 2675
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.1560693641618496e-05,
-      "loss": 4.351,
+      "learning_rate": 0.00015485549132947975,
+      "loss": 6.4379,
       "step": 2676
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.153179190751445e-05,
-      "loss": 6.6481,
+      "learning_rate": 0.0001547687861271676,
+      "loss": 7.0282,
       "step": 2677
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.150289017341041e-05,
-      "loss": 6.5852,
+      "learning_rate": 0.00015468208092485547,
+      "loss": 6.9963,
       "step": 2678
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.147398843930636e-05,
-      "loss": 6.6475,
+      "learning_rate": 0.00015459537572254332,
+      "loss": 6.9964,
       "step": 2679
     },
     {
       "epoch": 3.0,
-      "learning_rate": 5.1445086705202317e-05,
-      "loss": 6.5648,
+      "learning_rate": 0.00015450867052023122,
+      "loss": 6.9713,
       "step": 2680
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.141618497109827e-05,
-      "loss": 6.5235,
+      "learning_rate": 0.00015442196531791907,
+      "loss": 7.0208,
       "step": 2681
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.1387283236994223e-05,
-      "loss": 6.3275,
+      "learning_rate": 0.00015433526011560692,
+      "loss": 6.8826,
       "step": 2682
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.135838150289018e-05,
-      "loss": 6.2861,
+      "learning_rate": 0.0001542485549132948,
+      "loss": 6.864,
       "step": 2683
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.1329479768786124e-05,
-      "loss": 6.474,
+      "learning_rate": 0.00015416184971098264,
+      "loss": 6.977,
       "step": 2684
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.130057803468208e-05,
-      "loss": 6.3749,
+      "learning_rate": 0.0001540751445086705,
+      "loss": 6.9834,
       "step": 2685
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.1271676300578044e-05,
-      "loss": 6.3246,
+      "learning_rate": 0.00015398843930635836,
+      "loss": 6.928,
       "step": 2686
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.1242774566474e-05,
-      "loss": 6.2802,
+      "learning_rate": 0.0001539017341040462,
+      "loss": 6.9983,
       "step": 2687
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.1213872832369944e-05,
-      "loss": 6.0807,
+      "learning_rate": 0.0001538150289017341,
+      "loss": 6.8669,
       "step": 2688
     },
     {
       "epoch": 3.01,
-      "learning_rate": 5.11849710982659e-05,
-      "loss": 6.2046,
+      "learning_rate": 0.00015372832369942195,
+      "loss": 6.9261,
       "step": 2689
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.115606936416185e-05,
-      "loss": 6.0992,
+      "learning_rate": 0.00015364161849710983,
+      "loss": 6.9015,
       "step": 2690
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.1127167630057804e-05,
-      "loss": 6.198,
+      "learning_rate": 0.00015355491329479767,
+      "loss": 6.9676,
       "step": 2691
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.109826589595376e-05,
-      "loss": 6.0787,
+      "learning_rate": 0.00015346820809248555,
+      "loss": 7.0024,
       "step": 2692
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.106936416184971e-05,
-      "loss": 6.0601,
+      "learning_rate": 0.0001533815028901734,
+      "loss": 7.0064,
       "step": 2693
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.104046242774567e-05,
-      "loss": 6.0759,
+      "learning_rate": 0.00015329479768786124,
+      "loss": 6.9751,
       "step": 2694
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.1011560693641625e-05,
-      "loss": 6.1768,
+      "learning_rate": 0.00015320809248554911,
+      "loss": 6.9621,
       "step": 2695
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.098265895953758e-05,
-      "loss": 5.9922,
+      "learning_rate": 0.000153121387283237,
+      "loss": 6.8904,
       "step": 2696
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.095375722543353e-05,
-      "loss": 5.9309,
+      "learning_rate": 0.00015303468208092486,
+      "loss": 6.8997,
       "step": 2697
     },
     {
       "epoch": 3.02,
-      "learning_rate": 5.092485549132948e-05,
-      "loss": 5.835,
+      "learning_rate": 0.0001529479768786127,
+      "loss": 6.811,
       "step": 2698
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.089595375722543e-05,
-      "loss": 5.9609,
+      "learning_rate": 0.00015286127167630056,
+      "loss": 6.8951,
       "step": 2699
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.0867052023121385e-05,
-      "loss": 6.0764,
+      "learning_rate": 0.00015277456647398843,
+      "loss": 6.9721,
       "step": 2700
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.083815028901734e-05,
-      "loss": 5.9498,
+      "learning_rate": 0.00015268786127167628,
+      "loss": 6.9518,
       "step": 2701
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.08092485549133e-05,
-      "loss": 5.9088,
+      "learning_rate": 0.00015260115606936415,
+      "loss": 6.9845,
       "step": 2702
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.078034682080925e-05,
-      "loss": 6.1087,
+      "learning_rate": 0.000152514450867052,
+      "loss": 7.0931,
       "step": 2703
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.0751445086705206e-05,
-      "loss": 5.866,
+      "learning_rate": 0.0001524277456647399,
+      "loss": 6.9025,
       "step": 2704
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.072254335260116e-05,
-      "loss": 5.8948,
+      "learning_rate": 0.00015234104046242774,
+      "loss": 6.9882,
       "step": 2705
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.069364161849711e-05,
-      "loss": 5.8705,
+      "learning_rate": 0.0001522543352601156,
+      "loss": 7.0075,
       "step": 2706
     },
     {
       "epoch": 3.03,
-      "learning_rate": 5.0664739884393066e-05,
-      "loss": 5.663,
+      "learning_rate": 0.00015216763005780346,
+      "loss": 6.8593,
       "step": 2707
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.063583815028902e-05,
-      "loss": 5.5898,
+      "learning_rate": 0.0001520809248554913,
+      "loss": 6.8353,
       "step": 2708
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.0606936416184966e-05,
-      "loss": 5.6218,
+      "learning_rate": 0.00015199421965317919,
+      "loss": 6.8885,
       "step": 2709
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.0578034682080933e-05,
-      "loss": 5.7719,
+      "learning_rate": 0.00015190751445086703,
+      "loss": 6.9998,
       "step": 2710
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.054913294797689e-05,
-      "loss": 5.6615,
+      "learning_rate": 0.00015182080924855488,
+      "loss": 6.9102,
       "step": 2711
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.052023121387284e-05,
-      "loss": 5.7796,
+      "learning_rate": 0.00015173410404624278,
+      "loss": 7.0027,
       "step": 2712
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.049132947976879e-05,
-      "loss": 5.747,
+      "learning_rate": 0.00015164739884393063,
+      "loss": 6.9585,
       "step": 2713
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.046242774566474e-05,
-      "loss": 5.6908,
+      "learning_rate": 0.0001515606936416185,
+      "loss": 6.962,
       "step": 2714
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.0433526011560694e-05,
-      "loss": 5.6214,
+      "learning_rate": 0.00015147398843930635,
+      "loss": 6.8508,
       "step": 2715
     },
     {
       "epoch": 3.04,
-      "learning_rate": 5.040462427745665e-05,
-      "loss": 5.764,
+      "learning_rate": 0.0001513872832369942,
+      "loss": 6.9282,
       "step": 2716
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.03757225433526e-05,
-      "loss": 5.6908,
+      "learning_rate": 0.00015130057803468207,
+      "loss": 6.9453,
       "step": 2717
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.034682080924856e-05,
-      "loss": 5.582,
+      "learning_rate": 0.00015121387283236992,
+      "loss": 6.9386,
       "step": 2718
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.0317919075144514e-05,
-      "loss": 5.5728,
+      "learning_rate": 0.0001511271676300578,
+      "loss": 6.9526,
       "step": 2719
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.028901734104047e-05,
-      "loss": 5.6095,
+      "learning_rate": 0.00015104046242774566,
+      "loss": 7.0401,
       "step": 2720
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.026011560693642e-05,
-      "loss": 5.2349,
+      "learning_rate": 0.00015095375722543354,
+      "loss": 6.8054,
       "step": 2721
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.0231213872832375e-05,
-      "loss": 5.352,
+      "learning_rate": 0.00015086705202312138,
+      "loss": 6.9611,
       "step": 2722
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.020231213872832e-05,
-      "loss": 5.208,
+      "learning_rate": 0.00015078034682080923,
+      "loss": 6.88,
       "step": 2723
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.0173410404624275e-05,
-      "loss": 4.9876,
+      "learning_rate": 0.0001506936416184971,
+      "loss": 6.9115,
       "step": 2724
     },
     {
       "epoch": 3.05,
-      "learning_rate": 5.014450867052023e-05,
-      "loss": 4.9968,
+      "learning_rate": 0.00015060693641618495,
+      "loss": 6.9158,
       "step": 2725
     },
     {
       "epoch": 3.06,
-      "learning_rate": 5.0115606936416195e-05,
-      "loss": 4.0196,
+      "learning_rate": 0.0001505202312138728,
+      "loss": 6.4352,
       "step": 2726
     },
     {
       "epoch": 3.06,
-      "learning_rate": 5.008670520231214e-05,
-      "loss": 6.6179,
+      "learning_rate": 0.00015043352601156067,
+      "loss": 6.9066,
       "step": 2727
     },
     {
       "epoch": 3.06,
-      "learning_rate": 5.0057803468208096e-05,
-      "loss": 6.6999,
+      "learning_rate": 0.00015034682080924855,
+      "loss": 6.9606,
       "step": 2728
     },
     {
       "epoch": 3.06,
-      "learning_rate": 5.002890173410405e-05,
-      "loss": 6.4384,
+      "learning_rate": 0.00015026011560693642,
+      "loss": 6.8612,
       "step": 2729
     },
     {
       "epoch": 3.06,
-      "learning_rate": 5e-05,
-      "loss": 6.3645,
+      "learning_rate": 0.00015017341040462427,
+      "loss": 6.8328,
       "step": 2730
     },
     {
       "epoch": 3.06,
-      "learning_rate": 4.9971098265895956e-05,
-      "loss": 6.425,
+      "learning_rate": 0.00015008670520231214,
+      "loss": 6.9504,
       "step": 2731
     },
     {
       "epoch": 3.06,
-      "learning_rate": 4.994219653179191e-05,
-      "loss": 6.3657,
+      "learning_rate": 0.00015,
+      "loss": 6.9595,
       "step": 2732
     },
     {
       "epoch": 3.06,
-      "learning_rate": 4.991329479768786e-05,
-      "loss": 6.4264,
+      "learning_rate": 0.00014991329479768783,
+      "loss": 7.1207,
       "step": 2733
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.9884393063583816e-05,
-      "loss": 6.1868,
+      "learning_rate": 0.0001498265895953757,
+      "loss": 6.8753,
       "step": 2734
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.985549132947977e-05,
-      "loss": 6.3021,
+      "learning_rate": 0.00014973988439306358,
+      "loss": 6.7741,
       "step": 2735
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.982658959537572e-05,
-      "loss": 6.5286,
+      "learning_rate": 0.00014965317919075143,
+      "loss": 6.9421,
       "step": 2736
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.979768786127168e-05,
-      "loss": 6.3646,
+      "learning_rate": 0.0001495664739884393,
+      "loss": 6.8137,
       "step": 2737
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.976878612716763e-05,
-      "loss": 6.259,
+      "learning_rate": 0.00014947976878612715,
+      "loss": 6.8768,
       "step": 2738
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.9739884393063583e-05,
-      "loss": 6.1573,
+      "learning_rate": 0.00014939306358381502,
+      "loss": 7.0011,
       "step": 2739
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.971098265895954e-05,
-      "loss": 5.8613,
+      "learning_rate": 0.00014930635838150287,
+      "loss": 6.868,
       "step": 2740
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.96820809248555e-05,
-      "loss": 6.1575,
+      "learning_rate": 0.00014921965317919074,
+      "loss": 7.0259,
       "step": 2741
     },
     {
       "epoch": 3.07,
-      "learning_rate": 4.965317919075145e-05,
-      "loss": 6.1631,
+      "learning_rate": 0.0001491329479768786,
+      "loss": 7.0126,
       "step": 2742
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.96242774566474e-05,
-      "loss": 6.0242,
+      "learning_rate": 0.00014904624277456646,
+      "loss": 6.8693,
       "step": 2743
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.959537572254335e-05,
-      "loss": 6.0373,
+      "learning_rate": 0.0001489595375722543,
+      "loss": 6.9274,
       "step": 2744
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.956647398843931e-05,
-      "loss": 5.856,
+      "learning_rate": 0.00014887283236994218,
+      "loss": 6.9279,
       "step": 2745
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.9537572254335264e-05,
-      "loss": 6.0481,
+      "learning_rate": 0.00014878612716763003,
+      "loss": 7.0102,
       "step": 2746
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.950867052023122e-05,
-      "loss": 6.0283,
+      "learning_rate": 0.0001486994219653179,
+      "loss": 6.8484,
       "step": 2747
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.9479768786127164e-05,
-      "loss": 6.1561,
+      "learning_rate": 0.00014861271676300578,
+      "loss": 7.01,
       "step": 2748
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.9450867052023125e-05,
-      "loss": 6.0948,
+      "learning_rate": 0.00014852601156069363,
+      "loss": 6.9657,
       "step": 2749
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.942196531791908e-05,
-      "loss": 5.9288,
+      "learning_rate": 0.00014843930635838147,
+      "loss": 6.9734,
       "step": 2750
     },
     {
       "epoch": 3.08,
-      "learning_rate": 4.939306358381503e-05,
-      "loss": 5.9382,
+      "learning_rate": 0.00014835260115606935,
+      "loss": 6.9612,
       "step": 2751
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.9364161849710985e-05,
-      "loss": 5.9306,
+      "learning_rate": 0.00014826589595375722,
+      "loss": 6.9293,
       "step": 2752
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.933526011560694e-05,
-      "loss": 6.0014,
+      "learning_rate": 0.00014817919075144507,
+      "loss": 6.9065,
       "step": 2753
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.930635838150289e-05,
-      "loss": 5.857,
+      "learning_rate": 0.00014809248554913291,
+      "loss": 6.9515,
       "step": 2754
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.9277456647398845e-05,
-      "loss": 5.9719,
+      "learning_rate": 0.0001480057803468208,
+      "loss": 7.0732,
       "step": 2755
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.92485549132948e-05,
-      "loss": 5.8929,
+      "learning_rate": 0.00014791907514450866,
+      "loss": 6.9627,
       "step": 2756
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.921965317919075e-05,
-      "loss": 5.7996,
+      "learning_rate": 0.0001478323699421965,
+      "loss": 6.8877,
       "step": 2757
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.9190751445086706e-05,
-      "loss": 5.8664,
+      "learning_rate": 0.00014774566473988438,
+      "loss": 6.8894,
       "step": 2758
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.916184971098266e-05,
-      "loss": 5.917,
+      "learning_rate": 0.00014765895953757226,
+      "loss": 7.0144,
       "step": 2759
     },
     {
       "epoch": 3.09,
-      "learning_rate": 4.913294797687861e-05,
-      "loss": 5.6727,
+      "learning_rate": 0.0001475722543352601,
+      "loss": 6.8443,
       "step": 2760
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.910404624277457e-05,
-      "loss": 5.6899,
+      "learning_rate": 0.00014748554913294795,
+      "loss": 6.9003,
       "step": 2761
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.9075144508670526e-05,
-      "loss": 5.8207,
+      "learning_rate": 0.00014739884393063582,
+      "loss": 6.8467,
       "step": 2762
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.904624277456647e-05,
-      "loss": 5.9183,
+      "learning_rate": 0.0001473121387283237,
+      "loss": 6.9877,
       "step": 2763
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.9017341040462426e-05,
-      "loss": 5.8383,
+      "learning_rate": 0.00014722543352601154,
+      "loss": 7.0086,
       "step": 2764
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.8988439306358387e-05,
-      "loss": 5.6567,
+      "learning_rate": 0.00014713872832369942,
+      "loss": 6.9606,
       "step": 2765
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.895953757225434e-05,
-      "loss": 5.7764,
+      "learning_rate": 0.00014705202312138727,
+      "loss": 7.0052,
       "step": 2766
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.8930635838150293e-05,
-      "loss": 5.7175,
+      "learning_rate": 0.00014696531791907514,
+      "loss": 7.087,
       "step": 2767
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.890173410404624e-05,
-      "loss": 5.5926,
+      "learning_rate": 0.00014687861271676299,
+      "loss": 7.0095,
       "step": 2768
     },
     {
       "epoch": 3.1,
-      "learning_rate": 4.88728323699422e-05,
-      "loss": 5.4128,
+      "learning_rate": 0.00014679190751445086,
+      "loss": 6.8476,
       "step": 2769
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.8843930635838154e-05,
-      "loss": 5.4554,
+      "learning_rate": 0.0001467052023121387,
+      "loss": 6.9252,
       "step": 2770
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.881502890173411e-05,
-      "loss": 5.1075,
+      "learning_rate": 0.00014661849710982658,
+      "loss": 6.697,
       "step": 2771
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.878612716763006e-05,
-      "loss": 5.1137,
+      "learning_rate": 0.00014653179190751443,
+      "loss": 6.9037,
       "step": 2772
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.8757225433526014e-05,
-      "loss": 5.0511,
+      "learning_rate": 0.0001464450867052023,
+      "loss": 6.7284,
       "step": 2773
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.872832369942197e-05,
-      "loss": 4.8036,
+      "learning_rate": 0.00014635838150289015,
+      "loss": 6.7355,
       "step": 2774
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.869942196531792e-05,
-      "loss": 4.7409,
+      "learning_rate": 0.00014627167630057802,
+      "loss": 6.8232,
       "step": 2775
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.8670520231213874e-05,
-      "loss": 4.6177,
+      "learning_rate": 0.0001461849710982659,
+      "loss": 6.757,
       "step": 2776
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.864161849710983e-05,
-      "loss": 6.5539,
+      "learning_rate": 0.00014609826589595374,
+      "loss": 6.9968,
       "step": 2777
     },
     {
       "epoch": 3.11,
-      "learning_rate": 4.861271676300578e-05,
-      "loss": 6.3955,
+      "learning_rate": 0.0001460115606936416,
+      "loss": 6.8813,
       "step": 2778
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.8583815028901735e-05,
-      "loss": 6.2536,
+      "learning_rate": 0.00014592485549132946,
+      "loss": 6.7813,
       "step": 2779
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.855491329479769e-05,
-      "loss": 6.1885,
+      "learning_rate": 0.00014583815028901734,
+      "loss": 6.8792,
       "step": 2780
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.852601156069365e-05,
-      "loss": 6.1976,
+      "learning_rate": 0.00014575144508670518,
+      "loss": 6.8593,
       "step": 2781
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.8497109826589595e-05,
-      "loss": 6.2908,
+      "learning_rate": 0.00014566473988439306,
+      "loss": 6.9638,
       "step": 2782
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.846820809248555e-05,
-      "loss": 6.2794,
+      "learning_rate": 0.00014557803468208093,
+      "loss": 7.0117,
       "step": 2783
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.84393063583815e-05,
-      "loss": 6.0769,
+      "learning_rate": 0.00014549132947976878,
+      "loss": 6.8429,
       "step": 2784
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.841040462427746e-05,
-      "loss": 6.2239,
+      "learning_rate": 0.00014540462427745663,
+      "loss": 7.0039,
       "step": 2785
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.8381502890173416e-05,
-      "loss": 5.9683,
+      "learning_rate": 0.0001453179190751445,
+      "loss": 6.9044,
       "step": 2786
     },
     {
       "epoch": 3.12,
-      "learning_rate": 4.835260115606937e-05,
-      "loss": 5.8174,
+      "learning_rate": 0.00014523121387283237,
+      "loss": 6.8835,
       "step": 2787
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.8323699421965316e-05,
-      "loss": 6.1419,
+      "learning_rate": 0.00014514450867052022,
+      "loss": 7.0385,
       "step": 2788
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.8294797687861276e-05,
-      "loss": 6.0101,
+      "learning_rate": 0.00014505780346820807,
+      "loss": 6.8975,
       "step": 2789
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.826589595375723e-05,
-      "loss": 6.0222,
+      "learning_rate": 0.00014497109826589594,
+      "loss": 6.909,
       "step": 2790
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.823699421965318e-05,
-      "loss": 6.0236,
+      "learning_rate": 0.00014488439306358381,
+      "loss": 7.0129,
       "step": 2791
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.8208092485549136e-05,
-      "loss": 5.9521,
+      "learning_rate": 0.00014479768786127166,
+      "loss": 6.9189,
       "step": 2792
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.817919075144509e-05,
-      "loss": 6.0391,
+      "learning_rate": 0.00014471098265895954,
+      "loss": 7.0023,
       "step": 2793
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.815028901734104e-05,
-      "loss": 5.8623,
+      "learning_rate": 0.00014462427745664738,
+      "loss": 6.8388,
       "step": 2794
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.8121387283237e-05,
-      "loss": 5.7828,
+      "learning_rate": 0.00014453757225433526,
+      "loss": 6.8623,
       "step": 2795
     },
     {
       "epoch": 3.13,
-      "learning_rate": 4.809248554913295e-05,
-      "loss": 5.7238,
+      "learning_rate": 0.0001444508670520231,
+      "loss": 6.8658,
       "step": 2796
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.8063583815028904e-05,
-      "loss": 5.8274,
+      "learning_rate": 0.00014436416184971098,
+      "loss": 6.9896,
       "step": 2797
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.803468208092486e-05,
-      "loss": 5.9671,
+      "learning_rate": 0.00014427745664739882,
+      "loss": 7.0003,
       "step": 2798
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.800578034682081e-05,
-      "loss": 5.8223,
+      "learning_rate": 0.0001441907514450867,
+      "loss": 6.916,
       "step": 2799
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.7976878612716764e-05,
-      "loss": 5.8614,
+      "learning_rate": 0.00014410404624277454,
+      "loss": 6.9407,
       "step": 2800
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.7947976878612724e-05,
-      "loss": 5.7377,
+      "learning_rate": 0.00014401734104046242,
+      "loss": 6.8335,
       "step": 2801
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.791907514450867e-05,
-      "loss": 5.8664,
+      "learning_rate": 0.00014393063583815026,
+      "loss": 6.915,
       "step": 2802
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.7890173410404624e-05,
-      "loss": 5.8737,
+      "learning_rate": 0.00014384393063583814,
+      "loss": 6.9683,
       "step": 2803
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.786127167630058e-05,
-      "loss": 5.8412,
+      "learning_rate": 0.000143757225433526,
+      "loss": 6.981,
       "step": 2804
     },
     {
       "epoch": 3.14,
-      "learning_rate": 4.783236994219654e-05,
-      "loss": 5.7842,
+      "learning_rate": 0.00014367052023121386,
+      "loss": 6.9051,
       "step": 2805
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.780346820809249e-05,
-      "loss": 5.7199,
+      "learning_rate": 0.0001435838150289017,
+      "loss": 6.9106,
       "step": 2806
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.777456647398844e-05,
-      "loss": 5.5704,
+      "learning_rate": 0.00014349710982658958,
+      "loss": 6.8365,
       "step": 2807
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.774566473988439e-05,
-      "loss": 5.5666,
+      "learning_rate": 0.00014341040462427745,
+      "loss": 6.836,
       "step": 2808
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.771676300578035e-05,
-      "loss": 5.7356,
+      "learning_rate": 0.0001433236994219653,
+      "loss": 6.9493,
       "step": 2809
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.7687861271676305e-05,
-      "loss": 5.7095,
+      "learning_rate": 0.00014323699421965317,
+      "loss": 6.9355,
       "step": 2810
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.765895953757226e-05,
-      "loss": 5.642,
+      "learning_rate": 0.00014315028901734105,
+      "loss": 6.8941,
       "step": 2811
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.763005780346821e-05,
-      "loss": 5.6063,
+      "learning_rate": 0.0001430635838150289,
+      "loss": 6.8831,
       "step": 2812
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.7601156069364166e-05,
-      "loss": 5.4774,
+      "learning_rate": 0.00014297687861271674,
+      "loss": 6.8567,
       "step": 2813
     },
     {
       "epoch": 3.15,
-      "learning_rate": 4.757225433526012e-05,
-      "loss": 5.6387,
+      "learning_rate": 0.00014289017341040462,
+      "loss": 6.9442,
       "step": 2814
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.754335260115607e-05,
-      "loss": 5.6217,
+      "learning_rate": 0.0001428034682080925,
+      "loss": 6.9575,
       "step": 2815
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.7514450867052026e-05,
-      "loss": 5.6762,
+      "learning_rate": 0.00014271676300578034,
+      "loss": 6.9753,
       "step": 2816
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.748554913294798e-05,
-      "loss": 5.6547,
+      "learning_rate": 0.00014263005780346818,
+      "loss": 6.9884,
       "step": 2817
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.745664739884393e-05,
-      "loss": 5.4379,
+      "learning_rate": 0.00014254335260115606,
+      "loss": 6.8946,
       "step": 2818
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.7427745664739886e-05,
-      "loss": 5.4483,
+      "learning_rate": 0.00014245664739884393,
+      "loss": 6.8819,
       "step": 2819
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.739884393063584e-05,
-      "loss": 5.1947,
+      "learning_rate": 0.00014236994219653178,
+      "loss": 6.8281,
       "step": 2820
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.73699421965318e-05,
-      "loss": 5.246,
+      "learning_rate": 0.00014228323699421965,
+      "loss": 6.9682,
       "step": 2821
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.7341040462427747e-05,
-      "loss": 5.1394,
+      "learning_rate": 0.0001421965317919075,
+      "loss": 6.8184,
       "step": 2822
     },
     {
       "epoch": 3.16,
-      "learning_rate": 4.73121387283237e-05,
-      "loss": 5.0579,
+      "learning_rate": 0.00014210982658959537,
+      "loss": 6.8973,
       "step": 2823
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.7283236994219653e-05,
-      "loss": 4.7241,
+      "learning_rate": 0.00014202312138728322,
+      "loss": 6.8016,
       "step": 2824
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.7254335260115614e-05,
-      "loss": 4.7669,
+      "learning_rate": 0.0001419364161849711,
+      "loss": 6.8473,
       "step": 2825
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.722543352601157e-05,
-      "loss": 4.2285,
+      "learning_rate": 0.00014184971098265894,
+      "loss": 6.5775,
       "step": 2826
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.7196531791907514e-05,
-      "loss": 6.6451,
+      "learning_rate": 0.00014176300578034681,
+      "loss": 7.0971,
       "step": 2827
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.716763005780347e-05,
-      "loss": 6.4288,
+      "learning_rate": 0.00014167630057803466,
+      "loss": 6.9419,
       "step": 2828
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.713872832369942e-05,
-      "loss": 6.4752,
+      "learning_rate": 0.00014158959537572253,
+      "loss": 6.958,
       "step": 2829
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.710982658959538e-05,
-      "loss": 6.4793,
+      "learning_rate": 0.00014150289017341038,
+      "loss": 7.0075,
       "step": 2830
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.7080924855491334e-05,
-      "loss": 6.3106,
+      "learning_rate": 0.00014141618497109826,
+      "loss": 6.9904,
       "step": 2831
     },
     {
       "epoch": 3.17,
-      "learning_rate": 4.705202312138728e-05,
-      "loss": 6.3783,
+      "learning_rate": 0.00014132947976878613,
+      "loss": 7.1248,
       "step": 2832
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.7023121387283234e-05,
-      "loss": 6.1974,
+      "learning_rate": 0.00014124277456647398,
+      "loss": 6.9061,
       "step": 2833
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.6994219653179195e-05,
-      "loss": 6.2894,
+      "learning_rate": 0.00014115606936416182,
+      "loss": 7.1333,
       "step": 2834
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.696531791907515e-05,
-      "loss": 5.8744,
+      "learning_rate": 0.0001410693641618497,
+      "loss": 6.8658,
       "step": 2835
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.69364161849711e-05,
-      "loss": 5.8546,
+      "learning_rate": 0.00014098265895953757,
+      "loss": 6.8069,
       "step": 2836
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.690751445086705e-05,
-      "loss": 6.0944,
+      "learning_rate": 0.00014089595375722542,
+      "loss": 6.982,
       "step": 2837
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.687861271676301e-05,
-      "loss": 5.9481,
+      "learning_rate": 0.0001408092485549133,
+      "loss": 6.9443,
       "step": 2838
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.684971098265896e-05,
-      "loss": 6.0504,
+      "learning_rate": 0.00014072254335260114,
+      "loss": 7.0984,
       "step": 2839
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.6820809248554915e-05,
-      "loss": 5.966,
+      "learning_rate": 0.000140635838150289,
+      "loss": 7.0415,
       "step": 2840
     },
     {
       "epoch": 3.18,
-      "learning_rate": 4.679190751445087e-05,
-      "loss": 5.8251,
+      "learning_rate": 0.00014054913294797686,
+      "loss": 6.9672,
       "step": 2841
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.676300578034682e-05,
-      "loss": 5.6489,
+      "learning_rate": 0.00014046242774566473,
+      "loss": 6.8412,
       "step": 2842
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.6734104046242776e-05,
-      "loss": 5.8384,
+      "learning_rate": 0.00014037572254335258,
+      "loss": 6.8889,
       "step": 2843
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.670520231213873e-05,
-      "loss": 5.8543,
+      "learning_rate": 0.00014028901734104045,
+      "loss": 6.8965,
       "step": 2844
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.667630057803468e-05,
-      "loss": 5.8065,
+      "learning_rate": 0.0001402023121387283,
+      "loss": 7.0014,
       "step": 2845
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.664739884393064e-05,
-      "loss": 5.8103,
+      "learning_rate": 0.00014011560693641617,
+      "loss": 6.955,
       "step": 2846
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.661849710982659e-05,
-      "loss": 5.8901,
+      "learning_rate": 0.00014002890173410402,
+      "loss": 6.9948,
       "step": 2847
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.658959537572254e-05,
-      "loss": 5.9106,
+      "learning_rate": 0.0001399421965317919,
+      "loss": 7.1017,
       "step": 2848
     },
     {
       "epoch": 3.19,
-      "learning_rate": 4.6560693641618496e-05,
-      "loss": 5.7428,
+      "learning_rate": 0.00013985549132947977,
+      "loss": 6.9373,
       "step": 2849
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.653179190751446e-05,
-      "loss": 5.6545,
+      "learning_rate": 0.00013976878612716762,
+      "loss": 6.8825,
       "step": 2850
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.650289017341041e-05,
-      "loss": 5.7638,
+      "learning_rate": 0.00013968208092485546,
+      "loss": 6.9294,
       "step": 2851
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.647398843930636e-05,
-      "loss": 5.7432,
+      "learning_rate": 0.00013959537572254334,
+      "loss": 6.9387,
       "step": 2852
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.644508670520231e-05,
-      "loss": 5.6725,
+      "learning_rate": 0.0001395086705202312,
+      "loss": 6.8857,
       "step": 2853
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.641618497109827e-05,
-      "loss": 5.7771,
+      "learning_rate": 0.00013942196531791906,
+      "loss": 6.9948,
       "step": 2854
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.6387283236994224e-05,
-      "loss": 5.6537,
+      "learning_rate": 0.00013933526011560693,
+      "loss": 6.988,
       "step": 2855
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.635838150289018e-05,
-      "loss": 5.699,
+      "learning_rate": 0.00013924855491329478,
+      "loss": 6.9783,
       "step": 2856
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.6329479768786124e-05,
-      "loss": 5.9377,
+      "learning_rate": 0.00013916184971098265,
+      "loss": 7.1108,
       "step": 2857
     },
     {
       "epoch": 3.2,
-      "learning_rate": 4.6300578034682084e-05,
-      "loss": 5.5544,
+      "learning_rate": 0.0001390751445086705,
+      "loss": 6.8565,
       "step": 2858
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.627167630057804e-05,
-      "loss": 5.7176,
+      "learning_rate": 0.00013898843930635837,
+      "loss": 6.9955,
       "step": 2859
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.624277456647399e-05,
-      "loss": 5.6697,
+      "learning_rate": 0.00013890173410404625,
+      "loss": 6.8924,
       "step": 2860
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.6213872832369945e-05,
-      "loss": 5.4618,
+      "learning_rate": 0.0001388150289017341,
+      "loss": 6.7955,
       "step": 2861
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.61849710982659e-05,
-      "loss": 5.8038,
+      "learning_rate": 0.00013872832369942194,
+      "loss": 7.032,
       "step": 2862
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.615606936416185e-05,
-      "loss": 5.5553,
+      "learning_rate": 0.0001386416184971098,
+      "loss": 6.9219,
       "step": 2863
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.6127167630057805e-05,
-      "loss": 5.5247,
+      "learning_rate": 0.0001385549132947977,
+      "loss": 6.8677,
       "step": 2864
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.609826589595376e-05,
-      "loss": 5.3451,
+      "learning_rate": 0.00013846820809248553,
+      "loss": 6.8485,
       "step": 2865
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.606936416184972e-05,
-      "loss": 5.554,
+      "learning_rate": 0.0001383815028901734,
+      "loss": 6.9283,
       "step": 2866
     },
     {
       "epoch": 3.21,
-      "learning_rate": 4.6040462427745665e-05,
-      "loss": 5.4965,
+      "learning_rate": 0.00013829479768786125,
+      "loss": 6.9536,
       "step": 2867
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.601156069364162e-05,
-      "loss": 5.6552,
+      "learning_rate": 0.00013820809248554913,
+      "loss": 7.1133,
       "step": 2868
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.598265895953757e-05,
-      "loss": 5.1673,
+      "learning_rate": 0.00013812138728323698,
+      "loss": 6.8499,
       "step": 2869
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.595375722543353e-05,
-      "loss": 4.9926,
+      "learning_rate": 0.00013803468208092485,
+      "loss": 6.8048,
       "step": 2870
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.5924855491329486e-05,
-      "loss": 4.9543,
+      "learning_rate": 0.0001379479768786127,
+      "loss": 6.7442,
       "step": 2871
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.589595375722543e-05,
-      "loss": 4.8069,
+      "learning_rate": 0.00013786127167630057,
+      "loss": 6.7583,
       "step": 2872
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.5867052023121386e-05,
-      "loss": 4.7033,
+      "learning_rate": 0.00013777456647398842,
+      "loss": 6.7236,
       "step": 2873
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.5838150289017346e-05,
-      "loss": 4.844,
+      "learning_rate": 0.0001376878612716763,
+      "loss": 6.8546,
       "step": 2874
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.58092485549133e-05,
-      "loss": 4.791,
+      "learning_rate": 0.00013760115606936414,
+      "loss": 6.7365,
       "step": 2875
     },
     {
       "epoch": 3.22,
-      "learning_rate": 4.578034682080925e-05,
-      "loss": 4.4269,
+      "learning_rate": 0.000137514450867052,
+      "loss": 6.5237,
       "step": 2876
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.57514450867052e-05,
-      "loss": 6.6561,
+      "learning_rate": 0.00013742774566473989,
+      "loss": 6.9983,
       "step": 2877
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.572254335260116e-05,
-      "loss": 6.6763,
+      "learning_rate": 0.00013734104046242773,
+      "loss": 7.0332,
       "step": 2878
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.569364161849711e-05,
-      "loss": 6.6807,
+      "learning_rate": 0.00013725433526011558,
+      "loss": 7.0691,
       "step": 2879
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.566473988439307e-05,
-      "loss": 6.5243,
+      "learning_rate": 0.00013716763005780345,
+      "loss": 7.0029,
       "step": 2880
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.563583815028902e-05,
-      "loss": 6.2083,
+      "learning_rate": 0.00013708092485549133,
+      "loss": 6.8764,
       "step": 2881
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.5606936416184974e-05,
-      "loss": 6.2445,
+      "learning_rate": 0.00013699421965317917,
+      "loss": 6.9002,
       "step": 2882
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.557803468208093e-05,
-      "loss": 6.2394,
+      "learning_rate": 0.00013690751445086705,
+      "loss": 6.9155,
       "step": 2883
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.554913294797688e-05,
-      "loss": 6.26,
+      "learning_rate": 0.00013682080924855492,
+      "loss": 7.0302,
       "step": 2884
     },
     {
       "epoch": 3.23,
-      "learning_rate": 4.5520231213872834e-05,
-      "loss": 6.1045,
+      "learning_rate": 0.00013673410404624277,
+      "loss": 6.9119,
       "step": 2885
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.549132947976879e-05,
-      "loss": 5.8797,
+      "learning_rate": 0.00013664739884393061,
+      "loss": 6.7963,
       "step": 2886
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.546242774566474e-05,
-      "loss": 5.843,
+      "learning_rate": 0.0001365606936416185,
+      "loss": 6.874,
       "step": 2887
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.5433526011560694e-05,
-      "loss": 6.0325,
+      "learning_rate": 0.00013647398843930636,
+      "loss": 7.0494,
       "step": 2888
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.540462427745665e-05,
-      "loss": 6.0832,
+      "learning_rate": 0.0001363872832369942,
+      "loss": 6.9917,
       "step": 2889
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.537572254335261e-05,
-      "loss": 5.9001,
+      "learning_rate": 0.00013630057803468206,
+      "loss": 6.8582,
       "step": 2890
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.5346820809248555e-05,
-      "loss": 6.0777,
+      "learning_rate": 0.00013621387283236993,
+      "loss": 6.9938,
       "step": 2891
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.531791907514451e-05,
-      "loss": 6.0347,
+      "learning_rate": 0.0001361271676300578,
+      "loss": 7.0459,
       "step": 2892
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.528901734104046e-05,
-      "loss": 5.93,
+      "learning_rate": 0.00013604046242774565,
+      "loss": 6.9634,
       "step": 2893
     },
     {
       "epoch": 3.24,
-      "learning_rate": 4.526011560693642e-05,
-      "loss": 5.8202,
+      "learning_rate": 0.00013595375722543352,
+      "loss": 6.89,
       "step": 2894
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.5231213872832375e-05,
-      "loss": 5.7562,
+      "learning_rate": 0.00013586705202312137,
+      "loss": 6.9461,
       "step": 2895
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.520231213872833e-05,
-      "loss": 5.78,
+      "learning_rate": 0.00013578034682080925,
+      "loss": 6.9557,
       "step": 2896
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.5173410404624275e-05,
-      "loss": 5.9176,
+      "learning_rate": 0.0001356936416184971,
+      "loss": 6.9659,
       "step": 2897
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.5144508670520236e-05,
-      "loss": 5.8159,
+      "learning_rate": 0.00013560693641618497,
+      "loss": 6.9809,
       "step": 2898
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.511560693641619e-05,
-      "loss": 5.6599,
+      "learning_rate": 0.0001355202312138728,
+      "loss": 6.8698,
       "step": 2899
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.508670520231214e-05,
-      "loss": 5.7263,
+      "learning_rate": 0.0001354335260115607,
+      "loss": 6.8836,
       "step": 2900
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.5057803468208096e-05,
-      "loss": 5.6458,
+      "learning_rate": 0.00013534682080924853,
+      "loss": 6.8282,
       "step": 2901
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.502890173410405e-05,
-      "loss": 5.6836,
+      "learning_rate": 0.0001352601156069364,
+      "loss": 6.8879,
       "step": 2902
     },
     {
       "epoch": 3.25,
-      "learning_rate": 4.5e-05,
-      "loss": 5.8466,
+      "learning_rate": 0.00013517341040462425,
+      "loss": 6.9849,
       "step": 2903
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.4971098265895956e-05,
-      "loss": 5.9247,
+      "learning_rate": 0.00013508670520231213,
+      "loss": 6.9059,
       "step": 2904
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.494219653179191e-05,
-      "loss": 5.8677,
+      "learning_rate": 0.000135,
+      "loss": 7.0047,
       "step": 2905
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.491329479768786e-05,
-      "loss": 5.9755,
+      "learning_rate": 0.00013491329479768785,
+      "loss": 7.0898,
       "step": 2906
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.488439306358382e-05,
-      "loss": 5.7746,
+      "learning_rate": 0.0001348265895953757,
+      "loss": 6.967,
       "step": 2907
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.485549132947977e-05,
-      "loss": 5.7397,
+      "learning_rate": 0.00013473988439306357,
+      "loss": 6.926,
       "step": 2908
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.4826589595375724e-05,
-      "loss": 5.7989,
+      "learning_rate": 0.00013465317919075144,
+      "loss": 7.1086,
       "step": 2909
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.4797687861271684e-05,
-      "loss": 5.7788,
+      "learning_rate": 0.0001345664739884393,
+      "loss": 6.9848,
       "step": 2910
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.476878612716763e-05,
-      "loss": 5.5651,
+      "learning_rate": 0.00013447976878612716,
+      "loss": 6.923,
       "step": 2911
     },
     {
       "epoch": 3.26,
-      "learning_rate": 4.4739884393063584e-05,
-      "loss": 5.4928,
+      "learning_rate": 0.00013439306358381504,
+      "loss": 6.8952,
       "step": 2912
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.471098265895954e-05,
-      "loss": 5.6191,
+      "learning_rate": 0.00013430635838150288,
+      "loss": 6.986,
       "step": 2913
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.46820809248555e-05,
-      "loss": 5.5665,
+      "learning_rate": 0.00013421965317919073,
+      "loss": 6.9535,
       "step": 2914
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.465317919075145e-05,
-      "loss": 5.703,
+      "learning_rate": 0.0001341329479768786,
+      "loss": 6.9965,
       "step": 2915
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.46242774566474e-05,
-      "loss": 5.5411,
+      "learning_rate": 0.00013404624277456648,
+      "loss": 6.8511,
       "step": 2916
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.459537572254335e-05,
-      "loss": 5.6207,
+      "learning_rate": 0.00013395953757225433,
+      "loss": 6.937,
       "step": 2917
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.456647398843931e-05,
-      "loss": 5.532,
+      "learning_rate": 0.00013387283236994217,
+      "loss": 6.9593,
       "step": 2918
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.4537572254335265e-05,
-      "loss": 5.7633,
+      "learning_rate": 0.00013378612716763005,
+      "loss": 7.0278,
       "step": 2919
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.450867052023122e-05,
-      "loss": 5.3982,
+      "learning_rate": 0.00013369942196531792,
+      "loss": 6.932,
       "step": 2920
     },
     {
       "epoch": 3.27,
-      "learning_rate": 4.447976878612717e-05,
-      "loss": 5.2476,
+      "learning_rate": 0.00013361271676300577,
+      "loss": 6.8064,
       "step": 2921
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.4450867052023125e-05,
-      "loss": 5.0332,
+      "learning_rate": 0.00013352601156069364,
+      "loss": 6.7694,
       "step": 2922
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.442196531791908e-05,
-      "loss": 5.2644,
+      "learning_rate": 0.0001334393063583815,
+      "loss": 6.9744,
       "step": 2923
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.439306358381503e-05,
-      "loss": 4.4457,
+      "learning_rate": 0.00013335260115606936,
+      "loss": 6.6374,
       "step": 2924
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.4364161849710985e-05,
-      "loss": 4.9607,
+      "learning_rate": 0.0001332658959537572,
+      "loss": 6.8863,
       "step": 2925
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.433526011560694e-05,
-      "loss": 4.5378,
+      "learning_rate": 0.00013317919075144508,
+      "loss": 6.5937,
       "step": 2926
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.430635838150289e-05,
-      "loss": 6.6412,
+      "learning_rate": 0.00013309248554913293,
+      "loss": 7.1967,
       "step": 2927
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.4277456647398846e-05,
-      "loss": 6.45,
+      "learning_rate": 0.00013300578034682078,
+      "loss": 6.9759,
       "step": 2928
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.42485549132948e-05,
-      "loss": 6.3944,
+      "learning_rate": 0.00013291907514450865,
+      "loss": 7.0368,
       "step": 2929
     },
     {
       "epoch": 3.28,
-      "learning_rate": 4.421965317919075e-05,
-      "loss": 6.3235,
+      "learning_rate": 0.00013283236994219652,
+      "loss": 6.9539,
       "step": 2930
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.4190751445086706e-05,
-      "loss": 6.2702,
+      "learning_rate": 0.00013274566473988437,
+      "loss": 6.8646,
       "step": 2931
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.416184971098266e-05,
-      "loss": 6.1707,
+      "learning_rate": 0.00013265895953757224,
+      "loss": 6.9341,
       "step": 2932
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.413294797687861e-05,
-      "loss": 6.1509,
+      "learning_rate": 0.00013257225433526012,
+      "loss": 7.0294,
       "step": 2933
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.4104046242774566e-05,
-      "loss": 6.0791,
+      "learning_rate": 0.00013248554913294797,
+      "loss": 6.9846,
       "step": 2934
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.407514450867053e-05,
-      "loss": 6.0292,
+      "learning_rate": 0.0001323988439306358,
+      "loss": 7.0064,
       "step": 2935
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.404624277456647e-05,
-      "loss": 5.7625,
+      "learning_rate": 0.00013231213872832369,
+      "loss": 6.8472,
       "step": 2936
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.401734104046243e-05,
-      "loss": 5.8354,
+      "learning_rate": 0.00013222543352601156,
+      "loss": 6.9391,
       "step": 2937
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.398843930635838e-05,
-      "loss": 5.9493,
+      "learning_rate": 0.0001321387283236994,
+      "loss": 6.9321,
       "step": 2938
     },
     {
       "epoch": 3.29,
-      "learning_rate": 4.395953757225434e-05,
-      "loss": 5.8485,
+      "learning_rate": 0.00013205202312138728,
+      "loss": 6.9364,
       "step": 2939
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.3930635838150294e-05,
-      "loss": 5.9136,
+      "learning_rate": 0.00013196531791907513,
+      "loss": 6.9587,
       "step": 2940
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.390173410404624e-05,
-      "loss": 6.0035,
+      "learning_rate": 0.000131878612716763,
+      "loss": 7.0678,
       "step": 2941
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.3872832369942194e-05,
-      "loss": 5.9307,
+      "learning_rate": 0.00013179190751445085,
+      "loss": 6.9836,
       "step": 2942
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.3843930635838154e-05,
-      "loss": 5.7465,
+      "learning_rate": 0.00013170520231213872,
+      "loss": 6.9244,
       "step": 2943
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.381502890173411e-05,
-      "loss": 5.789,
+      "learning_rate": 0.00013161849710982657,
+      "loss": 6.9405,
       "step": 2944
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.378612716763006e-05,
-      "loss": 5.855,
+      "learning_rate": 0.00013153179190751444,
+      "loss": 6.994,
       "step": 2945
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.3757225433526015e-05,
-      "loss": 5.787,
+      "learning_rate": 0.0001314450867052023,
+      "loss": 6.9554,
       "step": 2946
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.372832369942197e-05,
-      "loss": 5.6477,
+      "learning_rate": 0.00013135838150289016,
+      "loss": 6.8755,
       "step": 2947
     },
     {
       "epoch": 3.3,
-      "learning_rate": 4.369942196531792e-05,
-      "loss": 5.9618,
+      "learning_rate": 0.000131271676300578,
+      "loss": 7.0571,
       "step": 2948
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.3670520231213875e-05,
-      "loss": 5.6153,
+      "learning_rate": 0.00013118497109826588,
+      "loss": 6.8505,
       "step": 2949
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.364161849710983e-05,
-      "loss": 5.6452,
+      "learning_rate": 0.00013109826589595376,
+      "loss": 6.9012,
       "step": 2950
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.361271676300578e-05,
-      "loss": 5.7472,
+      "learning_rate": 0.0001310115606936416,
+      "loss": 6.9987,
       "step": 2951
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.3583815028901735e-05,
-      "loss": 5.7144,
+      "learning_rate": 0.00013092485549132945,
+      "loss": 6.9566,
       "step": 2952
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.355491329479769e-05,
-      "loss": 5.5399,
+      "learning_rate": 0.00013083815028901733,
+      "loss": 6.8231,
       "step": 2953
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.352601156069364e-05,
-      "loss": 5.6919,
+      "learning_rate": 0.0001307514450867052,
+      "loss": 6.9535,
       "step": 2954
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.34971098265896e-05,
-      "loss": 5.6741,
+      "learning_rate": 0.00013066473988439305,
+      "loss": 6.9986,
       "step": 2955
     },
     {
       "epoch": 3.31,
-      "learning_rate": 4.346820809248555e-05,
-      "loss": 5.7706,
+      "learning_rate": 0.00013057803468208092,
+      "loss": 7.0692,
       "step": 2956
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.34393063583815e-05,
-      "loss": 5.7316,
+      "learning_rate": 0.0001304913294797688,
+      "loss": 7.0032,
       "step": 2957
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.3410404624277456e-05,
-      "loss": 5.684,
+      "learning_rate": 0.00013040462427745664,
+      "loss": 7.0389,
       "step": 2958
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.3381502890173416e-05,
-      "loss": 5.7266,
+      "learning_rate": 0.0001303179190751445,
+      "loss": 6.967,
       "step": 2959
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.335260115606937e-05,
-      "loss": 5.7802,
+      "learning_rate": 0.00013023121387283236,
+      "loss": 6.9638,
       "step": 2960
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.3323699421965316e-05,
-      "loss": 5.7636,
+      "learning_rate": 0.00013014450867052023,
+      "loss": 7.0297,
       "step": 2961
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.329479768786127e-05,
-      "loss": 5.7761,
+      "learning_rate": 0.00013005780346820808,
+      "loss": 7.1418,
       "step": 2962
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.326589595375723e-05,
-      "loss": 5.5738,
+      "learning_rate": 0.00012997109826589593,
+      "loss": 6.9949,
       "step": 2963
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.3236994219653183e-05,
-      "loss": 5.5763,
+      "learning_rate": 0.0001298843930635838,
+      "loss": 7.0003,
       "step": 2964
     },
     {
       "epoch": 3.32,
-      "learning_rate": 4.320809248554914e-05,
-      "loss": 5.5978,
+      "learning_rate": 0.00012979768786127168,
+      "loss": 7.0202,
       "step": 2965
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.3179190751445084e-05,
-      "loss": 5.6589,
+      "learning_rate": 0.00012971098265895952,
+      "loss": 7.0793,
       "step": 2966
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.3150289017341044e-05,
-      "loss": 5.4753,
+      "learning_rate": 0.0001296242774566474,
+      "loss": 6.9747,
       "step": 2967
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.3121387283237e-05,
-      "loss": 5.3502,
+      "learning_rate": 0.00012953757225433524,
+      "loss": 6.894,
       "step": 2968
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.309248554913295e-05,
-      "loss": 5.4888,
+      "learning_rate": 0.00012945086705202312,
+      "loss": 7.0205,
       "step": 2969
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.3063583815028904e-05,
-      "loss": 5.4107,
+      "learning_rate": 0.00012936416184971096,
+      "loss": 6.9975,
       "step": 2970
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.303468208092486e-05,
-      "loss": 5.1268,
+      "learning_rate": 0.00012927745664739884,
+      "loss": 6.8626,
       "step": 2971
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.300578034682081e-05,
-      "loss": 5.1046,
+      "learning_rate": 0.00012919075144508669,
+      "loss": 6.9145,
       "step": 2972
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.2976878612716764e-05,
-      "loss": 4.7382,
+      "learning_rate": 0.00012910404624277456,
+      "loss": 6.7994,
       "step": 2973
     },
     {
       "epoch": 3.33,
-      "learning_rate": 4.294797687861272e-05,
-      "loss": 4.7719,
+      "learning_rate": 0.0001290173410404624,
+      "loss": 6.8309,
       "step": 2974
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.291907514450868e-05,
-      "loss": 4.5536,
+      "learning_rate": 0.00012893063583815028,
+      "loss": 6.8047,
       "step": 2975
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.2890173410404625e-05,
-      "loss": 4.2415,
+      "learning_rate": 0.00012884393063583813,
+      "loss": 6.6702,
       "step": 2976
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.286127167630058e-05,
-      "loss": 6.6672,
+      "learning_rate": 0.000128757225433526,
+      "loss": 7.0755,
       "step": 2977
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.283236994219653e-05,
-      "loss": 6.5042,
+      "learning_rate": 0.00012867052023121387,
+      "loss": 6.9162,
       "step": 2978
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.280346820809249e-05,
-      "loss": 6.4641,
+      "learning_rate": 0.00012858381502890172,
+      "loss": 6.8238,
       "step": 2979
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.2774566473988445e-05,
-      "loss": 6.41,
+      "learning_rate": 0.00012849710982658957,
+      "loss": 6.9409,
       "step": 2980
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.274566473988439e-05,
-      "loss": 6.38,
+      "learning_rate": 0.00012841040462427744,
+      "loss": 6.9548,
       "step": 2981
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.2716763005780345e-05,
-      "loss": 6.2999,
+      "learning_rate": 0.00012832369942196532,
+      "loss": 6.9721,
       "step": 2982
     },
     {
       "epoch": 3.34,
-      "learning_rate": 4.2687861271676306e-05,
-      "loss": 6.1835,
+      "learning_rate": 0.00012823699421965316,
+      "loss": 6.8915,
       "step": 2983
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.265895953757226e-05,
-      "loss": 6.0936,
+      "learning_rate": 0.00012815028901734104,
+      "loss": 6.9119,
       "step": 2984
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.263005780346821e-05,
-      "loss": 6.1504,
+      "learning_rate": 0.0001280635838150289,
+      "loss": 6.956,
       "step": 2985
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.260115606936416e-05,
-      "loss": 5.9654,
+      "learning_rate": 0.00012797687861271676,
+      "loss": 6.8357,
       "step": 2986
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.257225433526012e-05,
-      "loss": 6.0643,
+      "learning_rate": 0.0001278901734104046,
+      "loss": 6.9673,
       "step": 2987
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.254335260115607e-05,
-      "loss": 5.771,
+      "learning_rate": 0.00012780346820809248,
+      "loss": 6.8069,
       "step": 2988
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.2514450867052026e-05,
-      "loss": 6.0689,
+      "learning_rate": 0.00012771676300578035,
+      "loss": 6.9628,
       "step": 2989
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.248554913294798e-05,
-      "loss": 6.0011,
+      "learning_rate": 0.0001276300578034682,
+      "loss": 6.8796,
       "step": 2990
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.245664739884393e-05,
-      "loss": 5.8727,
+      "learning_rate": 0.00012754335260115605,
+      "loss": 6.882,
       "step": 2991
     },
     {
       "epoch": 3.35,
-      "learning_rate": 4.242774566473989e-05,
-      "loss": 6.0036,
+      "learning_rate": 0.00012745664739884392,
+      "loss": 6.9016,
       "step": 2992
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.239884393063584e-05,
-      "loss": 6.0063,
+      "learning_rate": 0.0001273699421965318,
+      "loss": 6.9469,
       "step": 2993
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.2369942196531794e-05,
-      "loss": 5.8758,
+      "learning_rate": 0.00012728323699421964,
+      "loss": 6.9342,
       "step": 2994
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.234104046242775e-05,
-      "loss": 5.7761,
+      "learning_rate": 0.00012719653179190751,
+      "loss": 6.8909,
       "step": 2995
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.23121387283237e-05,
-      "loss": 5.8538,
+      "learning_rate": 0.00012710982658959536,
+      "loss": 7.0184,
       "step": 2996
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.2283236994219654e-05,
-      "loss": 5.7193,
+      "learning_rate": 0.00012702312138728323,
+      "loss": 6.9725,
       "step": 2997
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.225433526011561e-05,
-      "loss": 5.6887,
+      "learning_rate": 0.00012693641618497108,
+      "loss": 6.8989,
       "step": 2998
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.222543352601157e-05,
-      "loss": 5.7343,
+      "learning_rate": 0.00012684971098265895,
+      "loss": 6.8822,
       "step": 2999
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.2196531791907514e-05,
-      "loss": 5.7269,
+      "learning_rate": 0.0001267630057803468,
+      "loss": 6.9503,
       "step": 3000
     },
     {
       "epoch": 3.36,
-      "eval_loss": 6.421473026275635,
-      "eval_runtime": 970.4649,
-      "eval_samples_per_second": 2.722,
-      "eval_steps_per_second": 0.341,
-      "eval_wer": 1.0836756156120404,
+      "eval_loss": 6.947468280792236,
+      "eval_runtime": 965.2852,
+      "eval_samples_per_second": 2.737,
+      "eval_steps_per_second": 0.343,
+      "eval_wer": 2.0097029585094353,
       "step": 3000
     },
     {
       "epoch": 3.36,
-      "learning_rate": 4.216763005780347e-05,
-      "loss": 5.6985,
+      "learning_rate": 0.00012667630057803468,
+      "loss": 6.9005,
       "step": 3001
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.213872832369942e-05,
-      "loss": 5.6211,
+      "learning_rate": 0.00012658959537572252,
+      "loss": 6.9461,
       "step": 3002
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.210982658959538e-05,
-      "loss": 5.7508,
+      "learning_rate": 0.0001265028901734104,
+      "loss": 7.0424,
       "step": 3003
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.2080924855491335e-05,
-      "loss": 5.7948,
+      "learning_rate": 0.00012641618497109824,
+      "loss": 7.0936,
       "step": 3004
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.205202312138729e-05,
-      "loss": 5.6432,
+      "learning_rate": 0.00012632947976878612,
+      "loss": 7.0147,
       "step": 3005
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.2023121387283235e-05,
-      "loss": 5.7515,
+      "learning_rate": 0.000126242774566474,
+      "loss": 6.9976,
       "step": 3006
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.1994219653179195e-05,
-      "loss": 5.4916,
+      "learning_rate": 0.00012615606936416184,
+      "loss": 6.8571,
       "step": 3007
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.196531791907515e-05,
-      "loss": 5.4615,
+      "learning_rate": 0.00012606936416184968,
+      "loss": 6.8319,
       "step": 3008
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.19364161849711e-05,
-      "loss": 5.5424,
+      "learning_rate": 0.00012598265895953756,
+      "loss": 6.8402,
       "step": 3009
     },
     {
       "epoch": 3.37,
-      "learning_rate": 4.1907514450867055e-05,
-      "loss": 5.686,
+      "learning_rate": 0.00012589595375722543,
+      "loss": 6.9879,
       "step": 3010
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.187861271676301e-05,
-      "loss": 5.7514,
+      "learning_rate": 0.00012580924855491328,
+      "loss": 7.0996,
       "step": 3011
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.184971098265896e-05,
-      "loss": 5.6856,
+      "learning_rate": 0.00012572254335260115,
+      "loss": 6.9935,
       "step": 3012
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.1820809248554916e-05,
-      "loss": 5.5668,
+      "learning_rate": 0.00012563583815028903,
+      "loss": 6.9208,
       "step": 3013
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.179190751445087e-05,
-      "loss": 5.5959,
+      "learning_rate": 0.00012554913294797687,
+      "loss": 6.9242,
       "step": 3014
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.176300578034682e-05,
-      "loss": 5.6039,
+      "learning_rate": 0.00012546242774566472,
+      "loss": 6.9404,
       "step": 3015
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.1734104046242776e-05,
-      "loss": 5.6949,
+      "learning_rate": 0.0001253757225433526,
+      "loss": 6.9612,
       "step": 3016
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.170520231213873e-05,
-      "loss": 5.6516,
+      "learning_rate": 0.00012528901734104047,
+      "loss": 7.0741,
       "step": 3017
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.167630057803468e-05,
-      "loss": 5.2591,
+      "learning_rate": 0.00012520231213872831,
+      "loss": 6.7588,
       "step": 3018
     },
     {
       "epoch": 3.38,
-      "learning_rate": 4.1647398843930636e-05,
-      "loss": 5.5663,
+      "learning_rate": 0.00012511560693641616,
+      "loss": 7.1322,
       "step": 3019
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.161849710982659e-05,
-      "loss": 5.2978,
+      "learning_rate": 0.00012502890173410404,
+      "loss": 6.8169,
       "step": 3020
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.1589595375722543e-05,
-      "loss": 5.4186,
+      "learning_rate": 0.00012494219653179188,
+      "loss": 7.0394,
       "step": 3021
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.15606936416185e-05,
-      "loss": 5.317,
+      "learning_rate": 0.00012485549132947976,
+      "loss": 7.0196,
       "step": 3022
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.153179190751445e-05,
-      "loss": 4.7979,
+      "learning_rate": 0.00012476878612716763,
+      "loss": 6.6603,
       "step": 3023
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.150289017341041e-05,
-      "loss": 4.8499,
+      "learning_rate": 0.00012468208092485548,
+      "loss": 6.6194,
       "step": 3024
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.147398843930636e-05,
-      "loss": 4.4216,
+      "learning_rate": 0.00012459537572254332,
+      "loss": 6.5237,
       "step": 3025
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.144508670520231e-05,
-      "loss": 4.1681,
+      "learning_rate": 0.0001245086705202312,
+      "loss": 6.5603,
       "step": 3026
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.1416184971098264e-05,
-      "loss": 6.7429,
+      "learning_rate": 0.00012442196531791907,
+      "loss": 7.1478,
       "step": 3027
     },
     {
       "epoch": 3.39,
-      "learning_rate": 4.1416184971098264e-05,
-      "loss": 6.4915,
+      "learning_rate": 0.00012433526011560692,
+      "loss": 6.8843,
       "step": 3028
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.1387283236994224e-05,
-      "loss": 6.6019,
+      "learning_rate": 0.0001242485549132948,
+      "loss": 7.0229,
       "step": 3029
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.135838150289018e-05,
-      "loss": 6.4919,
+      "learning_rate": 0.00012416184971098267,
+      "loss": 7.0182,
       "step": 3030
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.132947976878613e-05,
-      "loss": 6.2878,
+      "learning_rate": 0.0001240751445086705,
+      "loss": 6.8557,
       "step": 3031
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.130057803468208e-05,
-      "loss": 6.3121,
+      "learning_rate": 0.00012398843930635836,
+      "loss": 6.9341,
       "step": 3032
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.127167630057804e-05,
-      "loss": 6.3966,
+      "learning_rate": 0.00012390173410404623,
+      "loss": 7.0173,
       "step": 3033
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.124277456647399e-05,
-      "loss": 6.2345,
+      "learning_rate": 0.0001238150289017341,
+      "loss": 6.9522,
       "step": 3034
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.1213872832369945e-05,
-      "loss": 6.1229,
+      "learning_rate": 0.00012372832369942195,
+      "loss": 7.0035,
       "step": 3035
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.11849710982659e-05,
-      "loss": 6.2163,
+      "learning_rate": 0.0001236416184971098,
+      "loss": 7.1242,
       "step": 3036
     },
     {
       "epoch": 3.4,
-      "learning_rate": 4.115606936416185e-05,
-      "loss": 5.9513,
+      "learning_rate": 0.00012355491329479767,
+      "loss": 6.9226,
       "step": 3037
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.1127167630057805e-05,
-      "loss": 5.8815,
+      "learning_rate": 0.00012346820809248555,
+      "loss": 6.9203,
       "step": 3038
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.109826589595376e-05,
-      "loss": 5.8337,
+      "learning_rate": 0.0001233815028901734,
+      "loss": 6.9889,
       "step": 3039
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.106936416184971e-05,
-      "loss": 5.9619,
+      "learning_rate": 0.00012329479768786127,
+      "loss": 7.0156,
       "step": 3040
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.1040462427745666e-05,
-      "loss": 5.7834,
+      "learning_rate": 0.00012320809248554912,
+      "loss": 6.8501,
       "step": 3041
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.101156069364162e-05,
-      "loss": 5.7704,
+      "learning_rate": 0.000123121387283237,
+      "loss": 6.8803,
       "step": 3042
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.098265895953757e-05,
-      "loss": 5.9058,
+      "learning_rate": 0.00012303468208092484,
+      "loss": 6.8974,
       "step": 3043
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.0953757225433526e-05,
-      "loss": 5.9243,
+      "learning_rate": 0.0001229479768786127,
+      "loss": 6.9644,
       "step": 3044
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.0924855491329486e-05,
-      "loss": 5.8113,
+      "learning_rate": 0.00012286127167630056,
+      "loss": 6.9029,
       "step": 3045
     },
     {
       "epoch": 3.41,
-      "learning_rate": 4.089595375722543e-05,
-      "loss": 5.8263,
+      "learning_rate": 0.00012277456647398843,
+      "loss": 6.9739,
       "step": 3046
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.0867052023121386e-05,
-      "loss": 5.5549,
+      "learning_rate": 0.00012268786127167628,
+      "loss": 6.7836,
       "step": 3047
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.083815028901734e-05,
-      "loss": 5.6818,
+      "learning_rate": 0.00012260115606936415,
+      "loss": 6.9183,
       "step": 3048
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.08092485549133e-05,
-      "loss": 5.6349,
+      "learning_rate": 0.000122514450867052,
+      "loss": 6.8832,
       "step": 3049
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.0780346820809253e-05,
-      "loss": 5.7643,
+      "learning_rate": 0.00012242774566473987,
+      "loss": 7.0231,
       "step": 3050
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.07514450867052e-05,
-      "loss": 5.7252,
+      "learning_rate": 0.00012234104046242775,
+      "loss": 6.9527,
       "step": 3051
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.0722543352601154e-05,
-      "loss": 5.646,
+      "learning_rate": 0.0001222543352601156,
+      "loss": 6.9096,
       "step": 3052
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.0693641618497114e-05,
-      "loss": 5.7113,
+      "learning_rate": 0.00012216763005780344,
+      "loss": 6.9914,
       "step": 3053
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.066473988439307e-05,
-      "loss": 5.7409,
+      "learning_rate": 0.00012208092485549131,
+      "loss": 7.008,
       "step": 3054
     },
     {
       "epoch": 3.42,
-      "learning_rate": 4.063583815028902e-05,
-      "loss": 5.6796,
+      "learning_rate": 0.00012199421965317919,
+      "loss": 6.9165,
       "step": 3055
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.0606936416184974e-05,
-      "loss": 5.3715,
+      "learning_rate": 0.00012190751445086703,
+      "loss": 6.767,
       "step": 3056
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.057803468208093e-05,
-      "loss": 5.619,
+      "learning_rate": 0.0001218208092485549,
+      "loss": 6.98,
       "step": 3057
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.054913294797688e-05,
-      "loss": 5.6725,
+      "learning_rate": 0.00012173410404624277,
+      "loss": 6.9731,
       "step": 3058
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.0520231213872834e-05,
-      "loss": 5.5262,
+      "learning_rate": 0.00012164739884393063,
+      "loss": 6.837,
       "step": 3059
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.049132947976879e-05,
-      "loss": 5.6084,
+      "learning_rate": 0.00012156069364161849,
+      "loss": 6.9264,
       "step": 3060
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.046242774566474e-05,
-      "loss": 5.9674,
+      "learning_rate": 0.00012147398843930634,
+      "loss": 7.139,
       "step": 3061
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.0433526011560695e-05,
-      "loss": 5.6601,
+      "learning_rate": 0.00012138728323699421,
+      "loss": 6.9647,
       "step": 3062
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.040462427745665e-05,
-      "loss": 5.5244,
+      "learning_rate": 0.00012130057803468207,
+      "loss": 6.8738,
       "step": 3063
     },
     {
       "epoch": 3.43,
-      "learning_rate": 4.03757225433526e-05,
-      "loss": 5.7707,
+      "learning_rate": 0.00012121387283236993,
+      "loss": 7.1349,
       "step": 3064
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.034682080924856e-05,
-      "loss": 5.583,
+      "learning_rate": 0.00012112716763005779,
+      "loss": 7.0233,
       "step": 3065
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.031791907514451e-05,
-      "loss": 5.313,
+      "learning_rate": 0.00012104046242774567,
+      "loss": 6.8303,
       "step": 3066
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.028901734104046e-05,
-      "loss": 5.4511,
+      "learning_rate": 0.00012095375722543351,
+      "loss": 6.9944,
       "step": 3067
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.0260115606936415e-05,
-      "loss": 5.5491,
+      "learning_rate": 0.00012086705202312137,
+      "loss": 7.0502,
       "step": 3068
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.0231213872832376e-05,
-      "loss": 5.6422,
+      "learning_rate": 0.00012078034682080923,
+      "loss": 7.0528,
       "step": 3069
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.020231213872833e-05,
-      "loss": 5.172,
+      "learning_rate": 0.0001206936416184971,
+      "loss": 6.8391,
       "step": 3070
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.0173410404624276e-05,
-      "loss": 5.0659,
+      "learning_rate": 0.00012060693641618497,
+      "loss": 6.8474,
       "step": 3071
     },
     {
       "epoch": 3.44,
-      "learning_rate": 4.014450867052023e-05,
-      "loss": 5.3763,
+      "learning_rate": 0.00012052023121387281,
+      "loss": 6.9472,
       "step": 3072
     },
     {
       "epoch": 3.45,
-      "learning_rate": 4.011560693641619e-05,
-      "loss": 4.8966,
+      "learning_rate": 0.00012043352601156067,
+      "loss": 6.6519,
       "step": 3073
     },
     {
       "epoch": 3.45,
-      "learning_rate": 4.008670520231214e-05,
-      "loss": 4.8614,
+      "learning_rate": 0.00012034682080924855,
+      "loss": 6.8449,
       "step": 3074
     },
     {
       "epoch": 3.45,
-      "learning_rate": 4.0057803468208096e-05,
-      "loss": 4.5573,
+      "learning_rate": 0.00012026011560693641,
+      "loss": 6.7784,
       "step": 3075
     },
     {
       "epoch": 3.45,
-      "learning_rate": 4.002890173410404e-05,
-      "loss": 4.3918,
+      "learning_rate": 0.00012017341040462427,
+      "loss": 6.7007,
       "step": 3076
     },
     {
       "epoch": 3.45,
-      "learning_rate": 4e-05,
-      "loss": 6.5772,
+      "learning_rate": 0.00012008670520231212,
+      "loss": 7.0193,
       "step": 3077
     },
     {
       "epoch": 3.45,
-      "learning_rate": 3.997109826589596e-05,
-      "loss": 6.6043,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 7.0726,
       "step": 3078
     },
     {
       "epoch": 3.45,
-      "learning_rate": 3.994219653179191e-05,
-      "loss": 6.4052,
+      "learning_rate": 0.00011991329479768785,
+      "loss": 7.0296,
       "step": 3079
     },
     {
       "epoch": 3.45,
-      "learning_rate": 3.9913294797687864e-05,
-      "loss": 6.2174,
+      "learning_rate": 0.00011982658959537571,
+      "loss": 6.9463,
       "step": 3080
     },
     {
       "epoch": 3.45,
-      "learning_rate": 3.988439306358382e-05,
-      "loss": 6.0199,
+      "learning_rate": 0.00011973988439306357,
+      "loss": 6.9082,
       "step": 3081
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.985549132947977e-05,
-      "loss": 6.174,
+      "learning_rate": 0.00011965317919075144,
+      "loss": 7.0247,
       "step": 3082
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.9826589595375724e-05,
-      "loss": 6.1906,
+      "learning_rate": 0.0001195664739884393,
+      "loss": 7.0034,
       "step": 3083
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.979768786127168e-05,
-      "loss": 6.2927,
+      "learning_rate": 0.00011947976878612715,
+      "loss": 6.9501,
       "step": 3084
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.976878612716764e-05,
-      "loss": 6.1482,
+      "learning_rate": 0.00011939306358381501,
+      "loss": 6.96,
       "step": 3085
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.9739884393063584e-05,
-      "loss": 5.9731,
+      "learning_rate": 0.00011930635838150289,
+      "loss": 6.9528,
       "step": 3086
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.971098265895954e-05,
-      "loss": 5.7534,
+      "learning_rate": 0.00011921965317919075,
+      "loss": 6.9952,
       "step": 3087
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.968208092485549e-05,
-      "loss": 5.8683,
+      "learning_rate": 0.0001191329479768786,
+      "loss": 6.9835,
       "step": 3088
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.965317919075145e-05,
-      "loss": 5.9972,
+      "learning_rate": 0.00011904624277456645,
+      "loss": 6.9715,
       "step": 3089
     },
     {
       "epoch": 3.46,
-      "learning_rate": 3.9624277456647405e-05,
-      "loss": 5.9717,
+      "learning_rate": 0.00011895953757225433,
+      "loss": 6.9015,
       "step": 3090
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.959537572254335e-05,
-      "loss": 5.9673,
+      "learning_rate": 0.00011887283236994219,
+      "loss": 6.8875,
       "step": 3091
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.9566473988439305e-05,
-      "loss": 5.6842,
+      "learning_rate": 0.00011878612716763005,
+      "loss": 6.8825,
       "step": 3092
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.9537572254335265e-05,
-      "loss": 5.6753,
+      "learning_rate": 0.00011869942196531791,
+      "loss": 6.9739,
       "step": 3093
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.950867052023122e-05,
-      "loss": 5.7621,
+      "learning_rate": 0.00011861271676300578,
+      "loss": 6.9405,
       "step": 3094
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.947976878612717e-05,
-      "loss": 5.6822,
+      "learning_rate": 0.00011852601156069363,
+      "loss": 6.905,
       "step": 3095
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.945086705202312e-05,
-      "loss": 5.7077,
+      "learning_rate": 0.00011843930635838149,
+      "loss": 7.0192,
       "step": 3096
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.942196531791908e-05,
-      "loss": 5.6647,
+      "learning_rate": 0.00011835260115606935,
+      "loss": 7.0277,
       "step": 3097
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.939306358381503e-05,
-      "loss": 5.6768,
+      "learning_rate": 0.00011826589595375722,
+      "loss": 6.9284,
       "step": 3098
     },
     {
       "epoch": 3.47,
-      "learning_rate": 3.9364161849710986e-05,
-      "loss": 5.7197,
+      "learning_rate": 0.00011817919075144508,
+      "loss": 6.9695,
       "step": 3099
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.933526011560694e-05,
-      "loss": 5.5098,
+      "learning_rate": 0.00011809248554913293,
+      "loss": 6.7742,
       "step": 3100
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.930635838150289e-05,
-      "loss": 5.8241,
+      "learning_rate": 0.00011800578034682079,
+      "loss": 7.0001,
       "step": 3101
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.9277456647398846e-05,
-      "loss": 5.5879,
+      "learning_rate": 0.00011791907514450866,
+      "loss": 6.9755,
       "step": 3102
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.92485549132948e-05,
-      "loss": 5.6919,
+      "learning_rate": 0.00011783236994219653,
+      "loss": 6.9757,
       "step": 3103
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.921965317919075e-05,
-      "loss": 5.5993,
+      "learning_rate": 0.00011774566473988439,
+      "loss": 6.9144,
       "step": 3104
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.9190751445086707e-05,
-      "loss": 5.8053,
+      "learning_rate": 0.00011765895953757225,
+      "loss": 7.0846,
       "step": 3105
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.916184971098266e-05,
-      "loss": 5.5036,
+      "learning_rate": 0.00011757225433526012,
+      "loss": 6.9167,
       "step": 3106
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.9132947976878613e-05,
-      "loss": 5.6325,
+      "learning_rate": 0.00011748554913294797,
+      "loss": 6.9963,
       "step": 3107
     },
     {
       "epoch": 3.48,
-      "learning_rate": 3.910404624277457e-05,
-      "loss": 5.4738,
+      "learning_rate": 0.00011739884393063583,
+      "loss": 6.8676,
       "step": 3108
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.907514450867053e-05,
-      "loss": 5.6235,
+      "learning_rate": 0.00011731213872832369,
+      "loss": 6.9873,
       "step": 3109
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.904624277456648e-05,
-      "loss": 5.6064,
+      "learning_rate": 0.00011722543352601156,
+      "loss": 7.0326,
       "step": 3110
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.901734104046243e-05,
-      "loss": 5.4408,
+      "learning_rate": 0.00011713872832369942,
+      "loss": 6.9049,
       "step": 3111
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.898843930635838e-05,
-      "loss": 5.6217,
+      "learning_rate": 0.00011705202312138727,
+      "loss": 6.9515,
       "step": 3112
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.895953757225434e-05,
-      "loss": 5.6117,
+      "learning_rate": 0.00011696531791907513,
+      "loss": 7.0012,
       "step": 3113
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.8930635838150294e-05,
-      "loss": 5.4931,
+      "learning_rate": 0.000116878612716763,
+      "loss": 6.9692,
       "step": 3114
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.890173410404625e-05,
-      "loss": 5.379,
+      "learning_rate": 0.00011679190751445086,
+      "loss": 6.8922,
       "step": 3115
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.8872832369942194e-05,
-      "loss": 5.2758,
+      "learning_rate": 0.00011670520231213872,
+      "loss": 6.7646,
       "step": 3116
     },
     {
       "epoch": 3.49,
-      "learning_rate": 3.884393063583815e-05,
-      "loss": 5.3127,
+      "learning_rate": 0.00011661849710982657,
+      "loss": 7.0114,
       "step": 3117
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.881502890173411e-05,
-      "loss": 5.2369,
+      "learning_rate": 0.00011653179190751443,
+      "loss": 6.9102,
       "step": 3118
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.878612716763006e-05,
-      "loss": 5.2385,
+      "learning_rate": 0.0001164450867052023,
+      "loss": 6.9086,
       "step": 3119
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.8757225433526015e-05,
-      "loss": 5.1275,
+      "learning_rate": 0.00011635838150289016,
+      "loss": 6.8517,
       "step": 3120
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.872832369942196e-05,
-      "loss": 5.0667,
+      "learning_rate": 0.00011627167630057802,
+      "loss": 6.8332,
       "step": 3121
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.869942196531792e-05,
-      "loss": 4.953,
+      "learning_rate": 0.00011618497109826587,
+      "loss": 6.8412,
       "step": 3122
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.8670520231213875e-05,
-      "loss": 5.114,
+      "learning_rate": 0.00011609826589595375,
+      "loss": 7.1077,
       "step": 3123
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.864161849710983e-05,
-      "loss": 4.6366,
+      "learning_rate": 0.0001160115606936416,
+      "loss": 6.7838,
       "step": 3124
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.861271676300578e-05,
-      "loss": 4.2476,
+      "learning_rate": 0.00011592485549132947,
+      "loss": 6.6164,
       "step": 3125
     },
     {
       "epoch": 3.5,
-      "learning_rate": 3.8583815028901736e-05,
-      "loss": 4.515,
+      "learning_rate": 0.00011583815028901733,
+      "loss": 6.8344,
       "step": 3126
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.855491329479769e-05,
-      "loss": 6.7223,
+      "learning_rate": 0.0001157514450867052,
+      "loss": 6.9698,
       "step": 3127
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.852601156069364e-05,
-      "loss": 6.7792,
+      "learning_rate": 0.00011566473988439306,
+      "loss": 7.0065,
       "step": 3128
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.8497109826589596e-05,
-      "loss": 6.6647,
+      "learning_rate": 0.00011557803468208091,
+      "loss": 7.0104,
       "step": 3129
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.846820809248555e-05,
-      "loss": 6.4839,
+      "learning_rate": 0.00011549132947976877,
+      "loss": 6.9161,
       "step": 3130
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.84393063583815e-05,
-      "loss": 6.2928,
+      "learning_rate": 0.00011540462427745664,
+      "loss": 6.8252,
       "step": 3131
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.8410404624277456e-05,
-      "loss": 6.2802,
+      "learning_rate": 0.0001153179190751445,
+      "loss": 6.9242,
       "step": 3132
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.838150289017341e-05,
-      "loss": 6.4586,
+      "learning_rate": 0.00011523121387283236,
+      "loss": 7.0713,
       "step": 3133
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.835260115606937e-05,
-      "loss": 6.3306,
+      "learning_rate": 0.00011514450867052021,
+      "loss": 7.0517,
       "step": 3134
     },
     {
       "epoch": 3.51,
-      "learning_rate": 3.832369942196532e-05,
-      "loss": 6.2626,
+      "learning_rate": 0.00011505780346820808,
+      "loss": 6.9459,
       "step": 3135
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.829479768786127e-05,
-      "loss": 6.1773,
+      "learning_rate": 0.00011497109826589594,
+      "loss": 6.9946,
       "step": 3136
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.8265895953757224e-05,
-      "loss": 6.2767,
+      "learning_rate": 0.0001148843930635838,
+      "loss": 7.0702,
       "step": 3137
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.8236994219653184e-05,
-      "loss": 6.0196,
+      "learning_rate": 0.00011479768786127166,
+      "loss": 6.981,
       "step": 3138
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.820809248554914e-05,
-      "loss": 5.9448,
+      "learning_rate": 0.00011471098265895954,
+      "loss": 6.8765,
       "step": 3139
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.817919075144509e-05,
-      "loss": 5.9634,
+      "learning_rate": 0.00011462427745664738,
+      "loss": 6.9463,
       "step": 3140
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.815028901734104e-05,
-      "loss": 6.032,
+      "learning_rate": 0.00011453757225433525,
+      "loss": 7.0176,
       "step": 3141
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.8121387283237e-05,
-      "loss": 5.9893,
+      "learning_rate": 0.0001144508670520231,
+      "loss": 7.0075,
       "step": 3142
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.809248554913295e-05,
-      "loss": 5.8925,
+      "learning_rate": 0.00011436416184971098,
+      "loss": 6.99,
       "step": 3143
     },
     {
       "epoch": 3.52,
-      "learning_rate": 3.8063583815028905e-05,
-      "loss": 5.8722,
+      "learning_rate": 0.00011427745664739884,
+      "loss": 6.9738,
       "step": 3144
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.803468208092486e-05,
-      "loss": 5.7664,
+      "learning_rate": 0.00011419075144508669,
+      "loss": 6.9598,
       "step": 3145
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.800578034682081e-05,
-      "loss": 5.8708,
+      "learning_rate": 0.00011410404624277455,
+      "loss": 7.0155,
       "step": 3146
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.7976878612716765e-05,
-      "loss": 5.6785,
+      "learning_rate": 0.00011401734104046242,
+      "loss": 6.9124,
       "step": 3147
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.794797687861272e-05,
-      "loss": 5.7291,
+      "learning_rate": 0.00011393063583815028,
+      "loss": 6.9968,
       "step": 3148
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.791907514450867e-05,
-      "loss": 5.673,
+      "learning_rate": 0.00011384393063583814,
+      "loss": 6.9675,
       "step": 3149
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.7890173410404625e-05,
-      "loss": 5.6927,
+      "learning_rate": 0.00011375722543352599,
+      "loss": 6.9215,
       "step": 3150
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.786127167630058e-05,
-      "loss": 5.7181,
+      "learning_rate": 0.00011367052023121386,
+      "loss": 6.9373,
       "step": 3151
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.783236994219653e-05,
-      "loss": 5.7005,
+      "learning_rate": 0.00011358381502890172,
+      "loss": 6.9133,
       "step": 3152
     },
     {
       "epoch": 3.53,
-      "learning_rate": 3.7803468208092486e-05,
-      "loss": 5.6466,
+      "learning_rate": 0.00011349710982658958,
+      "loss": 6.8938,
       "step": 3153
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.7774566473988446e-05,
-      "loss": 5.7038,
+      "learning_rate": 0.00011341040462427744,
+      "loss": 7.0057,
       "step": 3154
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.774566473988439e-05,
-      "loss": 5.7456,
+      "learning_rate": 0.00011332369942196532,
+      "loss": 7.0136,
       "step": 3155
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.7716763005780346e-05,
-      "loss": 5.7579,
+      "learning_rate": 0.00011323699421965318,
+      "loss": 7.0203,
       "step": 3156
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.76878612716763e-05,
-      "loss": 5.6608,
+      "learning_rate": 0.00011315028901734102,
+      "loss": 7.0284,
       "step": 3157
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.765895953757226e-05,
-      "loss": 5.5923,
+      "learning_rate": 0.00011306358381502888,
+      "loss": 6.8901,
       "step": 3158
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.763005780346821e-05,
-      "loss": 5.5962,
+      "learning_rate": 0.00011297687861271676,
+      "loss": 6.8607,
       "step": 3159
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.760115606936416e-05,
-      "loss": 5.6518,
+      "learning_rate": 0.00011289017341040462,
+      "loss": 6.9967,
       "step": 3160
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.757225433526011e-05,
-      "loss": 5.5659,
+      "learning_rate": 0.00011280346820809248,
+      "loss": 7.0163,
       "step": 3161
     },
     {
       "epoch": 3.54,
-      "learning_rate": 3.754335260115607e-05,
-      "loss": 5.4819,
+      "learning_rate": 0.00011271676300578033,
+      "loss": 6.8615,
       "step": 3162
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.751445086705203e-05,
-      "loss": 5.4615,
+      "learning_rate": 0.0001126300578034682,
+      "loss": 6.9525,
       "step": 3163
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.748554913294798e-05,
-      "loss": 5.2943,
+      "learning_rate": 0.00011254335260115606,
+      "loss": 6.8218,
       "step": 3164
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.7456647398843934e-05,
-      "loss": 5.5557,
+      "learning_rate": 0.00011245664739884392,
+      "loss": 7.024,
       "step": 3165
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.742774566473989e-05,
-      "loss": 5.4359,
+      "learning_rate": 0.00011236994219653178,
+      "loss": 6.9657,
       "step": 3166
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.739884393063584e-05,
-      "loss": 5.5582,
+      "learning_rate": 0.00011228323699421965,
+      "loss": 7.0287,
       "step": 3167
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.7369942196531794e-05,
-      "loss": 5.2184,
+      "learning_rate": 0.0001121965317919075,
+      "loss": 6.8304,
       "step": 3168
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.734104046242775e-05,
-      "loss": 5.056,
+      "learning_rate": 0.00011210982658959536,
+      "loss": 6.7944,
       "step": 3169
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.73121387283237e-05,
-      "loss": 5.0324,
+      "learning_rate": 0.00011202312138728322,
+      "loss": 6.7792,
       "step": 3170
     },
     {
       "epoch": 3.55,
-      "learning_rate": 3.7283236994219654e-05,
-      "loss": 5.1016,
+      "learning_rate": 0.0001119364161849711,
+      "loss": 6.9115,
       "step": 3171
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.725433526011561e-05,
-      "loss": 5.1002,
+      "learning_rate": 0.00011184971098265896,
+      "loss": 6.8905,
       "step": 3172
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.722543352601156e-05,
-      "loss": 4.8865,
+      "learning_rate": 0.0001117630057803468,
+      "loss": 6.6739,
       "step": 3173
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.719653179190752e-05,
-      "loss": 4.891,
+      "learning_rate": 0.00011167630057803466,
+      "loss": 6.9134,
       "step": 3174
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.716763005780347e-05,
-      "loss": 4.4016,
+      "learning_rate": 0.00011158959537572254,
+      "loss": 6.7277,
       "step": 3175
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.713872832369942e-05,
-      "loss": 4.1073,
+      "learning_rate": 0.0001115028901734104,
+      "loss": 6.5974,
       "step": 3176
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.7109826589595375e-05,
-      "loss": 6.6271,
+      "learning_rate": 0.00011141618497109826,
+      "loss": 6.8985,
       "step": 3177
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.7080924855491335e-05,
-      "loss": 6.6603,
+      "learning_rate": 0.00011132947976878612,
+      "loss": 6.8676,
       "step": 3178
     },
     {
       "epoch": 3.56,
-      "learning_rate": 3.705202312138729e-05,
-      "loss": 6.5593,
+      "learning_rate": 0.00011124277456647399,
+      "loss": 6.9169,
       "step": 3179
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.7023121387283235e-05,
-      "loss": 6.6417,
+      "learning_rate": 0.00011115606936416184,
+      "loss": 7.0474,
       "step": 3180
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.699421965317919e-05,
-      "loss": 6.6314,
+      "learning_rate": 0.0001110693641618497,
+      "loss": 7.1049,
       "step": 3181
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.696531791907515e-05,
-      "loss": 6.4941,
+      "learning_rate": 0.00011098265895953756,
+      "loss": 6.9687,
       "step": 3182
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.69364161849711e-05,
-      "loss": 6.3348,
+      "learning_rate": 0.00011089595375722543,
+      "loss": 6.9439,
       "step": 3183
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.6907514450867056e-05,
-      "loss": 6.1829,
+      "learning_rate": 0.0001108092485549133,
+      "loss": 6.8232,
       "step": 3184
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.6878612716763e-05,
-      "loss": 6.2995,
+      "learning_rate": 0.00011072254335260114,
+      "loss": 6.899,
       "step": 3185
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.684971098265896e-05,
-      "loss": 6.2576,
+      "learning_rate": 0.000110635838150289,
+      "loss": 6.8791,
       "step": 3186
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.6820809248554916e-05,
-      "loss": 6.2203,
+      "learning_rate": 0.00011054913294797687,
+      "loss": 6.916,
       "step": 3187
     },
     {
       "epoch": 3.57,
-      "learning_rate": 3.679190751445087e-05,
-      "loss": 6.155,
+      "learning_rate": 0.00011046242774566474,
+      "loss": 6.97,
       "step": 3188
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.676300578034682e-05,
-      "loss": 5.8867,
+      "learning_rate": 0.0001103757225433526,
+      "loss": 6.8492,
       "step": 3189
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.6734104046242777e-05,
-      "loss": 5.9032,
+      "learning_rate": 0.00011028901734104044,
+      "loss": 6.925,
       "step": 3190
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.670520231213873e-05,
-      "loss": 5.882,
+      "learning_rate": 0.00011020231213872832,
+      "loss": 6.9563,
       "step": 3191
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.6676300578034683e-05,
-      "loss": 6.0578,
+      "learning_rate": 0.00011011560693641618,
+      "loss": 7.131,
       "step": 3192
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.664739884393064e-05,
-      "loss": 5.9923,
+      "learning_rate": 0.00011002890173410404,
+      "loss": 7.0816,
       "step": 3193
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.66184971098266e-05,
-      "loss": 5.8848,
+      "learning_rate": 0.0001099421965317919,
+      "loss": 6.8995,
       "step": 3194
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.6589595375722544e-05,
-      "loss": 5.8296,
+      "learning_rate": 0.00010985549132947977,
+      "loss": 6.87,
       "step": 3195
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.65606936416185e-05,
-      "loss": 5.7808,
+      "learning_rate": 0.00010976878612716762,
+      "loss": 6.9171,
       "step": 3196
     },
     {
       "epoch": 3.58,
-      "learning_rate": 3.653179190751445e-05,
-      "loss": 5.8621,
+      "learning_rate": 0.00010968208092485548,
+      "loss": 6.9538,
       "step": 3197
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.650289017341041e-05,
-      "loss": 5.8936,
+      "learning_rate": 0.00010959537572254334,
+      "loss": 7.0117,
       "step": 3198
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.6473988439306364e-05,
-      "loss": 5.7174,
+      "learning_rate": 0.00010950867052023121,
+      "loss": 6.8955,
       "step": 3199
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.644508670520231e-05,
-      "loss": 5.8294,
+      "learning_rate": 0.00010942196531791907,
+      "loss": 6.9991,
       "step": 3200
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.6416184971098265e-05,
-      "loss": 5.6788,
+      "learning_rate": 0.00010933526011560692,
+      "loss": 6.9451,
       "step": 3201
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.6387283236994225e-05,
-      "loss": 5.6799,
+      "learning_rate": 0.00010924855491329478,
+      "loss": 6.97,
       "step": 3202
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.635838150289018e-05,
-      "loss": 5.5947,
+      "learning_rate": 0.00010916184971098265,
+      "loss": 6.88,
       "step": 3203
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.632947976878613e-05,
-      "loss": 5.4713,
+      "learning_rate": 0.00010907514450867051,
+      "loss": 6.8592,
       "step": 3204
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.630057803468208e-05,
-      "loss": 5.4671,
+      "learning_rate": 0.00010898843930635837,
+      "loss": 6.8412,
       "step": 3205
     },
     {
       "epoch": 3.59,
-      "learning_rate": 3.627167630057804e-05,
-      "loss": 5.807,
+      "learning_rate": 0.00010890173410404623,
+      "loss": 7.13,
       "step": 3206
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.624277456647399e-05,
-      "loss": 5.7598,
+      "learning_rate": 0.00010881502890173411,
+      "loss": 7.0393,
       "step": 3207
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.6213872832369945e-05,
-      "loss": 5.631,
+      "learning_rate": 0.00010872832369942196,
+      "loss": 6.953,
       "step": 3208
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.61849710982659e-05,
-      "loss": 5.5155,
+      "learning_rate": 0.00010864161849710982,
+      "loss": 6.9189,
       "step": 3209
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.615606936416185e-05,
-      "loss": 5.5665,
+      "learning_rate": 0.00010855491329479768,
+      "loss": 6.9341,
       "step": 3210
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.6127167630057806e-05,
-      "loss": 5.5452,
+      "learning_rate": 0.00010846820809248555,
+      "loss": 6.9102,
       "step": 3211
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.609826589595376e-05,
-      "loss": 5.6003,
+      "learning_rate": 0.00010838150289017341,
+      "loss": 6.9629,
       "step": 3212
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.606936416184971e-05,
-      "loss": 5.393,
+      "learning_rate": 0.00010829479768786126,
+      "loss": 6.835,
       "step": 3213
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.6040462427745666e-05,
-      "loss": 5.4751,
+      "learning_rate": 0.00010820809248554912,
+      "loss": 6.9173,
       "step": 3214
     },
     {
       "epoch": 3.6,
-      "learning_rate": 3.601156069364162e-05,
-      "loss": 5.4122,
+      "learning_rate": 0.00010812138728323698,
+      "loss": 6.9131,
       "step": 3215
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.598265895953757e-05,
-      "loss": 5.6729,
+      "learning_rate": 0.00010803468208092485,
+      "loss": 7.161,
       "step": 3216
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.5953757225433526e-05,
-      "loss": 5.2523,
+      "learning_rate": 0.00010794797687861271,
+      "loss": 6.9009,
       "step": 3217
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.592485549132948e-05,
-      "loss": 5.0913,
+      "learning_rate": 0.00010786127167630056,
+      "loss": 6.7834,
       "step": 3218
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.589595375722544e-05,
-      "loss": 5.3576,
+      "learning_rate": 0.00010777456647398842,
+      "loss": 6.9907,
       "step": 3219
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.586705202312139e-05,
-      "loss": 5.1211,
+      "learning_rate": 0.00010768786127167629,
+      "loss": 6.82,
       "step": 3220
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.583815028901734e-05,
-      "loss": 5.2233,
+      "learning_rate": 0.00010760115606936415,
+      "loss": 6.9597,
       "step": 3221
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.5809248554913294e-05,
-      "loss": 4.9888,
+      "learning_rate": 0.00010751445086705201,
+      "loss": 6.8769,
       "step": 3222
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.5780346820809254e-05,
-      "loss": 4.8,
+      "learning_rate": 0.00010742774566473986,
+      "loss": 6.7884,
       "step": 3223
     },
     {
       "epoch": 3.61,
-      "learning_rate": 3.575144508670521e-05,
-      "loss": 4.7608,
+      "learning_rate": 0.00010734104046242773,
+      "loss": 6.8133,
       "step": 3224
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.5722543352601154e-05,
-      "loss": 4.7674,
+      "learning_rate": 0.0001072543352601156,
+      "loss": 6.9896,
       "step": 3225
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.569364161849711e-05,
-      "loss": 4.1745,
+      "learning_rate": 0.00010716763005780346,
+      "loss": 6.427,
       "step": 3226
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.566473988439307e-05,
-      "loss": 6.8219,
+      "learning_rate": 0.00010708092485549132,
+      "loss": 6.9661,
       "step": 3227
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.563583815028902e-05,
-      "loss": 6.9519,
+      "learning_rate": 0.00010699421965317919,
+      "loss": 7.1771,
       "step": 3228
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.5606936416184975e-05,
-      "loss": 6.7231,
+      "learning_rate": 0.00010690751445086705,
+      "loss": 7.017,
       "step": 3229
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.557803468208092e-05,
-      "loss": 6.4331,
+      "learning_rate": 0.0001068208092485549,
+      "loss": 6.8201,
       "step": 3230
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.554913294797688e-05,
-      "loss": 6.4903,
+      "learning_rate": 0.00010673410404624276,
+      "loss": 6.9195,
       "step": 3231
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.5520231213872835e-05,
-      "loss": 6.5446,
+      "learning_rate": 0.00010664739884393063,
+      "loss": 7.0103,
       "step": 3232
     },
     {
       "epoch": 3.62,
-      "learning_rate": 3.549132947976879e-05,
-      "loss": 6.4909,
+      "learning_rate": 0.00010656069364161849,
+      "loss": 6.9796,
       "step": 3233
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.546242774566474e-05,
-      "loss": 6.286,
+      "learning_rate": 0.00010647398843930635,
+      "loss": 6.9488,
       "step": 3234
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.5433526011560695e-05,
-      "loss": 6.4256,
+      "learning_rate": 0.0001063872832369942,
+      "loss": 7.0202,
       "step": 3235
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.540462427745665e-05,
-      "loss": 6.3037,
+      "learning_rate": 0.00010630057803468207,
+      "loss": 7.0162,
       "step": 3236
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.53757225433526e-05,
-      "loss": 6.2871,
+      "learning_rate": 0.00010621387283236993,
+      "loss": 6.9998,
       "step": 3237
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.5346820809248556e-05,
-      "loss": 6.3198,
+      "learning_rate": 0.00010612716763005779,
+      "loss": 7.0726,
       "step": 3238
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.531791907514451e-05,
-      "loss": 6.1607,
+      "learning_rate": 0.00010604046242774565,
+      "loss": 6.9949,
       "step": 3239
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.528901734104046e-05,
-      "loss": 6.039,
+      "learning_rate": 0.00010595375722543353,
+      "loss": 6.8791,
       "step": 3240
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.5260115606936416e-05,
-      "loss": 6.0312,
+      "learning_rate": 0.00010586705202312137,
+      "loss": 6.955,
       "step": 3241
     },
     {
       "epoch": 3.63,
-      "learning_rate": 3.523121387283237e-05,
-      "loss": 5.9622,
+      "learning_rate": 0.00010578034682080923,
+      "loss": 6.9895,
       "step": 3242
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.520231213872833e-05,
-      "loss": 5.8609,
+      "learning_rate": 0.0001056936416184971,
+      "loss": 6.8816,
       "step": 3243
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.5173410404624276e-05,
-      "loss": 5.8697,
+      "learning_rate": 0.00010560693641618497,
+      "loss": 6.8876,
       "step": 3244
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.514450867052023e-05,
-      "loss": 5.8515,
+      "learning_rate": 0.00010552023121387283,
+      "loss": 6.8485,
       "step": 3245
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.511560693641618e-05,
-      "loss": 5.8284,
+      "learning_rate": 0.00010543352601156068,
+      "loss": 6.9635,
       "step": 3246
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.508670520231214e-05,
-      "loss": 5.8684,
+      "learning_rate": 0.00010534682080924854,
+      "loss": 6.9855,
       "step": 3247
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.50578034682081e-05,
-      "loss": 5.8444,
+      "learning_rate": 0.00010526011560693641,
+      "loss": 6.8915,
       "step": 3248
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.502890173410405e-05,
-      "loss": 6.067,
+      "learning_rate": 0.00010517341040462427,
+      "loss": 7.0414,
       "step": 3249
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.5e-05,
-      "loss": 5.953,
+      "learning_rate": 0.00010508670520231213,
+      "loss": 7.1033,
       "step": 3250
     },
     {
       "epoch": 3.64,
-      "learning_rate": 3.497109826589596e-05,
-      "loss": 5.8511,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 6.9511,
       "step": 3251
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.494219653179191e-05,
-      "loss": 5.759,
+      "learning_rate": 0.00010491329479768786,
+      "loss": 6.9852,
       "step": 3252
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.4913294797687864e-05,
-      "loss": 5.7168,
+      "learning_rate": 0.00010482658959537571,
+      "loss": 6.8764,
       "step": 3253
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.488439306358382e-05,
-      "loss": 5.6958,
+      "learning_rate": 0.00010473988439306357,
+      "loss": 6.9875,
       "step": 3254
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.485549132947977e-05,
-      "loss": 5.8839,
+      "learning_rate": 0.00010465317919075143,
+      "loss": 7.1382,
       "step": 3255
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.4826589595375724e-05,
-      "loss": 5.5239,
+      "learning_rate": 0.0001045664739884393,
+      "loss": 6.8815,
       "step": 3256
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.479768786127168e-05,
-      "loss": 5.7605,
+      "learning_rate": 0.00010447976878612717,
+      "loss": 7.0884,
       "step": 3257
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.476878612716763e-05,
-      "loss": 5.6363,
+      "learning_rate": 0.00010439306358381501,
+      "loss": 6.9276,
       "step": 3258
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.4739884393063585e-05,
-      "loss": 5.6338,
+      "learning_rate": 0.00010430635838150287,
+      "loss": 7.023,
       "step": 3259
     },
     {
       "epoch": 3.65,
-      "learning_rate": 3.471098265895954e-05,
-      "loss": 5.6076,
+      "learning_rate": 0.00010421965317919075,
+      "loss": 6.9778,
       "step": 3260
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.468208092485549e-05,
-      "loss": 5.7088,
+      "learning_rate": 0.00010413294797687861,
+      "loss": 7.0239,
       "step": 3261
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.4653179190751445e-05,
-      "loss": 5.4379,
+      "learning_rate": 0.00010404624277456647,
+      "loss": 6.8194,
       "step": 3262
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.4624277456647405e-05,
-      "loss": 5.4409,
+      "learning_rate": 0.00010395953757225431,
+      "loss": 6.9444,
       "step": 3263
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.459537572254335e-05,
-      "loss": 5.509,
+      "learning_rate": 0.00010387283236994219,
+      "loss": 7.0104,
       "step": 3264
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.4566473988439305e-05,
-      "loss": 5.3936,
+      "learning_rate": 0.00010378612716763005,
+      "loss": 6.959,
       "step": 3265
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.453757225433526e-05,
-      "loss": 5.2955,
+      "learning_rate": 0.00010369942196531791,
+      "loss": 6.8627,
       "step": 3266
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.450867052023122e-05,
-      "loss": 5.5096,
+      "learning_rate": 0.00010361271676300577,
+      "loss": 7.1101,
       "step": 3267
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.447976878612717e-05,
-      "loss": 4.9549,
+      "learning_rate": 0.00010352601156069364,
+      "loss": 6.6313,
       "step": 3268
     },
     {
       "epoch": 3.66,
-      "learning_rate": 3.445086705202312e-05,
-      "loss": 5.2643,
+      "learning_rate": 0.00010343930635838149,
+      "loss": 6.966,
       "step": 3269
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.442196531791907e-05,
-      "loss": 5.3609,
+      "learning_rate": 0.00010335260115606935,
+      "loss": 6.9615,
       "step": 3270
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.439306358381503e-05,
-      "loss": 5.2851,
+      "learning_rate": 0.00010326589595375721,
+      "loss": 6.9997,
       "step": 3271
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.4364161849710986e-05,
-      "loss": 4.8627,
+      "learning_rate": 0.00010317919075144509,
+      "loss": 6.7831,
       "step": 3272
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.433526011560694e-05,
-      "loss": 4.829,
+      "learning_rate": 0.00010309248554913295,
+      "loss": 6.8398,
       "step": 3273
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.430635838150289e-05,
-      "loss": 4.7314,
+      "learning_rate": 0.00010300578034682079,
+      "loss": 6.8111,
       "step": 3274
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.427745664739885e-05,
-      "loss": 4.7056,
+      "learning_rate": 0.00010291907514450865,
+      "loss": 6.8788,
       "step": 3275
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.42485549132948e-05,
-      "loss": 4.2564,
+      "learning_rate": 0.00010283236994219653,
+      "loss": 6.5684,
       "step": 3276
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.4219653179190754e-05,
-      "loss": 6.8315,
+      "learning_rate": 0.00010274566473988439,
+      "loss": 7.2001,
       "step": 3277
     },
     {
       "epoch": 3.67,
-      "learning_rate": 3.419075144508671e-05,
-      "loss": 6.3619,
+      "learning_rate": 0.00010265895953757225,
+      "loss": 6.842,
       "step": 3278
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.416184971098266e-05,
-      "loss": 6.4135,
+      "learning_rate": 0.00010257225433526011,
+      "loss": 6.8556,
       "step": 3279
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.4132947976878614e-05,
-      "loss": 6.4295,
+      "learning_rate": 0.00010248554913294798,
+      "loss": 6.9969,
       "step": 3280
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.410404624277457e-05,
-      "loss": 6.25,
+      "learning_rate": 0.00010239884393063583,
+      "loss": 6.9653,
       "step": 3281
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.407514450867052e-05,
-      "loss": 6.1448,
+      "learning_rate": 0.00010231213872832369,
+      "loss": 6.8303,
       "step": 3282
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.404624277456648e-05,
-      "loss": 6.2086,
+      "learning_rate": 0.00010222543352601155,
+      "loss": 7.0245,
       "step": 3283
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.401734104046243e-05,
-      "loss": 6.169,
+      "learning_rate": 0.00010213872832369942,
+      "loss": 6.9837,
       "step": 3284
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.398843930635838e-05,
-      "loss": 6.0334,
+      "learning_rate": 0.00010205202312138728,
+      "loss": 6.8857,
       "step": 3285
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.3959537572254335e-05,
-      "loss": 6.0659,
+      "learning_rate": 0.00010196531791907513,
+      "loss": 6.9276,
       "step": 3286
     },
     {
       "epoch": 3.68,
-      "learning_rate": 3.3930635838150295e-05,
-      "loss": 6.0309,
+      "learning_rate": 0.00010187861271676299,
+      "loss": 6.9852,
       "step": 3287
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.390173410404625e-05,
-      "loss": 5.8093,
+      "learning_rate": 0.00010179190751445086,
+      "loss": 6.8388,
       "step": 3288
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.3872832369942195e-05,
-      "loss": 5.9169,
+      "learning_rate": 0.00010170520231213872,
+      "loss": 6.9746,
       "step": 3289
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.384393063583815e-05,
-      "loss": 5.7167,
+      "learning_rate": 0.00010161849710982658,
+      "loss": 6.9131,
       "step": 3290
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.381502890173411e-05,
-      "loss": 5.6016,
+      "learning_rate": 0.00010153179190751443,
+      "loss": 6.786,
       "step": 3291
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.378612716763006e-05,
-      "loss": 5.8648,
+      "learning_rate": 0.0001014450867052023,
+      "loss": 7.0481,
       "step": 3292
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.3757225433526015e-05,
-      "loss": 5.7461,
+      "learning_rate": 0.00010135838150289017,
+      "loss": 6.9661,
       "step": 3293
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.372832369942196e-05,
-      "loss": 5.6823,
+      "learning_rate": 0.00010127167630057803,
+      "loss": 6.8567,
       "step": 3294
     },
     {
       "epoch": 3.69,
-      "learning_rate": 3.369942196531792e-05,
-      "loss": 5.7597,
+      "learning_rate": 0.00010118497109826589,
+      "loss": 6.9619,
       "step": 3295
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.3670520231213876e-05,
-      "loss": 5.6318,
+      "learning_rate": 0.00010109826589595376,
+      "loss": 6.885,
       "step": 3296
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.364161849710983e-05,
-      "loss": 5.6345,
+      "learning_rate": 0.00010101156069364161,
+      "loss": 6.8713,
       "step": 3297
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.361271676300578e-05,
-      "loss": 5.7597,
+      "learning_rate": 0.00010092485549132947,
+      "loss": 7.0405,
       "step": 3298
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.3583815028901736e-05,
-      "loss": 5.7098,
+      "learning_rate": 0.00010083815028901733,
+      "loss": 6.9577,
       "step": 3299
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.355491329479769e-05,
-      "loss": 5.6952,
+      "learning_rate": 0.0001007514450867052,
+      "loss": 6.9791,
       "step": 3300
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.352601156069364e-05,
-      "loss": 5.6173,
+      "learning_rate": 0.00010066473988439306,
+      "loss": 6.9296,
       "step": 3301
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.3497109826589596e-05,
-      "loss": 5.6827,
+      "learning_rate": 0.00010057803468208092,
+      "loss": 6.9456,
       "step": 3302
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.346820809248556e-05,
-      "loss": 5.8103,
+      "learning_rate": 0.00010049132947976877,
+      "loss": 7.1295,
       "step": 3303
     },
     {
       "epoch": 3.7,
-      "learning_rate": 3.34393063583815e-05,
-      "loss": 5.6391,
+      "learning_rate": 0.00010040462427745664,
+      "loss": 6.9778,
       "step": 3304
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.341040462427746e-05,
-      "loss": 5.6507,
+      "learning_rate": 0.0001003179190751445,
+      "loss": 7.0144,
       "step": 3305
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.338150289017341e-05,
-      "loss": 5.4575,
+      "learning_rate": 0.00010023121387283236,
+      "loss": 6.9138,
       "step": 3306
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.335260115606937e-05,
-      "loss": 5.5514,
+      "learning_rate": 0.00010014450867052022,
+      "loss": 6.9912,
       "step": 3307
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.3323699421965324e-05,
-      "loss": 5.5702,
+      "learning_rate": 0.0001000578034682081,
+      "loss": 6.9467,
       "step": 3308
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.329479768786127e-05,
-      "loss": 5.3092,
+      "learning_rate": 9.997109826589594e-05,
+      "loss": 6.7897,
       "step": 3309
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.3265895953757224e-05,
-      "loss": 5.378,
+      "learning_rate": 9.98843930635838e-05,
+      "loss": 6.8087,
       "step": 3310
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.323699421965318e-05,
-      "loss": 5.5337,
+      "learning_rate": 9.979768786127167e-05,
+      "loss": 6.9187,
       "step": 3311
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.320809248554914e-05,
-      "loss": 5.4207,
+      "learning_rate": 9.971098265895953e-05,
+      "loss": 6.9466,
       "step": 3312
     },
     {
       "epoch": 3.71,
-      "learning_rate": 3.317919075144509e-05,
-      "loss": 5.5454,
+      "learning_rate": 9.96242774566474e-05,
+      "loss": 6.9526,
       "step": 3313
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.315028901734104e-05,
-      "loss": 5.4107,
+      "learning_rate": 9.953757225433525e-05,
+      "loss": 6.8968,
       "step": 3314
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.312138728323699e-05,
-      "loss": 5.3955,
+      "learning_rate": 9.94508670520231e-05,
+      "loss": 6.9531,
       "step": 3315
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.309248554913295e-05,
-      "loss": 5.5455,
+      "learning_rate": 9.936416184971097e-05,
+      "loss": 6.9886,
       "step": 3316
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.3063583815028905e-05,
-      "loss": 5.3798,
+      "learning_rate": 9.927745664739884e-05,
+      "loss": 6.8414,
       "step": 3317
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.303468208092486e-05,
-      "loss": 5.5366,
+      "learning_rate": 9.91907514450867e-05,
+      "loss": 7.0102,
       "step": 3318
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.3005780346820805e-05,
-      "loss": 5.5776,
+      "learning_rate": 9.910404624277455e-05,
+      "loss": 7.0216,
       "step": 3319
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.2976878612716765e-05,
-      "loss": 5.3759,
+      "learning_rate": 9.901734104046241e-05,
+      "loss": 6.9103,
       "step": 3320
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.294797687861272e-05,
-      "loss": 4.973,
+      "learning_rate": 9.893063583815028e-05,
+      "loss": 6.8419,
       "step": 3321
     },
     {
       "epoch": 3.72,
-      "learning_rate": 3.291907514450867e-05,
-      "loss": 4.8999,
+      "learning_rate": 9.884393063583814e-05,
+      "loss": 6.8704,
       "step": 3322
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.2890173410404626e-05,
-      "loss": 4.5715,
+      "learning_rate": 9.8757225433526e-05,
+      "loss": 6.6997,
       "step": 3323
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.286127167630058e-05,
-      "loss": 5.1184,
+      "learning_rate": 9.867052023121385e-05,
+      "loss": 7.0903,
       "step": 3324
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.283236994219653e-05,
-      "loss": 4.5693,
+      "learning_rate": 9.858381502890172e-05,
+      "loss": 6.663,
       "step": 3325
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.2803468208092486e-05,
-      "loss": 3.9867,
+      "learning_rate": 9.849710982658958e-05,
+      "loss": 6.4811,
       "step": 3326
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.277456647398844e-05,
-      "loss": 6.6003,
+      "learning_rate": 9.841040462427744e-05,
+      "loss": 6.9222,
       "step": 3327
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.27456647398844e-05,
-      "loss": 6.45,
+      "learning_rate": 9.83236994219653e-05,
+      "loss": 6.7625,
       "step": 3328
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.2716763005780346e-05,
-      "loss": 6.6662,
+      "learning_rate": 9.823699421965318e-05,
+      "loss": 7.1315,
       "step": 3329
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.26878612716763e-05,
-      "loss": 6.5463,
+      "learning_rate": 9.815028901734104e-05,
+      "loss": 7.0418,
       "step": 3330
     },
     {
       "epoch": 3.73,
-      "learning_rate": 3.265895953757225e-05,
-      "loss": 6.3725,
+      "learning_rate": 9.806358381502889e-05,
+      "loss": 6.9667,
       "step": 3331
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.2630057803468213e-05,
-      "loss": 6.4034,
+      "learning_rate": 9.797687861271675e-05,
+      "loss": 7.0393,
       "step": 3332
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.260115606936417e-05,
-      "loss": 6.2748,
+      "learning_rate": 9.789017341040462e-05,
+      "loss": 7.0275,
       "step": 3333
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.2572254335260114e-05,
-      "loss": 6.206,
+      "learning_rate": 9.780346820809248e-05,
+      "loss": 7.0061,
       "step": 3334
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.254335260115607e-05,
-      "loss": 6.1969,
+      "learning_rate": 9.771676300578034e-05,
+      "loss": 7.0042,
       "step": 3335
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.251445086705203e-05,
-      "loss": 6.1778,
+      "learning_rate": 9.763005780346819e-05,
+      "loss": 6.9564,
       "step": 3336
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.248554913294798e-05,
-      "loss": 6.2494,
+      "learning_rate": 9.754335260115606e-05,
+      "loss": 6.9374,
       "step": 3337
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.2456647398843934e-05,
-      "loss": 6.1107,
+      "learning_rate": 9.745664739884392e-05,
+      "loss": 6.9126,
       "step": 3338
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.242774566473988e-05,
-      "loss": 6.2284,
+      "learning_rate": 9.736994219653178e-05,
+      "loss": 7.0371,
       "step": 3339
     },
     {
       "epoch": 3.74,
-      "learning_rate": 3.239884393063584e-05,
-      "loss": 5.9643,
+      "learning_rate": 9.728323699421964e-05,
+      "loss": 6.9576,
       "step": 3340
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.2369942196531794e-05,
-      "loss": 5.9079,
+      "learning_rate": 9.719653179190752e-05,
+      "loss": 7.0964,
       "step": 3341
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.234104046242775e-05,
-      "loss": 5.6741,
+      "learning_rate": 9.710982658959536e-05,
+      "loss": 6.8088,
       "step": 3342
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.23121387283237e-05,
-      "loss": 5.8351,
+      "learning_rate": 9.702312138728322e-05,
+      "loss": 6.9754,
       "step": 3343
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.2283236994219655e-05,
-      "loss": 5.7366,
+      "learning_rate": 9.693641618497108e-05,
+      "loss": 6.8401,
       "step": 3344
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.225433526011561e-05,
-      "loss": 5.8936,
+      "learning_rate": 9.684971098265896e-05,
+      "loss": 7.0174,
       "step": 3345
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.222543352601156e-05,
-      "loss": 5.6282,
+      "learning_rate": 9.676300578034682e-05,
+      "loss": 6.8884,
       "step": 3346
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.2196531791907515e-05,
-      "loss": 5.681,
+      "learning_rate": 9.667630057803466e-05,
+      "loss": 6.9402,
       "step": 3347
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.216763005780347e-05,
-      "loss": 5.6737,
+      "learning_rate": 9.658959537572253e-05,
+      "loss": 6.8461,
       "step": 3348
     },
     {
       "epoch": 3.75,
-      "learning_rate": 3.213872832369942e-05,
-      "loss": 5.679,
+      "learning_rate": 9.65028901734104e-05,
+      "loss": 6.8284,
       "step": 3349
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.2109826589595375e-05,
-      "loss": 5.5979,
+      "learning_rate": 9.641618497109826e-05,
+      "loss": 6.8668,
       "step": 3350
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.208092485549133e-05,
-      "loss": 5.6729,
+      "learning_rate": 9.632947976878612e-05,
+      "loss": 6.94,
       "step": 3351
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.205202312138729e-05,
-      "loss": 5.7746,
+      "learning_rate": 9.624277456647398e-05,
+      "loss": 7.1159,
       "step": 3352
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.202312138728324e-05,
-      "loss": 5.6829,
+      "learning_rate": 9.615606936416185e-05,
+      "loss": 6.9423,
       "step": 3353
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.199421965317919e-05,
-      "loss": 5.7522,
+      "learning_rate": 9.60693641618497e-05,
+      "loss": 7.0743,
       "step": 3354
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.196531791907514e-05,
-      "loss": 5.6528,
+      "learning_rate": 9.598265895953756e-05,
+      "loss": 7.0513,
       "step": 3355
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.19364161849711e-05,
-      "loss": 5.4941,
+      "learning_rate": 9.589595375722542e-05,
+      "loss": 6.8931,
       "step": 3356
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.1907514450867056e-05,
-      "loss": 5.7363,
+      "learning_rate": 9.58092485549133e-05,
+      "loss": 7.0865,
       "step": 3357
     },
     {
       "epoch": 3.76,
-      "learning_rate": 3.187861271676301e-05,
-      "loss": 5.4951,
+      "learning_rate": 9.572254335260116e-05,
+      "loss": 6.8853,
       "step": 3358
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.1849710982658956e-05,
-      "loss": 5.4796,
+      "learning_rate": 9.5635838150289e-05,
+      "loss": 6.9566,
       "step": 3359
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.182080924855492e-05,
-      "loss": 5.5092,
+      "learning_rate": 9.554913294797686e-05,
+      "loss": 6.9063,
       "step": 3360
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.179190751445087e-05,
-      "loss": 5.5353,
+      "learning_rate": 9.546242774566474e-05,
+      "loss": 6.9501,
       "step": 3361
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.1763005780346824e-05,
-      "loss": 5.4579,
+      "learning_rate": 9.53757225433526e-05,
+      "loss": 6.916,
       "step": 3362
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.173410404624278e-05,
-      "loss": 5.3075,
+      "learning_rate": 9.528901734104046e-05,
+      "loss": 6.8705,
       "step": 3363
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.170520231213873e-05,
-      "loss": 5.4789,
+      "learning_rate": 9.52023121387283e-05,
+      "loss": 6.9339,
       "step": 3364
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.1676300578034684e-05,
-      "loss": 5.5613,
+      "learning_rate": 9.511560693641618e-05,
+      "loss": 7.0632,
       "step": 3365
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.164739884393064e-05,
-      "loss": 5.3011,
+      "learning_rate": 9.502890173410404e-05,
+      "loss": 6.9097,
       "step": 3366
     },
     {
       "epoch": 3.77,
-      "learning_rate": 3.161849710982659e-05,
-      "loss": 5.4345,
+      "learning_rate": 9.49421965317919e-05,
+      "loss": 7.0208,
       "step": 3367
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.1589595375722544e-05,
-      "loss": 5.4541,
+      "learning_rate": 9.485549132947976e-05,
+      "loss": 7.0523,
       "step": 3368
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.15606936416185e-05,
-      "loss": 5.4113,
+      "learning_rate": 9.476878612716763e-05,
+      "loss": 7.0082,
       "step": 3369
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.153179190751445e-05,
-      "loss": 5.1738,
+      "learning_rate": 9.468208092485548e-05,
+      "loss": 6.9431,
       "step": 3370
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.1502890173410405e-05,
-      "loss": 4.8345,
+      "learning_rate": 9.459537572254334e-05,
+      "loss": 6.8363,
       "step": 3371
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.1473988439306365e-05,
-      "loss": 4.4258,
+      "learning_rate": 9.45086705202312e-05,
+      "loss": 6.5576,
       "step": 3372
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.144508670520231e-05,
-      "loss": 4.7768,
+      "learning_rate": 9.442196531791907e-05,
+      "loss": 6.8418,
       "step": 3373
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.1416184971098265e-05,
-      "loss": 4.4195,
+      "learning_rate": 9.433526011560693e-05,
+      "loss": 6.7799,
       "step": 3374
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.138728323699422e-05,
-      "loss": 4.5741,
+      "learning_rate": 9.42485549132948e-05,
+      "loss": 6.7234,
       "step": 3375
     },
     {
       "epoch": 3.78,
-      "learning_rate": 3.135838150289018e-05,
-      "loss": 4.1956,
+      "learning_rate": 9.416184971098264e-05,
+      "loss": 6.4303,
       "step": 3376
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.132947976878613e-05,
-      "loss": 6.5653,
+      "learning_rate": 9.407514450867052e-05,
+      "loss": 6.9178,
       "step": 3377
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.130057803468208e-05,
-      "loss": 6.6174,
+      "learning_rate": 9.398843930635838e-05,
+      "loss": 7.0763,
       "step": 3378
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.127167630057803e-05,
-      "loss": 6.5778,
+      "learning_rate": 9.390173410404624e-05,
+      "loss": 7.0493,
       "step": 3379
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.124277456647399e-05,
-      "loss": 6.6901,
+      "learning_rate": 9.38150289017341e-05,
+      "loss": 7.092,
       "step": 3380
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.1213872832369946e-05,
-      "loss": 6.4539,
+      "learning_rate": 9.372832369942197e-05,
+      "loss": 6.9851,
       "step": 3381
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.11849710982659e-05,
-      "loss": 6.385,
+      "learning_rate": 9.364161849710982e-05,
+      "loss": 6.9888,
       "step": 3382
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.115606936416185e-05,
-      "loss": 6.4684,
+      "learning_rate": 9.355491329479768e-05,
+      "loss": 7.1392,
       "step": 3383
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.1127167630057806e-05,
-      "loss": 6.2235,
+      "learning_rate": 9.346820809248554e-05,
+      "loss": 6.9728,
       "step": 3384
     },
     {
       "epoch": 3.79,
-      "learning_rate": 3.109826589595376e-05,
-      "loss": 6.3449,
+      "learning_rate": 9.338150289017341e-05,
+      "loss": 7.0665,
       "step": 3385
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.106936416184971e-05,
-      "loss": 6.1041,
+      "learning_rate": 9.329479768786127e-05,
+      "loss": 6.9288,
       "step": 3386
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.1040462427745667e-05,
-      "loss": 5.9429,
+      "learning_rate": 9.320809248554912e-05,
+      "loss": 6.8693,
       "step": 3387
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.101156069364162e-05,
-      "loss": 6.0175,
+      "learning_rate": 9.312138728323698e-05,
+      "loss": 6.9395,
       "step": 3388
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.0982658959537573e-05,
-      "loss": 5.7044,
+      "learning_rate": 9.303468208092485e-05,
+      "loss": 6.8104,
       "step": 3389
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.095375722543353e-05,
-      "loss": 5.7466,
+      "learning_rate": 9.294797687861271e-05,
+      "loss": 6.9048,
       "step": 3390
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.092485549132948e-05,
-      "loss": 5.902,
+      "learning_rate": 9.286127167630057e-05,
+      "loss": 7.0385,
       "step": 3391
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.089595375722544e-05,
-      "loss": 5.8159,
+      "learning_rate": 9.277456647398842e-05,
+      "loss": 6.9855,
       "step": 3392
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.086705202312139e-05,
-      "loss": 5.6838,
+      "learning_rate": 9.26878612716763e-05,
+      "loss": 6.8512,
       "step": 3393
     },
     {
       "epoch": 3.8,
-      "learning_rate": 3.083815028901734e-05,
-      "loss": 5.5947,
+      "learning_rate": 9.260115606936415e-05,
+      "loss": 6.7844,
       "step": 3394
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.0809248554913294e-05,
-      "loss": 5.7871,
+      "learning_rate": 9.251445086705202e-05,
+      "loss": 6.9159,
       "step": 3395
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.0780346820809254e-05,
-      "loss": 5.9015,
+      "learning_rate": 9.242774566473988e-05,
+      "loss": 7.0346,
       "step": 3396
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.075144508670521e-05,
-      "loss": 5.7582,
+      "learning_rate": 9.234104046242775e-05,
+      "loss": 6.986,
       "step": 3397
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.0722543352601154e-05,
-      "loss": 5.5872,
+      "learning_rate": 9.22543352601156e-05,
+      "loss": 6.9128,
       "step": 3398
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.069364161849711e-05,
-      "loss": 5.6391,
+      "learning_rate": 9.216763005780346e-05,
+      "loss": 6.9302,
       "step": 3399
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.066473988439307e-05,
-      "loss": 5.5829,
+      "learning_rate": 9.208092485549132e-05,
+      "loss": 6.9213,
       "step": 3400
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.063583815028902e-05,
-      "loss": 5.6197,
+      "learning_rate": 9.199421965317919e-05,
+      "loss": 6.9838,
       "step": 3401
     },
     {
       "epoch": 3.81,
-      "learning_rate": 3.0606936416184975e-05,
-      "loss": 5.4253,
+      "learning_rate": 9.190751445086705e-05,
+      "loss": 6.8188,
       "step": 3402
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.057803468208092e-05,
-      "loss": 5.5996,
+      "learning_rate": 9.182080924855491e-05,
+      "loss": 6.9628,
       "step": 3403
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.0549132947976875e-05,
-      "loss": 5.5713,
+      "learning_rate": 9.173410404624276e-05,
+      "loss": 6.8594,
       "step": 3404
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.0520231213872835e-05,
-      "loss": 5.4223,
+      "learning_rate": 9.164739884393062e-05,
+      "loss": 6.8248,
       "step": 3405
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.049132947976879e-05,
-      "loss": 5.5381,
+      "learning_rate": 9.156069364161849e-05,
+      "loss": 6.9311,
       "step": 3406
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.046242774566474e-05,
-      "loss": 5.5878,
+      "learning_rate": 9.147398843930635e-05,
+      "loss": 6.9892,
       "step": 3407
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.0433526011560692e-05,
-      "loss": 5.4942,
+      "learning_rate": 9.138728323699421e-05,
+      "loss": 6.9237,
       "step": 3408
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.040462427745665e-05,
-      "loss": 5.6171,
+      "learning_rate": 9.130057803468206e-05,
+      "loss": 7.0265,
       "step": 3409
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.0375722543352603e-05,
-      "loss": 5.5272,
+      "learning_rate": 9.121387283236993e-05,
+      "loss": 7.0232,
       "step": 3410
     },
     {
       "epoch": 3.82,
-      "learning_rate": 3.0346820809248556e-05,
-      "loss": 5.169,
+      "learning_rate": 9.11271676300578e-05,
+      "loss": 6.8047,
       "step": 3411
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.0317919075144506e-05,
-      "loss": 5.3816,
+      "learning_rate": 9.104046242774565e-05,
+      "loss": 6.8933,
       "step": 3412
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.0289017341040466e-05,
-      "loss": 5.6123,
+      "learning_rate": 9.095375722543351e-05,
+      "loss": 7.1024,
       "step": 3413
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.0260115606936416e-05,
-      "loss": 5.4941,
+      "learning_rate": 9.086705202312139e-05,
+      "loss": 7.0577,
       "step": 3414
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.023121387283237e-05,
-      "loss": 5.2038,
+      "learning_rate": 9.078034682080924e-05,
+      "loss": 6.8031,
       "step": 3415
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.0202312138728323e-05,
-      "loss": 5.2672,
+      "learning_rate": 9.06936416184971e-05,
+      "loss": 6.863,
       "step": 3416
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.017341040462428e-05,
-      "loss": 5.1894,
+      "learning_rate": 9.060693641618496e-05,
+      "loss": 6.9563,
       "step": 3417
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.0144508670520234e-05,
-      "loss": 5.11,
+      "learning_rate": 9.052023121387283e-05,
+      "loss": 6.857,
       "step": 3418
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.0115606936416184e-05,
-      "loss": 4.9257,
+      "learning_rate": 9.043352601156069e-05,
+      "loss": 6.8371,
       "step": 3419
     },
     {
       "epoch": 3.83,
-      "learning_rate": 3.0086705202312137e-05,
-      "loss": 5.3022,
+      "learning_rate": 9.034682080924854e-05,
+      "loss": 7.0227,
       "step": 3420
     },
     {
       "epoch": 3.84,
-      "learning_rate": 3.0057803468208097e-05,
-      "loss": 5.0088,
+      "learning_rate": 9.02601156069364e-05,
+      "loss": 6.8487,
       "step": 3421
     },
     {
       "epoch": 3.84,
-      "learning_rate": 3.0028901734104047e-05,
-      "loss": 5.0952,
+      "learning_rate": 9.017341040462427e-05,
+      "loss": 7.0208,
       "step": 3422
     },
     {
       "epoch": 3.84,
-      "learning_rate": 3e-05,
-      "loss": 4.7782,
+      "learning_rate": 9.008670520231213e-05,
+      "loss": 6.8391,
       "step": 3423
     },
     {
       "epoch": 3.84,
-      "learning_rate": 2.9971098265895954e-05,
-      "loss": 4.9202,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 7.0022,
       "step": 3424
     },
     {
       "epoch": 3.84,
-      "learning_rate": 2.994219653179191e-05,
-      "loss": 4.4308,
+      "learning_rate": 8.991329479768785e-05,
+      "loss": 6.7108,
       "step": 3425
     },
     {
       "epoch": 3.84,
-      "learning_rate": 2.9913294797687864e-05,
-      "loss": 4.4415,
+      "learning_rate": 8.982658959537573e-05,
+      "loss": 6.8226,
       "step": 3426
     },
     {
       "epoch": 3.84,
-      "learning_rate": 2.9884393063583815e-05,
-      "loss": 6.7225,
+      "learning_rate": 8.973988439306357e-05,
+      "loss": 6.9525,
       "step": 3427
     },
     {
       "epoch": 3.84,
-      "learning_rate": 2.9855491329479768e-05,
-      "loss": 6.5528,
+      "learning_rate": 8.965317919075143e-05,
+      "loss": 6.8547,
       "step": 3428
     },
     {
       "epoch": 3.84,
-      "learning_rate": 2.9826589595375725e-05,
-      "loss": 6.4853,
+      "learning_rate": 8.95664739884393e-05,
+      "loss": 6.8432,
       "step": 3429
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.9797687861271678e-05,
-      "loss": 6.5298,
+      "learning_rate": 8.947976878612717e-05,
+      "loss": 6.9912,
       "step": 3430
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.9768786127167632e-05,
-      "loss": 6.2862,
+      "learning_rate": 8.939306358381503e-05,
+      "loss": 6.8989,
       "step": 3431
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.9739884393063582e-05,
-      "loss": 6.3852,
+      "learning_rate": 8.930635838150287e-05,
+      "loss": 6.9709,
       "step": 3432
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.9710982658959542e-05,
-      "loss": 6.2827,
+      "learning_rate": 8.921965317919074e-05,
+      "loss": 6.9067,
       "step": 3433
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.9682080924855492e-05,
-      "loss": 6.3544,
+      "learning_rate": 8.913294797687861e-05,
+      "loss": 7.0683,
       "step": 3434
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.9653179190751446e-05,
-      "loss": 6.237,
+      "learning_rate": 8.904624277456647e-05,
+      "loss": 6.9033,
       "step": 3435
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.96242774566474e-05,
-      "loss": 6.0076,
+      "learning_rate": 8.895953757225433e-05,
+      "loss": 6.8125,
       "step": 3436
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.9595375722543356e-05,
-      "loss": 5.9789,
+      "learning_rate": 8.887283236994218e-05,
+      "loss": 6.8962,
       "step": 3437
     },
     {
       "epoch": 3.85,
-      "learning_rate": 2.956647398843931e-05,
-      "loss": 5.9731,
+      "learning_rate": 8.878612716763005e-05,
+      "loss": 6.8555,
       "step": 3438
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.953757225433526e-05,
-      "loss": 6.0193,
+      "learning_rate": 8.869942196531791e-05,
+      "loss": 7.004,
       "step": 3439
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.9508670520231213e-05,
-      "loss": 5.8731,
+      "learning_rate": 8.861271676300577e-05,
+      "loss": 6.9943,
       "step": 3440
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.947976878612717e-05,
-      "loss": 5.744,
+      "learning_rate": 8.852601156069363e-05,
+      "loss": 6.8712,
       "step": 3441
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.9450867052023123e-05,
-      "loss": 5.6733,
+      "learning_rate": 8.84393063583815e-05,
+      "loss": 6.8801,
       "step": 3442
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.9421965317919076e-05,
-      "loss": 5.7053,
+      "learning_rate": 8.835260115606935e-05,
+      "loss": 6.9215,
       "step": 3443
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.9393063583815027e-05,
-      "loss": 5.8118,
+      "learning_rate": 8.826589595375721e-05,
+      "loss": 6.9709,
       "step": 3444
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.9364161849710987e-05,
-      "loss": 5.5926,
+      "learning_rate": 8.817919075144507e-05,
+      "loss": 6.8661,
       "step": 3445
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.9335260115606937e-05,
-      "loss": 5.7114,
+      "learning_rate": 8.809248554913295e-05,
+      "loss": 6.8988,
       "step": 3446
     },
     {
       "epoch": 3.86,
-      "learning_rate": 2.930635838150289e-05,
-      "loss": 5.8406,
+      "learning_rate": 8.800578034682081e-05,
+      "loss": 7.0801,
       "step": 3447
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.9277456647398844e-05,
-      "loss": 5.5935,
+      "learning_rate": 8.791907514450865e-05,
+      "loss": 6.8862,
       "step": 3448
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.92485549132948e-05,
-      "loss": 5.6831,
+      "learning_rate": 8.783236994219651e-05,
+      "loss": 6.963,
       "step": 3449
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.9219653179190754e-05,
-      "loss": 5.5316,
+      "learning_rate": 8.774566473988439e-05,
+      "loss": 6.9185,
       "step": 3450
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.9190751445086707e-05,
-      "loss": 5.5391,
+      "learning_rate": 8.765895953757225e-05,
+      "loss": 6.9145,
       "step": 3451
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.9161849710982657e-05,
-      "loss": 5.5363,
+      "learning_rate": 8.757225433526011e-05,
+      "loss": 6.8994,
       "step": 3452
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.9132947976878618e-05,
-      "loss": 5.4976,
+      "learning_rate": 8.748554913294797e-05,
+      "loss": 6.9425,
       "step": 3453
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.9104046242774568e-05,
-      "loss": 5.4457,
+      "learning_rate": 8.739884393063584e-05,
+      "loss": 6.854,
       "step": 3454
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.907514450867052e-05,
-      "loss": 5.4826,
+      "learning_rate": 8.731213872832369e-05,
+      "loss": 6.8721,
       "step": 3455
     },
     {
       "epoch": 3.87,
-      "learning_rate": 2.9046242774566475e-05,
-      "loss": 5.6601,
+      "learning_rate": 8.722543352601155e-05,
+      "loss": 6.9971,
       "step": 3456
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.901734104046243e-05,
-      "loss": 5.5088,
+      "learning_rate": 8.713872832369941e-05,
+      "loss": 6.9073,
       "step": 3457
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.8988439306358385e-05,
-      "loss": 5.4895,
+      "learning_rate": 8.705202312138728e-05,
+      "loss": 6.9095,
       "step": 3458
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.8959537572254335e-05,
-      "loss": 5.4042,
+      "learning_rate": 8.696531791907514e-05,
+      "loss": 6.9277,
       "step": 3459
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.893063583815029e-05,
-      "loss": 5.5039,
+      "learning_rate": 8.687861271676299e-05,
+      "loss": 6.9733,
       "step": 3460
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.8901734104046245e-05,
-      "loss": 5.5691,
+      "learning_rate": 8.679190751445085e-05,
+      "loss": 6.917,
       "step": 3461
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.88728323699422e-05,
-      "loss": 5.4174,
+      "learning_rate": 8.670520231213873e-05,
+      "loss": 6.824,
       "step": 3462
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.8843930635838152e-05,
-      "loss": 5.6211,
+      "learning_rate": 8.661849710982659e-05,
+      "loss": 7.1416,
       "step": 3463
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.8815028901734102e-05,
-      "loss": 5.4022,
+      "learning_rate": 8.653179190751445e-05,
+      "loss": 6.9202,
       "step": 3464
     },
     {
       "epoch": 3.88,
-      "learning_rate": 2.8786127167630062e-05,
-      "loss": 5.2764,
+      "learning_rate": 8.644508670520229e-05,
+      "loss": 6.9463,
       "step": 3465
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.8757225433526013e-05,
-      "loss": 5.3453,
+      "learning_rate": 8.635838150289017e-05,
+      "loss": 6.9054,
       "step": 3466
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.8728323699421966e-05,
-      "loss": 5.4133,
+      "learning_rate": 8.627167630057803e-05,
+      "loss": 6.8859,
       "step": 3467
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.869942196531792e-05,
-      "loss": 5.3223,
+      "learning_rate": 8.618497109826589e-05,
+      "loss": 6.9419,
       "step": 3468
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.8670520231213876e-05,
-      "loss": 4.8857,
+      "learning_rate": 8.609826589595375e-05,
+      "loss": 6.7233,
       "step": 3469
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.864161849710983e-05,
-      "loss": 4.9402,
+      "learning_rate": 8.601156069364162e-05,
+      "loss": 6.8229,
       "step": 3470
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.861271676300578e-05,
-      "loss": 5.2155,
+      "learning_rate": 8.592485549132947e-05,
+      "loss": 6.7774,
       "step": 3471
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.8583815028901733e-05,
-      "loss": 5.0858,
+      "learning_rate": 8.583815028901733e-05,
+      "loss": 6.6833,
       "step": 3472
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.855491329479769e-05,
-      "loss": 5.0552,
+      "learning_rate": 8.575144508670519e-05,
+      "loss": 6.8494,
       "step": 3473
     },
     {
       "epoch": 3.89,
-      "learning_rate": 2.8526011560693643e-05,
-      "loss": 4.9235,
+      "learning_rate": 8.566473988439306e-05,
+      "loss": 6.8986,
       "step": 3474
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.8497109826589597e-05,
-      "loss": 4.2173,
+      "learning_rate": 8.557803468208092e-05,
+      "loss": 6.5254,
       "step": 3475
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.846820809248555e-05,
-      "loss": 3.9945,
+      "learning_rate": 8.549132947976878e-05,
+      "loss": 6.4351,
       "step": 3476
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.8439306358381507e-05,
-      "loss": 6.807,
+      "learning_rate": 8.540462427745663e-05,
+      "loss": 6.988,
       "step": 3477
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.841040462427746e-05,
-      "loss": 6.7285,
+      "learning_rate": 8.53179190751445e-05,
+      "loss": 6.9573,
       "step": 3478
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.838150289017341e-05,
-      "loss": 6.6942,
+      "learning_rate": 8.523121387283236e-05,
+      "loss": 6.9875,
       "step": 3479
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.8352601156069364e-05,
-      "loss": 6.7693,
+      "learning_rate": 8.514450867052023e-05,
+      "loss": 6.9868,
       "step": 3480
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.832369942196532e-05,
-      "loss": 6.617,
+      "learning_rate": 8.505780346820809e-05,
+      "loss": 6.9032,
       "step": 3481
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.8294797687861274e-05,
-      "loss": 6.4547,
+      "learning_rate": 8.497109826589596e-05,
+      "loss": 6.7965,
       "step": 3482
     },
     {
       "epoch": 3.9,
-      "learning_rate": 2.8265895953757228e-05,
-      "loss": 6.4444,
+      "learning_rate": 8.48843930635838e-05,
+      "loss": 6.8511,
       "step": 3483
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.8236994219653178e-05,
-      "loss": 6.6003,
+      "learning_rate": 8.479768786127167e-05,
+      "loss": 6.9965,
       "step": 3484
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.8208092485549138e-05,
-      "loss": 6.5267,
+      "learning_rate": 8.471098265895953e-05,
+      "loss": 6.9971,
       "step": 3485
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.8179190751445088e-05,
-      "loss": 6.4767,
+      "learning_rate": 8.46242774566474e-05,
+      "loss": 7.0059,
       "step": 3486
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.815028901734104e-05,
-      "loss": 6.3602,
+      "learning_rate": 8.453757225433526e-05,
+      "loss": 6.9012,
       "step": 3487
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.8121387283236995e-05,
-      "loss": 6.4169,
+      "learning_rate": 8.445086705202311e-05,
+      "loss": 6.9161,
       "step": 3488
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.8092485549132952e-05,
-      "loss": 6.3471,
+      "learning_rate": 8.436416184971097e-05,
+      "loss": 6.9912,
       "step": 3489
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.8063583815028905e-05,
-      "loss": 6.3879,
+      "learning_rate": 8.427745664739884e-05,
+      "loss": 7.0854,
       "step": 3490
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.8034682080924855e-05,
-      "loss": 6.1766,
+      "learning_rate": 8.41907514450867e-05,
+      "loss": 6.9357,
       "step": 3491
     },
     {
       "epoch": 3.91,
-      "learning_rate": 2.800578034682081e-05,
-      "loss": 6.1161,
+      "learning_rate": 8.410404624277456e-05,
+      "loss": 6.9532,
       "step": 3492
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.7976878612716766e-05,
-      "loss": 6.1352,
+      "learning_rate": 8.401734104046241e-05,
+      "loss": 6.9709,
       "step": 3493
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.794797687861272e-05,
-      "loss": 5.9167,
+      "learning_rate": 8.393063583815028e-05,
+      "loss": 6.8741,
       "step": 3494
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.7919075144508673e-05,
-      "loss": 6.2858,
+      "learning_rate": 8.384393063583814e-05,
+      "loss": 7.1401,
       "step": 3495
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.7890173410404623e-05,
-      "loss": 5.8726,
+      "learning_rate": 8.3757225433526e-05,
+      "loss": 6.8519,
       "step": 3496
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.7861271676300583e-05,
-      "loss": 5.9033,
+      "learning_rate": 8.367052023121386e-05,
+      "loss": 6.9683,
       "step": 3497
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.7832369942196533e-05,
-      "loss": 5.9026,
+      "learning_rate": 8.358381502890174e-05,
+      "loss": 6.9166,
       "step": 3498
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.7803468208092486e-05,
-      "loss": 5.9418,
+      "learning_rate": 8.34971098265896e-05,
+      "loss": 7.0192,
       "step": 3499
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.777456647398844e-05,
-      "loss": 5.9739,
+      "learning_rate": 8.341040462427745e-05,
+      "loss": 6.9823,
       "step": 3500
     },
     {
       "epoch": 3.92,
-      "learning_rate": 2.7745664739884393e-05,
-      "loss": 5.9502,
+      "learning_rate": 8.33236994219653e-05,
+      "loss": 7.0363,
       "step": 3501
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.771676300578035e-05,
-      "loss": 5.9345,
+      "learning_rate": 8.323699421965317e-05,
+      "loss": 6.9968,
       "step": 3502
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.7687861271676304e-05,
-      "loss": 5.748,
+      "learning_rate": 8.315028901734104e-05,
+      "loss": 6.9161,
       "step": 3503
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.7658959537572254e-05,
-      "loss": 5.7392,
+      "learning_rate": 8.30635838150289e-05,
+      "loss": 6.9019,
       "step": 3504
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.7630057803468207e-05,
-      "loss": 5.7934,
+      "learning_rate": 8.297687861271675e-05,
+      "loss": 6.9788,
       "step": 3505
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.7601156069364164e-05,
-      "loss": 5.5893,
+      "learning_rate": 8.289017341040461e-05,
+      "loss": 6.8525,
       "step": 3506
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.7572254335260117e-05,
-      "loss": 5.6744,
+      "learning_rate": 8.280346820809248e-05,
+      "loss": 6.8796,
       "step": 3507
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.754335260115607e-05,
-      "loss": 5.4706,
+      "learning_rate": 8.271676300578034e-05,
+      "loss": 6.8319,
       "step": 3508
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.751445086705202e-05,
-      "loss": 5.6207,
+      "learning_rate": 8.26300578034682e-05,
+      "loss": 6.892,
       "step": 3509
     },
     {
       "epoch": 3.93,
-      "learning_rate": 2.748554913294798e-05,
-      "loss": 5.5856,
+      "learning_rate": 8.254335260115605e-05,
+      "loss": 6.9033,
       "step": 3510
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.745664739884393e-05,
-      "loss": 5.5927,
+      "learning_rate": 8.245664739884392e-05,
+      "loss": 6.9526,
       "step": 3511
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.7427745664739885e-05,
-      "loss": 5.4905,
+      "learning_rate": 8.236994219653178e-05,
+      "loss": 6.9222,
       "step": 3512
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.7398843930635838e-05,
-      "loss": 5.4233,
+      "learning_rate": 8.228323699421964e-05,
+      "loss": 6.8827,
       "step": 3513
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.7369942196531795e-05,
-      "loss": 5.3989,
+      "learning_rate": 8.21965317919075e-05,
+      "loss": 6.8438,
       "step": 3514
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.734104046242775e-05,
-      "loss": 5.5984,
+      "learning_rate": 8.210982658959538e-05,
+      "loss": 7.0015,
       "step": 3515
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.73121387283237e-05,
-      "loss": 5.4248,
+      "learning_rate": 8.202312138728322e-05,
+      "loss": 6.9578,
       "step": 3516
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.7283236994219652e-05,
-      "loss": 5.3162,
+      "learning_rate": 8.193641618497108e-05,
+      "loss": 6.8577,
       "step": 3517
     },
     {
       "epoch": 3.94,
-      "learning_rate": 2.725433526011561e-05,
-      "loss": 5.5014,
+      "learning_rate": 8.184971098265895e-05,
+      "loss": 7.0006,
       "step": 3518
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.7225433526011562e-05,
-      "loss": 5.248,
+      "learning_rate": 8.176300578034682e-05,
+      "loss": 6.8948,
       "step": 3519
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.7196531791907516e-05,
-      "loss": 5.0012,
+      "learning_rate": 8.167630057803468e-05,
+      "loss": 6.7669,
       "step": 3520
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.7167630057803466e-05,
-      "loss": 4.9062,
+      "learning_rate": 8.158959537572253e-05,
+      "loss": 6.8055,
       "step": 3521
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.7138728323699426e-05,
-      "loss": 4.9968,
+      "learning_rate": 8.150289017341039e-05,
+      "loss": 6.9408,
       "step": 3522
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.7109826589595376e-05,
-      "loss": 4.7358,
+      "learning_rate": 8.141618497109826e-05,
+      "loss": 6.7583,
       "step": 3523
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.708092485549133e-05,
-      "loss": 4.4911,
+      "learning_rate": 8.132947976878612e-05,
+      "loss": 6.7596,
       "step": 3524
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.7052023121387283e-05,
-      "loss": 4.0858,
+      "learning_rate": 8.124277456647398e-05,
+      "loss": 6.554,
       "step": 3525
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.702312138728324e-05,
-      "loss": 3.9942,
+      "learning_rate": 8.115606936416184e-05,
+      "loss": 6.5716,
       "step": 3526
     },
     {
       "epoch": 3.95,
-      "learning_rate": 2.6994219653179193e-05,
-      "loss": 6.5906,
+      "learning_rate": 8.106936416184972e-05,
+      "loss": 7.0449,
       "step": 3527
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.6965317919075143e-05,
-      "loss": 6.3362,
+      "learning_rate": 8.098265895953756e-05,
+      "loss": 6.811,
       "step": 3528
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.6936416184971097e-05,
-      "loss": 6.4671,
+      "learning_rate": 8.089595375722542e-05,
+      "loss": 6.9522,
       "step": 3529
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.6907514450867057e-05,
-      "loss": 6.4544,
+      "learning_rate": 8.080924855491328e-05,
+      "loss": 6.9564,
       "step": 3530
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.6878612716763007e-05,
-      "loss": 6.2912,
+      "learning_rate": 8.072254335260116e-05,
+      "loss": 6.9103,
       "step": 3531
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.684971098265896e-05,
-      "loss": 6.2342,
+      "learning_rate": 8.063583815028902e-05,
+      "loss": 6.953,
       "step": 3532
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.6820809248554914e-05,
-      "loss": 6.1703,
+      "learning_rate": 8.054913294797686e-05,
+      "loss": 6.8899,
       "step": 3533
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.679190751445087e-05,
-      "loss": 6.118,
+      "learning_rate": 8.046242774566472e-05,
+      "loss": 6.9235,
       "step": 3534
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.6763005780346824e-05,
-      "loss": 6.0146,
+      "learning_rate": 8.03757225433526e-05,
+      "loss": 6.88,
       "step": 3535
     },
     {
       "epoch": 3.96,
-      "learning_rate": 2.6734104046242774e-05,
-      "loss": 6.0629,
+      "learning_rate": 8.028901734104046e-05,
+      "loss": 6.9287,
       "step": 3536
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.6705202312138728e-05,
-      "loss": 6.0616,
+      "learning_rate": 8.020231213872832e-05,
+      "loss": 6.9323,
       "step": 3537
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.6676300578034684e-05,
-      "loss": 5.9266,
+      "learning_rate": 8.011560693641617e-05,
+      "loss": 6.9301,
       "step": 3538
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.6647398843930638e-05,
-      "loss": 5.8319,
+      "learning_rate": 8.002890173410404e-05,
+      "loss": 6.8659,
       "step": 3539
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.661849710982659e-05,
-      "loss": 5.8814,
+      "learning_rate": 7.99421965317919e-05,
+      "loss": 6.9917,
       "step": 3540
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.658959537572254e-05,
-      "loss": 5.8063,
+      "learning_rate": 7.985549132947976e-05,
+      "loss": 6.9694,
       "step": 3541
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.65606936416185e-05,
-      "loss": 5.7783,
+      "learning_rate": 7.976878612716762e-05,
+      "loss": 6.9856,
       "step": 3542
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.653179190751445e-05,
-      "loss": 5.8857,
+      "learning_rate": 7.96820809248555e-05,
+      "loss": 7.1248,
       "step": 3543
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.6502890173410405e-05,
-      "loss": 5.8285,
+      "learning_rate": 7.959537572254334e-05,
+      "loss": 7.0151,
       "step": 3544
     },
     {
       "epoch": 3.97,
-      "learning_rate": 2.647398843930636e-05,
-      "loss": 5.6414,
+      "learning_rate": 7.95086705202312e-05,
+      "loss": 6.9304,
       "step": 3545
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.6445086705202315e-05,
-      "loss": 5.7204,
+      "learning_rate": 7.942196531791906e-05,
+      "loss": 6.9135,
       "step": 3546
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.641618497109827e-05,
-      "loss": 5.6493,
+      "learning_rate": 7.933526011560694e-05,
+      "loss": 6.9219,
       "step": 3547
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.638728323699422e-05,
-      "loss": 5.5828,
+      "learning_rate": 7.92485549132948e-05,
+      "loss": 6.9109,
       "step": 3548
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.6358381502890172e-05,
-      "loss": 5.6409,
+      "learning_rate": 7.916184971098266e-05,
+      "loss": 6.9533,
       "step": 3549
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.632947976878613e-05,
-      "loss": 5.4222,
+      "learning_rate": 7.90751445086705e-05,
+      "loss": 6.849,
       "step": 3550
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.6300578034682083e-05,
-      "loss": 5.7336,
+      "learning_rate": 7.898843930635838e-05,
+      "loss": 7.1183,
       "step": 3551
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.6271676300578036e-05,
-      "loss": 5.7722,
+      "learning_rate": 7.890173410404624e-05,
+      "loss": 7.0645,
       "step": 3552
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.6242774566473986e-05,
-      "loss": 5.4519,
+      "learning_rate": 7.88150289017341e-05,
+      "loss": 6.942,
       "step": 3553
     },
     {
       "epoch": 3.98,
-      "learning_rate": 2.6213872832369946e-05,
-      "loss": 5.4769,
+      "learning_rate": 7.872832369942196e-05,
+      "loss": 6.9667,
       "step": 3554
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.6184971098265896e-05,
-      "loss": 5.7153,
+      "learning_rate": 7.864161849710983e-05,
+      "loss": 7.0767,
       "step": 3555
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.615606936416185e-05,
-      "loss": 5.5773,
+      "learning_rate": 7.855491329479768e-05,
+      "loss": 6.8794,
       "step": 3556
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.6127167630057803e-05,
-      "loss": 5.5697,
+      "learning_rate": 7.846820809248554e-05,
+      "loss": 6.9251,
       "step": 3557
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.609826589595376e-05,
-      "loss": 5.4797,
+      "learning_rate": 7.83815028901734e-05,
+      "loss": 6.9615,
       "step": 3558
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.6069364161849714e-05,
-      "loss": 5.5339,
+      "learning_rate": 7.829479768786127e-05,
+      "loss": 7.1067,
       "step": 3559
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.6040462427745667e-05,
-      "loss": 5.5235,
+      "learning_rate": 7.820809248554913e-05,
+      "loss": 7.1583,
       "step": 3560
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.6011560693641617e-05,
-      "loss": 5.2254,
+      "learning_rate": 7.812138728323698e-05,
+      "loss": 6.9358,
       "step": 3561
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.5982658959537577e-05,
-      "loss": 5.3352,
+      "learning_rate": 7.803468208092484e-05,
+      "loss": 7.0677,
       "step": 3562
     },
     {
       "epoch": 3.99,
-      "learning_rate": 2.5953757225433527e-05,
-      "loss": 5.1421,
+      "learning_rate": 7.794797687861271e-05,
+      "loss": 6.895,
       "step": 3563
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.592485549132948e-05,
-      "loss": 4.934,
+      "learning_rate": 7.786127167630058e-05,
+      "loss": 6.8874,
       "step": 3564
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.5895953757225434e-05,
-      "loss": 4.6501,
+      "learning_rate": 7.777456647398844e-05,
+      "loss": 6.7598,
       "step": 3565
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.586705202312139e-05,
-      "loss": 4.5022,
+      "learning_rate": 7.768786127167628e-05,
+      "loss": 6.7982,
       "step": 3566
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.5838150289017344e-05,
-      "loss": 4.2806,
+      "learning_rate": 7.760115606936416e-05,
+      "loss": 6.6764,
       "step": 3567
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.5809248554913295e-05,
-      "loss": 4.2039,
+      "learning_rate": 7.751445086705202e-05,
+      "loss": 6.7002,
       "step": 3568
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.5780346820809248e-05,
-      "loss": 6.7141,
+      "learning_rate": 7.742774566473988e-05,
+      "loss": 7.0624,
       "step": 3569
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.5751445086705205e-05,
-      "loss": 6.6321,
+      "learning_rate": 7.734104046242774e-05,
+      "loss": 7.0686,
       "step": 3570
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.5722543352601158e-05,
-      "loss": 6.3952,
+      "learning_rate": 7.725433526011561e-05,
+      "loss": 6.8804,
       "step": 3571
     },
     {
       "epoch": 4.0,
-      "learning_rate": 2.5693641618497112e-05,
-      "loss": 6.4987,
+      "learning_rate": 7.716763005780346e-05,
+      "loss": 7.0076,
       "step": 3572
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.5664739884393062e-05,
-      "loss": 6.2566,
+      "learning_rate": 7.708092485549132e-05,
+      "loss": 6.8853,
       "step": 3573
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.5635838150289022e-05,
-      "loss": 6.3289,
+      "learning_rate": 7.699421965317918e-05,
+      "loss": 7.0372,
       "step": 3574
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.5606936416184972e-05,
-      "loss": 6.1318,
+      "learning_rate": 7.690751445086705e-05,
+      "loss": 7.0066,
       "step": 3575
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.5578034682080925e-05,
-      "loss": 5.952,
+      "learning_rate": 7.682080924855491e-05,
+      "loss": 7.0166,
       "step": 3576
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.554913294797688e-05,
-      "loss": 5.8687,
+      "learning_rate": 7.673410404624277e-05,
+      "loss": 6.9352,
       "step": 3577
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.5520231213872836e-05,
-      "loss": 5.8383,
+      "learning_rate": 7.664739884393062e-05,
+      "loss": 6.8799,
       "step": 3578
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.549132947976879e-05,
-      "loss": 6.0817,
+      "learning_rate": 7.65606936416185e-05,
+      "loss": 6.8862,
       "step": 3579
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.546242774566474e-05,
-      "loss": 6.1188,
+      "learning_rate": 7.647398843930635e-05,
+      "loss": 6.8478,
       "step": 3580
     },
     {
       "epoch": 4.01,
-      "learning_rate": 2.5433526011560693e-05,
-      "loss": 6.2177,
+      "learning_rate": 7.638728323699421e-05,
+      "loss": 6.8784,
       "step": 3581
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.540462427745665e-05,
-      "loss": 6.3255,
+      "learning_rate": 7.630057803468207e-05,
+      "loss": 6.8748,
       "step": 3582
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.5375722543352603e-05,
-      "loss": 6.122,
+      "learning_rate": 7.621387283236995e-05,
+      "loss": 6.9627,
       "step": 3583
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.5346820809248556e-05,
-      "loss": 6.0617,
+      "learning_rate": 7.61271676300578e-05,
+      "loss": 7.043,
       "step": 3584
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.531791907514451e-05,
-      "loss": 5.8157,
+      "learning_rate": 7.604046242774566e-05,
+      "loss": 6.9294,
       "step": 3585
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.5289017341040467e-05,
-      "loss": 5.7214,
+      "learning_rate": 7.595375722543352e-05,
+      "loss": 6.9689,
       "step": 3586
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.526011560693642e-05,
-      "loss": 5.7333,
+      "learning_rate": 7.586705202312139e-05,
+      "loss": 6.9653,
       "step": 3587
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.523121387283237e-05,
-      "loss": 5.7087,
+      "learning_rate": 7.578034682080925e-05,
+      "loss": 6.8732,
       "step": 3588
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.5202312138728324e-05,
-      "loss": 5.717,
+      "learning_rate": 7.56936416184971e-05,
+      "loss": 6.9132,
       "step": 3589
     },
     {
       "epoch": 4.02,
-      "learning_rate": 2.517341040462428e-05,
-      "loss": 5.9284,
+      "learning_rate": 7.560693641618496e-05,
+      "loss": 6.9616,
       "step": 3590
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.5144508670520234e-05,
-      "loss": 5.9431,
+      "learning_rate": 7.552023121387283e-05,
+      "loss": 7.0856,
       "step": 3591
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.5115606936416187e-05,
-      "loss": 5.5278,
+      "learning_rate": 7.543352601156069e-05,
+      "loss": 6.8631,
       "step": 3592
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.5086705202312137e-05,
-      "loss": 5.4409,
+      "learning_rate": 7.534682080924855e-05,
+      "loss": 6.8856,
       "step": 3593
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.5057803468208098e-05,
-      "loss": 5.4167,
+      "learning_rate": 7.52601156069364e-05,
+      "loss": 6.8453,
       "step": 3594
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.5028901734104048e-05,
-      "loss": 5.6539,
+      "learning_rate": 7.517341040462427e-05,
+      "loss": 6.9564,
       "step": 3595
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.5e-05,
-      "loss": 5.4763,
+      "learning_rate": 7.508670520231213e-05,
+      "loss": 6.8253,
       "step": 3596
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.4971098265895955e-05,
-      "loss": 5.7778,
+      "learning_rate": 7.5e-05,
+      "loss": 7.0389,
       "step": 3597
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.4942196531791908e-05,
-      "loss": 5.6501,
+      "learning_rate": 7.491329479768785e-05,
+      "loss": 7.0128,
       "step": 3598
     },
     {
       "epoch": 4.03,
-      "learning_rate": 2.491329479768786e-05,
-      "loss": 5.5325,
+      "learning_rate": 7.482658959537571e-05,
+      "loss": 7.0146,
       "step": 3599
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.4884393063583815e-05,
-      "loss": 5.4527,
+      "learning_rate": 7.473988439306357e-05,
+      "loss": 6.9781,
       "step": 3600
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.485549132947977e-05,
-      "loss": 5.3832,
+      "learning_rate": 7.465317919075143e-05,
+      "loss": 6.8081,
       "step": 3601
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.4826589595375725e-05,
-      "loss": 5.4814,
+      "learning_rate": 7.45664739884393e-05,
+      "loss": 6.886,
       "step": 3602
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.4797687861271675e-05,
-      "loss": 5.4953,
+      "learning_rate": 7.447976878612716e-05,
+      "loss": 6.9431,
       "step": 3603
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.4768786127167632e-05,
-      "loss": 5.4067,
+      "learning_rate": 7.439306358381502e-05,
+      "loss": 6.9914,
       "step": 3604
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.4739884393063582e-05,
-      "loss": 5.3146,
+      "learning_rate": 7.430635838150289e-05,
+      "loss": 6.9057,
       "step": 3605
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.471098265895954e-05,
-      "loss": 5.4841,
+      "learning_rate": 7.421965317919074e-05,
+      "loss": 7.0448,
       "step": 3606
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.4682080924855492e-05,
-      "loss": 5.3597,
+      "learning_rate": 7.413294797687861e-05,
+      "loss": 6.9174,
       "step": 3607
     },
     {
       "epoch": 4.04,
-      "learning_rate": 2.4653179190751446e-05,
-      "loss": 5.2563,
+      "learning_rate": 7.404624277456646e-05,
+      "loss": 6.8715,
       "step": 3608
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.46242774566474e-05,
-      "loss": 5.1591,
+      "learning_rate": 7.395953757225433e-05,
+      "loss": 6.8857,
       "step": 3609
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.4595375722543353e-05,
-      "loss": 5.3594,
+      "learning_rate": 7.387283236994219e-05,
+      "loss": 7.0687,
       "step": 3610
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.4566473988439306e-05,
-      "loss": 5.1041,
+      "learning_rate": 7.378612716763005e-05,
+      "loss": 6.9719,
       "step": 3611
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.4537572254335263e-05,
-      "loss": 4.9924,
+      "learning_rate": 7.369942196531791e-05,
+      "loss": 6.9503,
       "step": 3612
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.4508670520231213e-05,
-      "loss": 4.8589,
+      "learning_rate": 7.361271676300577e-05,
+      "loss": 6.8963,
       "step": 3613
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.447976878612717e-05,
-      "loss": 4.956,
+      "learning_rate": 7.352601156069363e-05,
+      "loss": 6.9857,
       "step": 3614
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.445086705202312e-05,
-      "loss": 4.6206,
+      "learning_rate": 7.343930635838149e-05,
+      "loss": 6.7984,
       "step": 3615
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.4421965317919077e-05,
-      "loss": 4.3996,
+      "learning_rate": 7.335260115606935e-05,
+      "loss": 6.8271,
       "step": 3616
     },
     {
       "epoch": 4.05,
-      "learning_rate": 2.439306358381503e-05,
-      "loss": 4.54,
+      "learning_rate": 7.326589595375721e-05,
+      "loss": 6.9797,
       "step": 3617
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.4364161849710984e-05,
-      "loss": 3.7919,
+      "learning_rate": 7.317919075144507e-05,
+      "loss": 6.4174,
       "step": 3618
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.4335260115606937e-05,
-      "loss": 6.5658,
+      "learning_rate": 7.309248554913295e-05,
+      "loss": 6.998,
       "step": 3619
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.430635838150289e-05,
-      "loss": 6.4823,
+      "learning_rate": 7.30057803468208e-05,
+      "loss": 6.868,
       "step": 3620
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.4277456647398844e-05,
-      "loss": 6.4512,
+      "learning_rate": 7.291907514450867e-05,
+      "loss": 6.931,
       "step": 3621
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.4248554913294798e-05,
-      "loss": 6.5247,
+      "learning_rate": 7.283236994219653e-05,
+      "loss": 6.9128,
       "step": 3622
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.421965317919075e-05,
-      "loss": 6.3722,
+      "learning_rate": 7.274566473988439e-05,
+      "loss": 6.8694,
       "step": 3623
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.4190751445086708e-05,
-      "loss": 6.3945,
+      "learning_rate": 7.265895953757225e-05,
+      "loss": 7.0234,
       "step": 3624
     },
     {
       "epoch": 4.06,
-      "learning_rate": 2.4161849710982658e-05,
-      "loss": 6.3302,
+      "learning_rate": 7.257225433526011e-05,
+      "loss": 7.0583,
       "step": 3625
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.4132947976878615e-05,
-      "loss": 6.148,
+      "learning_rate": 7.248554913294797e-05,
+      "loss": 6.9629,
       "step": 3626
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.4104046242774568e-05,
-      "loss": 6.0382,
+      "learning_rate": 7.239884393063583e-05,
+      "loss": 6.8813,
       "step": 3627
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.407514450867052e-05,
-      "loss": 6.0478,
+      "learning_rate": 7.231213872832369e-05,
+      "loss": 7.0126,
       "step": 3628
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.4046242774566475e-05,
-      "loss": 5.9305,
+      "learning_rate": 7.222543352601155e-05,
+      "loss": 6.9588,
       "step": 3629
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.401734104046243e-05,
-      "loss": 5.9585,
+      "learning_rate": 7.213872832369941e-05,
+      "loss": 6.8871,
       "step": 3630
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.3988439306358382e-05,
-      "loss": 6.0226,
+      "learning_rate": 7.205202312138727e-05,
+      "loss": 6.8823,
       "step": 3631
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.3959537572254335e-05,
-      "loss": 6.1637,
+      "learning_rate": 7.196531791907513e-05,
+      "loss": 7.0355,
       "step": 3632
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.393063583815029e-05,
-      "loss": 6.2469,
+      "learning_rate": 7.1878612716763e-05,
+      "loss": 7.0359,
       "step": 3633
     },
     {
       "epoch": 4.07,
-      "learning_rate": 2.3901734104046246e-05,
-      "loss": 6.0979,
+      "learning_rate": 7.179190751445085e-05,
+      "loss": 7.0187,
       "step": 3634
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.3872832369942196e-05,
-      "loss": 5.8992,
+      "learning_rate": 7.170520231213873e-05,
+      "loss": 6.9206,
       "step": 3635
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.3843930635838153e-05,
-      "loss": 5.6381,
+      "learning_rate": 7.161849710982659e-05,
+      "loss": 6.8472,
       "step": 3636
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.3815028901734106e-05,
-      "loss": 5.391,
+      "learning_rate": 7.153179190751445e-05,
+      "loss": 6.7181,
       "step": 3637
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.378612716763006e-05,
-      "loss": 5.4923,
+      "learning_rate": 7.144508670520231e-05,
+      "loss": 6.8762,
       "step": 3638
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.3757225433526013e-05,
-      "loss": 5.6729,
+      "learning_rate": 7.135838150289017e-05,
+      "loss": 6.9842,
       "step": 3639
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.3728323699421966e-05,
-      "loss": 5.4955,
+      "learning_rate": 7.127167630057803e-05,
+      "loss": 6.855,
       "step": 3640
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.369942196531792e-05,
-      "loss": 5.5478,
+      "learning_rate": 7.118497109826589e-05,
+      "loss": 6.856,
       "step": 3641
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.3670520231213873e-05,
-      "loss": 5.6194,
+      "learning_rate": 7.109826589595375e-05,
+      "loss": 6.9564,
       "step": 3642
     },
     {
       "epoch": 4.08,
-      "learning_rate": 2.3641618497109827e-05,
-      "loss": 5.4703,
+      "learning_rate": 7.101156069364161e-05,
+      "loss": 6.9358,
       "step": 3643
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.3612716763005784e-05,
-      "loss": 5.5188,
+      "learning_rate": 7.092485549132947e-05,
+      "loss": 6.9312,
       "step": 3644
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.3583815028901734e-05,
-      "loss": 5.5047,
+      "learning_rate": 7.083815028901733e-05,
+      "loss": 6.8643,
       "step": 3645
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.355491329479769e-05,
-      "loss": 5.5032,
+      "learning_rate": 7.075144508670519e-05,
+      "loss": 6.9824,
       "step": 3646
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.352601156069364e-05,
-      "loss": 5.6191,
+      "learning_rate": 7.066473988439306e-05,
+      "loss": 6.9827,
       "step": 3647
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.3497109826589597e-05,
-      "loss": 5.4026,
+      "learning_rate": 7.057803468208091e-05,
+      "loss": 6.9262,
       "step": 3648
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.346820809248555e-05,
-      "loss": 5.3747,
+      "learning_rate": 7.049132947976879e-05,
+      "loss": 6.8913,
       "step": 3649
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.3439306358381504e-05,
-      "loss": 5.4775,
+      "learning_rate": 7.040462427745665e-05,
+      "loss": 7.012,
       "step": 3650
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.3410404624277458e-05,
-      "loss": 5.4561,
+      "learning_rate": 7.03179190751445e-05,
+      "loss": 6.9546,
       "step": 3651
     },
     {
       "epoch": 4.09,
-      "learning_rate": 2.338150289017341e-05,
-      "loss": 5.3009,
+      "learning_rate": 7.023121387283237e-05,
+      "loss": 6.9517,
       "step": 3652
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.3352601156069365e-05,
-      "loss": 5.2025,
+      "learning_rate": 7.014450867052023e-05,
+      "loss": 6.7977,
       "step": 3653
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.332369942196532e-05,
-      "loss": 5.3928,
+      "learning_rate": 7.005780346820809e-05,
+      "loss": 6.9611,
       "step": 3654
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.329479768786127e-05,
-      "loss": 5.3654,
+      "learning_rate": 6.997109826589595e-05,
+      "loss": 6.9328,
       "step": 3655
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.326589595375723e-05,
-      "loss": 5.3492,
+      "learning_rate": 6.988439306358381e-05,
+      "loss": 6.9467,
       "step": 3656
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.323699421965318e-05,
-      "loss": 5.3293,
+      "learning_rate": 6.979768786127167e-05,
+      "loss": 7.0084,
       "step": 3657
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.3208092485549135e-05,
-      "loss": 5.3543,
+      "learning_rate": 6.971098265895953e-05,
+      "loss": 7.0582,
       "step": 3658
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.317919075144509e-05,
-      "loss": 5.098,
+      "learning_rate": 6.962427745664739e-05,
+      "loss": 6.8726,
       "step": 3659
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.3150289017341042e-05,
-      "loss": 4.9798,
+      "learning_rate": 6.953757225433525e-05,
+      "loss": 6.831,
       "step": 3660
     },
     {
       "epoch": 4.1,
-      "learning_rate": 2.3121387283236996e-05,
-      "loss": 4.8648,
+      "learning_rate": 6.945086705202312e-05,
+      "loss": 6.7949,
       "step": 3661
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.309248554913295e-05,
-      "loss": 4.7863,
+      "learning_rate": 6.936416184971097e-05,
+      "loss": 6.8714,
       "step": 3662
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.3063583815028902e-05,
-      "loss": 4.9495,
+      "learning_rate": 6.927745664739884e-05,
+      "loss": 6.9335,
       "step": 3663
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.303468208092486e-05,
-      "loss": 4.5006,
+      "learning_rate": 6.91907514450867e-05,
+      "loss": 6.7484,
       "step": 3664
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.300578034682081e-05,
-      "loss": 4.2424,
+      "learning_rate": 6.910404624277456e-05,
+      "loss": 6.6647,
       "step": 3665
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.2976878612716766e-05,
-      "loss": 4.1985,
+      "learning_rate": 6.901734104046242e-05,
+      "loss": 6.7365,
       "step": 3666
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.2947976878612716e-05,
-      "loss": 4.4271,
+      "learning_rate": 6.893063583815028e-05,
+      "loss": 6.8466,
       "step": 3667
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.2919075144508673e-05,
-      "loss": 3.5912,
+      "learning_rate": 6.884393063583815e-05,
+      "loss": 6.5066,
       "step": 3668
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.2890173410404627e-05,
-      "loss": 6.5155,
+      "learning_rate": 6.8757225433526e-05,
+      "loss": 6.9878,
       "step": 3669
     },
     {
       "epoch": 4.11,
-      "learning_rate": 2.286127167630058e-05,
-      "loss": 6.5081,
+      "learning_rate": 6.867052023121387e-05,
+      "loss": 7.0257,
       "step": 3670
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.2832369942196533e-05,
-      "loss": 6.4572,
+      "learning_rate": 6.858381502890173e-05,
+      "loss": 6.979,
       "step": 3671
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.2803468208092487e-05,
-      "loss": 6.333,
+      "learning_rate": 6.849710982658959e-05,
+      "loss": 6.9768,
       "step": 3672
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.277456647398844e-05,
-      "loss": 6.0303,
+      "learning_rate": 6.841040462427746e-05,
+      "loss": 6.8317,
       "step": 3673
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.2745664739884394e-05,
-      "loss": 6.0506,
+      "learning_rate": 6.832369942196531e-05,
+      "loss": 6.8511,
       "step": 3674
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.2716763005780347e-05,
-      "loss": 6.1549,
+      "learning_rate": 6.823699421965318e-05,
+      "loss": 7.041,
       "step": 3675
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.2687861271676304e-05,
-      "loss": 6.0321,
+      "learning_rate": 6.815028901734103e-05,
+      "loss": 6.933,
       "step": 3676
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.2658959537572254e-05,
-      "loss": 6.0422,
+      "learning_rate": 6.80635838150289e-05,
+      "loss": 6.9499,
       "step": 3677
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.263005780346821e-05,
-      "loss": 5.8937,
+      "learning_rate": 6.797687861271676e-05,
+      "loss": 6.771,
       "step": 3678
     },
     {
       "epoch": 4.12,
-      "learning_rate": 2.2601156069364164e-05,
-      "loss": 6.0523,
+      "learning_rate": 6.789017341040462e-05,
+      "loss": 6.9424,
       "step": 3679
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.2572254335260118e-05,
-      "loss": 5.8994,
+      "learning_rate": 6.780346820809248e-05,
+      "loss": 6.8462,
       "step": 3680
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.254335260115607e-05,
-      "loss": 5.9659,
+      "learning_rate": 6.771676300578034e-05,
+      "loss": 6.8983,
       "step": 3681
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.2514450867052025e-05,
-      "loss": 5.866,
+      "learning_rate": 6.76300578034682e-05,
+      "loss": 6.8978,
       "step": 3682
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.2485549132947978e-05,
-      "loss": 5.7765,
+      "learning_rate": 6.754335260115606e-05,
+      "loss": 6.9739,
       "step": 3683
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.245664739884393e-05,
-      "loss": 5.5946,
+      "learning_rate": 6.745664739884392e-05,
+      "loss": 6.9093,
       "step": 3684
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.2427745664739885e-05,
-      "loss": 5.5499,
+      "learning_rate": 6.736994219653178e-05,
+      "loss": 6.8759,
       "step": 3685
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.2398843930635842e-05,
-      "loss": 5.8191,
+      "learning_rate": 6.728323699421964e-05,
+      "loss": 7.008,
       "step": 3686
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.2369942196531792e-05,
-      "loss": 5.6596,
+      "learning_rate": 6.719653179190752e-05,
+      "loss": 6.9592,
       "step": 3687
     },
     {
       "epoch": 4.13,
-      "learning_rate": 2.234104046242775e-05,
-      "loss": 5.505,
+      "learning_rate": 6.710982658959537e-05,
+      "loss": 6.8853,
       "step": 3688
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.23121387283237e-05,
-      "loss": 5.5112,
+      "learning_rate": 6.702312138728324e-05,
+      "loss": 6.8727,
       "step": 3689
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.2283236994219656e-05,
-      "loss": 5.5404,
+      "learning_rate": 6.693641618497109e-05,
+      "loss": 6.8997,
       "step": 3690
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.225433526011561e-05,
-      "loss": 5.8148,
+      "learning_rate": 6.684971098265896e-05,
+      "loss": 7.0329,
       "step": 3691
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.2225433526011563e-05,
-      "loss": 5.4573,
+      "learning_rate": 6.676300578034682e-05,
+      "loss": 6.7747,
       "step": 3692
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.2196531791907516e-05,
-      "loss": 5.4866,
+      "learning_rate": 6.667630057803468e-05,
+      "loss": 6.8131,
       "step": 3693
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.216763005780347e-05,
-      "loss": 5.4617,
+      "learning_rate": 6.658959537572254e-05,
+      "loss": 6.8831,
       "step": 3694
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.2138728323699423e-05,
-      "loss": 5.3662,
+      "learning_rate": 6.650289017341039e-05,
+      "loss": 6.8266,
       "step": 3695
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.2109826589595376e-05,
-      "loss": 5.2912,
+      "learning_rate": 6.641618497109826e-05,
+      "loss": 6.7917,
       "step": 3696
     },
     {
       "epoch": 4.14,
-      "learning_rate": 2.208092485549133e-05,
-      "loss": 5.4562,
+      "learning_rate": 6.632947976878612e-05,
+      "loss": 6.943,
       "step": 3697
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.2052023121387283e-05,
-      "loss": 5.7212,
+      "learning_rate": 6.624277456647398e-05,
+      "loss": 7.1945,
       "step": 3698
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.2023121387283237e-05,
-      "loss": 5.4432,
+      "learning_rate": 6.615606936416184e-05,
+      "loss": 6.9582,
       "step": 3699
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.199421965317919e-05,
-      "loss": 5.3639,
+      "learning_rate": 6.60693641618497e-05,
+      "loss": 6.9691,
       "step": 3700
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.1965317919075147e-05,
-      "loss": 5.3482,
+      "learning_rate": 6.598265895953756e-05,
+      "loss": 6.8532,
       "step": 3701
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.1936416184971097e-05,
-      "loss": 5.3963,
+      "learning_rate": 6.589595375722542e-05,
+      "loss": 6.8912,
       "step": 3702
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.1907514450867054e-05,
-      "loss": 5.3025,
+      "learning_rate": 6.580924855491328e-05,
+      "loss": 6.9396,
       "step": 3703
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.1878612716763007e-05,
-      "loss": 5.2745,
+      "learning_rate": 6.572254335260114e-05,
+      "loss": 6.9073,
       "step": 3704
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.184971098265896e-05,
-      "loss": 5.4875,
+      "learning_rate": 6.5635838150289e-05,
+      "loss": 7.0158,
       "step": 3705
     },
     {
       "epoch": 4.15,
-      "learning_rate": 2.1820809248554914e-05,
-      "loss": 5.3923,
+      "learning_rate": 6.554913294797688e-05,
+      "loss": 6.9749,
       "step": 3706
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.1791907514450868e-05,
-      "loss": 5.5267,
+      "learning_rate": 6.546242774566473e-05,
+      "loss": 7.0987,
       "step": 3707
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.176300578034682e-05,
-      "loss": 5.2431,
+      "learning_rate": 6.53757225433526e-05,
+      "loss": 6.9336,
       "step": 3708
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.1734104046242775e-05,
-      "loss": 5.0404,
+      "learning_rate": 6.528901734104046e-05,
+      "loss": 6.773,
       "step": 3709
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.1705202312138728e-05,
-      "loss": 5.4369,
+      "learning_rate": 6.520231213872832e-05,
+      "loss": 7.0427,
       "step": 3710
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.1676300578034685e-05,
-      "loss": 5.2531,
+      "learning_rate": 6.511560693641618e-05,
+      "loss": 6.9185,
       "step": 3711
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.1647398843930635e-05,
-      "loss": 5.4547,
+      "learning_rate": 6.502890173410404e-05,
+      "loss": 7.0125,
       "step": 3712
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.1618497109826592e-05,
-      "loss": 5.166,
+      "learning_rate": 6.49421965317919e-05,
+      "loss": 6.9541,
       "step": 3713
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.1589595375722542e-05,
-      "loss": 4.7408,
+      "learning_rate": 6.485549132947976e-05,
+      "loss": 6.8194,
       "step": 3714
     },
     {
       "epoch": 4.16,
-      "learning_rate": 2.15606936416185e-05,
-      "loss": 4.473,
+      "learning_rate": 6.476878612716762e-05,
+      "loss": 6.8023,
       "step": 3715
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.1531791907514452e-05,
-      "loss": 4.2133,
+      "learning_rate": 6.468208092485548e-05,
+      "loss": 6.6076,
       "step": 3716
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.1502890173410405e-05,
-      "loss": 3.9946,
+      "learning_rate": 6.459537572254334e-05,
+      "loss": 6.6521,
       "step": 3717
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.147398843930636e-05,
-      "loss": 4.0836,
+      "learning_rate": 6.45086705202312e-05,
+      "loss": 6.6705,
       "step": 3718
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.1445086705202312e-05,
-      "loss": 6.5834,
+      "learning_rate": 6.442196531791906e-05,
+      "loss": 7.0104,
       "step": 3719
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.1416184971098266e-05,
-      "loss": 6.6397,
+      "learning_rate": 6.433526011560694e-05,
+      "loss": 7.0139,
       "step": 3720
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.1387283236994223e-05,
-      "loss": 6.4812,
+      "learning_rate": 6.424855491329478e-05,
+      "loss": 6.9713,
       "step": 3721
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.1358381502890173e-05,
-      "loss": 6.5259,
+      "learning_rate": 6.416184971098266e-05,
+      "loss": 7.091,
       "step": 3722
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.132947976878613e-05,
-      "loss": 6.4196,
+      "learning_rate": 6.407514450867052e-05,
+      "loss": 6.9938,
       "step": 3723
     },
     {
       "epoch": 4.17,
-      "learning_rate": 2.130057803468208e-05,
-      "loss": 6.1659,
+      "learning_rate": 6.398843930635838e-05,
+      "loss": 6.9004,
       "step": 3724
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.1271676300578036e-05,
-      "loss": 6.274,
+      "learning_rate": 6.390173410404624e-05,
+      "loss": 6.9733,
       "step": 3725
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.124277456647399e-05,
-      "loss": 6.1598,
+      "learning_rate": 6.38150289017341e-05,
+      "loss": 7.0235,
       "step": 3726
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.1213872832369943e-05,
-      "loss": 5.8667,
+      "learning_rate": 6.372832369942196e-05,
+      "loss": 6.8565,
       "step": 3727
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.1184971098265897e-05,
-      "loss": 6.0368,
+      "learning_rate": 6.364161849710982e-05,
+      "loss": 7.0179,
       "step": 3728
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.115606936416185e-05,
-      "loss": 5.9911,
+      "learning_rate": 6.355491329479768e-05,
+      "loss": 6.9896,
       "step": 3729
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.1127167630057804e-05,
-      "loss": 6.0203,
+      "learning_rate": 6.346820809248554e-05,
+      "loss": 6.938,
       "step": 3730
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.1098265895953757e-05,
-      "loss": 6.024,
+      "learning_rate": 6.33815028901734e-05,
+      "loss": 6.9329,
       "step": 3731
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.106936416184971e-05,
-      "loss": 6.0574,
+      "learning_rate": 6.329479768786126e-05,
+      "loss": 6.9966,
       "step": 3732
     },
     {
       "epoch": 4.18,
-      "learning_rate": 2.1040462427745667e-05,
-      "loss": 5.9631,
+      "learning_rate": 6.320809248554912e-05,
+      "loss": 6.9129,
       "step": 3733
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.1011560693641617e-05,
-      "loss": 5.8301,
+      "learning_rate": 6.3121387283237e-05,
+      "loss": 6.902,
       "step": 3734
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.0982658959537574e-05,
-      "loss": 5.7569,
+      "learning_rate": 6.303468208092484e-05,
+      "loss": 6.9343,
       "step": 3735
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.0953757225433528e-05,
-      "loss": 5.4568,
+      "learning_rate": 6.294797687861272e-05,
+      "loss": 6.8273,
       "step": 3736
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.092485549132948e-05,
-      "loss": 5.6431,
+      "learning_rate": 6.286127167630058e-05,
+      "loss": 6.9749,
       "step": 3737
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.0895953757225435e-05,
-      "loss": 5.4383,
+      "learning_rate": 6.277456647398844e-05,
+      "loss": 6.8845,
       "step": 3738
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.0867052023121388e-05,
-      "loss": 5.6445,
+      "learning_rate": 6.26878612716763e-05,
+      "loss": 7.0338,
       "step": 3739
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.083815028901734e-05,
-      "loss": 5.5511,
+      "learning_rate": 6.260115606936416e-05,
+      "loss": 6.8807,
       "step": 3740
     },
     {
       "epoch": 4.19,
-      "learning_rate": 2.0809248554913295e-05,
-      "loss": 5.5655,
+      "learning_rate": 6.251445086705202e-05,
+      "loss": 6.9534,
       "step": 3741
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.078034682080925e-05,
-      "loss": 5.3784,
+      "learning_rate": 6.242774566473988e-05,
+      "loss": 6.9147,
       "step": 3742
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.0751445086705205e-05,
-      "loss": 5.6323,
+      "learning_rate": 6.234104046242774e-05,
+      "loss": 7.0554,
       "step": 3743
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.0722543352601155e-05,
-      "loss": 5.7378,
+      "learning_rate": 6.22543352601156e-05,
+      "loss": 7.1228,
       "step": 3744
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.0693641618497112e-05,
-      "loss": 5.4582,
+      "learning_rate": 6.216763005780346e-05,
+      "loss": 6.8456,
       "step": 3745
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.0664739884393066e-05,
-      "loss": 5.3318,
+      "learning_rate": 6.208092485549133e-05,
+      "loss": 6.8419,
       "step": 3746
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.063583815028902e-05,
-      "loss": 5.4517,
+      "learning_rate": 6.199421965317918e-05,
+      "loss": 6.9376,
       "step": 3747
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.0606936416184972e-05,
-      "loss": 5.3782,
+      "learning_rate": 6.190751445086705e-05,
+      "loss": 6.9456,
       "step": 3748
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.0578034682080926e-05,
-      "loss": 5.2712,
+      "learning_rate": 6.18208092485549e-05,
+      "loss": 6.8524,
       "step": 3749
     },
     {
       "epoch": 4.2,
-      "learning_rate": 2.054913294797688e-05,
-      "loss": 5.4604,
+      "learning_rate": 6.173410404624277e-05,
+      "loss": 6.9419,
       "step": 3750
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.0520231213872833e-05,
-      "loss": 5.4099,
+      "learning_rate": 6.164739884393063e-05,
+      "loss": 6.9562,
       "step": 3751
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.0491329479768786e-05,
-      "loss": 5.4074,
+      "learning_rate": 6.15606936416185e-05,
+      "loss": 6.9831,
       "step": 3752
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.0462427745664743e-05,
-      "loss": 5.3424,
+      "learning_rate": 6.147398843930636e-05,
+      "loss": 6.918,
       "step": 3753
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.0433526011560693e-05,
-      "loss": 5.3895,
+      "learning_rate": 6.138728323699422e-05,
+      "loss": 6.9837,
       "step": 3754
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.040462427745665e-05,
-      "loss": 5.4174,
+      "learning_rate": 6.130057803468208e-05,
+      "loss": 6.949,
       "step": 3755
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.03757225433526e-05,
-      "loss": 5.3372,
+      "learning_rate": 6.121387283236994e-05,
+      "loss": 6.9652,
       "step": 3756
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.0346820809248557e-05,
-      "loss": 5.4256,
+      "learning_rate": 6.11271676300578e-05,
+      "loss": 6.9829,
       "step": 3757
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.031791907514451e-05,
-      "loss": 5.198,
+      "learning_rate": 6.104046242774566e-05,
+      "loss": 7.0162,
       "step": 3758
     },
     {
       "epoch": 4.21,
-      "learning_rate": 2.0289017341040464e-05,
-      "loss": 4.902,
+      "learning_rate": 6.095375722543352e-05,
+      "loss": 6.8315,
       "step": 3759
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.0260115606936417e-05,
-      "loss": 5.1105,
+      "learning_rate": 6.0867052023121385e-05,
+      "loss": 6.9604,
       "step": 3760
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.023121387283237e-05,
-      "loss": 4.8741,
+      "learning_rate": 6.0780346820809245e-05,
+      "loss": 6.9258,
       "step": 3761
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.0202312138728324e-05,
-      "loss": 4.9546,
+      "learning_rate": 6.0693641618497105e-05,
+      "loss": 6.9723,
       "step": 3762
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.017341040462428e-05,
-      "loss": 4.6065,
+      "learning_rate": 6.0606936416184966e-05,
+      "loss": 6.7321,
       "step": 3763
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.014450867052023e-05,
-      "loss": 4.7421,
+      "learning_rate": 6.052023121387283e-05,
+      "loss": 6.8748,
       "step": 3764
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.0115606936416188e-05,
-      "loss": 4.5229,
+      "learning_rate": 6.0433526011560686e-05,
+      "loss": 6.7687,
       "step": 3765
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.0086705202312138e-05,
-      "loss": 4.3702,
+      "learning_rate": 6.034682080924855e-05,
+      "loss": 6.7653,
       "step": 3766
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.0057803468208095e-05,
-      "loss": 4.5311,
+      "learning_rate": 6.026011560693641e-05,
+      "loss": 6.9013,
       "step": 3767
     },
     {
       "epoch": 4.22,
-      "learning_rate": 2.0028901734104048e-05,
-      "loss": 3.6107,
+      "learning_rate": 6.0173410404624274e-05,
+      "loss": 6.4894,
       "step": 3768
     },
     {
       "epoch": 4.23,
-      "learning_rate": 2e-05,
-      "loss": 6.7742,
+      "learning_rate": 6.0086705202312134e-05,
+      "loss": 7.0869,
       "step": 3769
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.9971098265895955e-05,
-      "loss": 6.6717,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 6.9647,
       "step": 3770
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.994219653179191e-05,
-      "loss": 6.7112,
+      "learning_rate": 5.9913294797687855e-05,
+      "loss": 6.9477,
       "step": 3771
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.9913294797687862e-05,
-      "loss": 6.6827,
+      "learning_rate": 5.982658959537572e-05,
+      "loss": 6.9717,
       "step": 3772
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.988439306358382e-05,
-      "loss": 6.7247,
+      "learning_rate": 5.9739884393063576e-05,
+      "loss": 6.9811,
       "step": 3773
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.985549132947977e-05,
-      "loss": 6.5046,
+      "learning_rate": 5.965317919075144e-05,
+      "loss": 6.8579,
       "step": 3774
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.9826589595375726e-05,
-      "loss": 6.5177,
+      "learning_rate": 5.95664739884393e-05,
+      "loss": 7.0453,
       "step": 3775
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.9797687861271676e-05,
-      "loss": 6.5889,
+      "learning_rate": 5.9479768786127164e-05,
+      "loss": 7.1022,
       "step": 3776
     },
     {
       "epoch": 4.23,
-      "learning_rate": 1.9768786127167633e-05,
-      "loss": 6.267,
+      "learning_rate": 5.9393063583815024e-05,
+      "loss": 6.8793,
       "step": 3777
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.9739884393063586e-05,
-      "loss": 6.3677,
+      "learning_rate": 5.930635838150289e-05,
+      "loss": 7.0557,
       "step": 3778
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.971098265895954e-05,
-      "loss": 6.0859,
+      "learning_rate": 5.9219653179190745e-05,
+      "loss": 6.8511,
       "step": 3779
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.9682080924855493e-05,
-      "loss": 6.056,
+      "learning_rate": 5.913294797687861e-05,
+      "loss": 6.963,
       "step": 3780
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.9653179190751446e-05,
-      "loss": 6.0011,
+      "learning_rate": 5.9046242774566465e-05,
+      "loss": 7.0057,
       "step": 3781
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.96242774566474e-05,
-      "loss": 5.8206,
+      "learning_rate": 5.895953757225433e-05,
+      "loss": 6.8021,
       "step": 3782
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.9595375722543353e-05,
-      "loss": 5.8119,
+      "learning_rate": 5.887283236994219e-05,
+      "loss": 6.775,
       "step": 3783
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.9566473988439307e-05,
-      "loss": 6.1209,
+      "learning_rate": 5.878612716763006e-05,
+      "loss": 6.9104,
       "step": 3784
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.9537572254335264e-05,
-      "loss": 6.3103,
+      "learning_rate": 5.869942196531791e-05,
+      "loss": 6.9029,
       "step": 3785
     },
     {
       "epoch": 4.24,
-      "learning_rate": 1.9508670520231214e-05,
-      "loss": 6.3862,
+      "learning_rate": 5.861271676300578e-05,
+      "loss": 6.9776,
       "step": 3786
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.947976878612717e-05,
-      "loss": 6.2683,
+      "learning_rate": 5.8526011560693634e-05,
+      "loss": 6.9952,
       "step": 3787
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.9450867052023124e-05,
-      "loss": 6.3144,
+      "learning_rate": 5.84393063583815e-05,
+      "loss": 7.0208,
       "step": 3788
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.9421965317919074e-05,
-      "loss": 6.1327,
+      "learning_rate": 5.835260115606936e-05,
+      "loss": 7.0608,
       "step": 3789
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.939306358381503e-05,
-      "loss": 5.7405,
+      "learning_rate": 5.8265895953757215e-05,
+      "loss": 6.8155,
       "step": 3790
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.936416184971098e-05,
-      "loss": 5.8379,
+      "learning_rate": 5.817919075144508e-05,
+      "loss": 7.0778,
       "step": 3791
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.9335260115606938e-05,
-      "loss": 5.7062,
+      "learning_rate": 5.8092485549132936e-05,
+      "loss": 7.0062,
       "step": 3792
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.930635838150289e-05,
-      "loss": 5.4604,
+      "learning_rate": 5.80057803468208e-05,
+      "loss": 6.8247,
       "step": 3793
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.9277456647398845e-05,
-      "loss": 5.5429,
+      "learning_rate": 5.791907514450866e-05,
+      "loss": 6.8716,
       "step": 3794
     },
     {
       "epoch": 4.25,
-      "learning_rate": 1.9248554913294798e-05,
-      "loss": 5.4875,
+      "learning_rate": 5.783236994219653e-05,
+      "loss": 6.8548,
       "step": 3795
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.921965317919075e-05,
-      "loss": 5.6432,
+      "learning_rate": 5.7745664739884384e-05,
+      "loss": 7.0577,
       "step": 3796
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.9190751445086705e-05,
-      "loss": 5.4127,
+      "learning_rate": 5.765895953757225e-05,
+      "loss": 6.9186,
       "step": 3797
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.916184971098266e-05,
-      "loss": 5.31,
+      "learning_rate": 5.7572254335260105e-05,
+      "loss": 6.8665,
       "step": 3798
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.9132947976878612e-05,
-      "loss": 5.4163,
+      "learning_rate": 5.748554913294797e-05,
+      "loss": 6.9472,
       "step": 3799
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.910404624277457e-05,
-      "loss": 5.4207,
+      "learning_rate": 5.739884393063583e-05,
+      "loss": 6.7832,
       "step": 3800
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.907514450867052e-05,
-      "loss": 5.4633,
+      "learning_rate": 5.731213872832369e-05,
+      "loss": 6.9288,
       "step": 3801
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.9046242774566476e-05,
-      "loss": 5.6844,
+      "learning_rate": 5.722543352601155e-05,
+      "loss": 7.0565,
       "step": 3802
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.901734104046243e-05,
-      "loss": 5.4448,
+      "learning_rate": 5.713872832369942e-05,
+      "loss": 6.9312,
       "step": 3803
     },
     {
       "epoch": 4.26,
-      "learning_rate": 1.8988439306358382e-05,
-      "loss": 5.4582,
+      "learning_rate": 5.705202312138727e-05,
+      "loss": 7.0457,
       "step": 3804
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.8959537572254336e-05,
-      "loss": 5.4259,
+      "learning_rate": 5.696531791907514e-05,
+      "loss": 7.0041,
       "step": 3805
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.893063583815029e-05,
-      "loss": 5.213,
+      "learning_rate": 5.6878612716762994e-05,
+      "loss": 6.8855,
       "step": 3806
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.8901734104046243e-05,
-      "loss": 5.2928,
+      "learning_rate": 5.679190751445086e-05,
+      "loss": 6.9253,
       "step": 3807
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.8872832369942196e-05,
-      "loss": 5.3339,
+      "learning_rate": 5.670520231213872e-05,
+      "loss": 7.0612,
       "step": 3808
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.884393063583815e-05,
-      "loss": 5.0116,
+      "learning_rate": 5.661849710982659e-05,
+      "loss": 6.8557,
       "step": 3809
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.8815028901734106e-05,
-      "loss": 5.3742,
+      "learning_rate": 5.653179190751444e-05,
+      "loss": 7.0709,
       "step": 3810
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.8786127167630057e-05,
-      "loss": 4.8466,
+      "learning_rate": 5.644508670520231e-05,
+      "loss": 6.7385,
       "step": 3811
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.8757225433526013e-05,
-      "loss": 5.1285,
+      "learning_rate": 5.635838150289016e-05,
+      "loss": 6.8919,
       "step": 3812
     },
     {
       "epoch": 4.27,
-      "learning_rate": 1.8728323699421967e-05,
-      "loss": 5.1354,
+      "learning_rate": 5.627167630057803e-05,
+      "loss": 6.9685,
       "step": 3813
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.869942196531792e-05,
-      "loss": 4.6542,
+      "learning_rate": 5.618497109826589e-05,
+      "loss": 6.725,
       "step": 3814
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.8670520231213874e-05,
-      "loss": 4.3565,
+      "learning_rate": 5.609826589595375e-05,
+      "loss": 6.7464,
       "step": 3815
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.8641618497109827e-05,
-      "loss": 4.4044,
+      "learning_rate": 5.601156069364161e-05,
+      "loss": 6.9303,
       "step": 3816
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.861271676300578e-05,
-      "loss": 4.0881,
+      "learning_rate": 5.592485549132948e-05,
+      "loss": 6.6991,
       "step": 3817
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.8583815028901734e-05,
-      "loss": 3.6768,
+      "learning_rate": 5.583815028901733e-05,
+      "loss": 6.4706,
       "step": 3818
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.8554913294797688e-05,
-      "loss": 6.474,
+      "learning_rate": 5.57514450867052e-05,
+      "loss": 7.0148,
       "step": 3819
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.8526011560693644e-05,
-      "loss": 6.3357,
+      "learning_rate": 5.566473988439306e-05,
+      "loss": 6.9101,
       "step": 3820
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.8497109826589594e-05,
-      "loss": 6.2608,
+      "learning_rate": 5.557803468208092e-05,
+      "loss": 6.8649,
       "step": 3821
     },
     {
       "epoch": 4.28,
-      "learning_rate": 1.846820809248555e-05,
-      "loss": 6.2672,
+      "learning_rate": 5.549132947976878e-05,
+      "loss": 6.8908,
       "step": 3822
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.84393063583815e-05,
-      "loss": 6.2217,
+      "learning_rate": 5.540462427745665e-05,
+      "loss": 6.9743,
       "step": 3823
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.8410404624277458e-05,
-      "loss": 5.9319,
+      "learning_rate": 5.53179190751445e-05,
+      "loss": 6.8302,
       "step": 3824
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.838150289017341e-05,
-      "loss": 5.903,
+      "learning_rate": 5.523121387283237e-05,
+      "loss": 6.8531,
       "step": 3825
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.8352601156069365e-05,
-      "loss": 6.0654,
+      "learning_rate": 5.514450867052022e-05,
+      "loss": 7.0209,
       "step": 3826
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.832369942196532e-05,
-      "loss": 5.7196,
+      "learning_rate": 5.505780346820809e-05,
+      "loss": 6.7972,
       "step": 3827
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.8294797687861272e-05,
-      "loss": 5.7025,
+      "learning_rate": 5.497109826589595e-05,
+      "loss": 6.8181,
       "step": 3828
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.8265895953757225e-05,
-      "loss": 5.7505,
+      "learning_rate": 5.488439306358381e-05,
+      "loss": 6.8807,
       "step": 3829
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.8236994219653182e-05,
-      "loss": 5.9735,
+      "learning_rate": 5.479768786127167e-05,
+      "loss": 7.0296,
       "step": 3830
     },
     {
       "epoch": 4.29,
-      "learning_rate": 1.8208092485549132e-05,
-      "loss": 5.9745,
+      "learning_rate": 5.4710982658959536e-05,
+      "loss": 7.0741,
       "step": 3831
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.817919075144509e-05,
-      "loss": 6.0312,
+      "learning_rate": 5.462427745664739e-05,
+      "loss": 7.0098,
       "step": 3832
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.815028901734104e-05,
-      "loss": 6.0525,
+      "learning_rate": 5.453757225433526e-05,
+      "loss": 7.0376,
       "step": 3833
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.8121387283236996e-05,
-      "loss": 5.8772,
+      "learning_rate": 5.445086705202312e-05,
+      "loss": 6.9414,
       "step": 3834
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.809248554913295e-05,
-      "loss": 5.6951,
+      "learning_rate": 5.436416184971098e-05,
+      "loss": 6.9124,
       "step": 3835
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.8063583815028903e-05,
-      "loss": 5.6024,
+      "learning_rate": 5.427745664739884e-05,
+      "loss": 6.9402,
       "step": 3836
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.8034682080924856e-05,
-      "loss": 5.4366,
+      "learning_rate": 5.4190751445086705e-05,
+      "loss": 6.8743,
       "step": 3837
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.800578034682081e-05,
-      "loss": 5.5501,
+      "learning_rate": 5.410404624277456e-05,
+      "loss": 6.9482,
       "step": 3838
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.7976878612716763e-05,
-      "loss": 5.4846,
+      "learning_rate": 5.4017341040462426e-05,
+      "loss": 6.9328,
       "step": 3839
     },
     {
       "epoch": 4.3,
-      "learning_rate": 1.794797687861272e-05,
-      "loss": 5.5299,
+      "learning_rate": 5.393063583815028e-05,
+      "loss": 6.8317,
       "step": 3840
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.791907514450867e-05,
-      "loss": 5.5464,
+      "learning_rate": 5.3843930635838147e-05,
+      "loss": 6.8939,
       "step": 3841
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.7890173410404627e-05,
-      "loss": 5.4894,
+      "learning_rate": 5.375722543352601e-05,
+      "loss": 6.9227,
       "step": 3842
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.7861271676300577e-05,
-      "loss": 5.3768,
+      "learning_rate": 5.367052023121387e-05,
+      "loss": 6.8907,
       "step": 3843
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.7832369942196534e-05,
-      "loss": 5.5695,
+      "learning_rate": 5.358381502890173e-05,
+      "loss": 7.0043,
       "step": 3844
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.7803468208092487e-05,
-      "loss": 5.4174,
+      "learning_rate": 5.3497109826589595e-05,
+      "loss": 6.8566,
       "step": 3845
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.777456647398844e-05,
-      "loss": 5.5527,
+      "learning_rate": 5.341040462427745e-05,
+      "loss": 6.9793,
       "step": 3846
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.7745664739884394e-05,
-      "loss": 5.4385,
+      "learning_rate": 5.3323699421965315e-05,
+      "loss": 6.9138,
       "step": 3847
     },
     {
       "epoch": 4.31,
-      "learning_rate": 1.7716763005780348e-05,
-      "loss": 5.3811,
+      "learning_rate": 5.3236994219653176e-05,
+      "loss": 6.9007,
       "step": 3848
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.76878612716763e-05,
-      "loss": 5.5022,
+      "learning_rate": 5.3150289017341036e-05,
+      "loss": 6.9893,
       "step": 3849
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.7658959537572255e-05,
-      "loss": 5.5565,
+      "learning_rate": 5.3063583815028896e-05,
+      "loss": 7.102,
       "step": 3850
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.7630057803468208e-05,
-      "loss": 5.301,
+      "learning_rate": 5.2976878612716764e-05,
+      "loss": 6.8972,
       "step": 3851
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.7601156069364165e-05,
-      "loss": 5.3661,
+      "learning_rate": 5.289017341040462e-05,
+      "loss": 6.9669,
       "step": 3852
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.7572254335260115e-05,
-      "loss": 5.4404,
+      "learning_rate": 5.2803468208092484e-05,
+      "loss": 6.9638,
       "step": 3853
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.754335260115607e-05,
-      "loss": 5.294,
+      "learning_rate": 5.271676300578034e-05,
+      "loss": 6.9619,
       "step": 3854
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.7514450867052025e-05,
-      "loss": 5.2237,
+      "learning_rate": 5.2630057803468205e-05,
+      "loss": 6.9052,
       "step": 3855
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.748554913294798e-05,
-      "loss": 5.2023,
+      "learning_rate": 5.2543352601156065e-05,
+      "loss": 6.8051,
       "step": 3856
     },
     {
       "epoch": 4.32,
-      "learning_rate": 1.7456647398843932e-05,
-      "loss": 5.179,
+      "learning_rate": 5.245664739884393e-05,
+      "loss": 6.944,
       "step": 3857
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.7427745664739885e-05,
-      "loss": 5.0626,
+      "learning_rate": 5.2369942196531786e-05,
+      "loss": 6.8603,
       "step": 3858
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.739884393063584e-05,
-      "loss": 5.5925,
+      "learning_rate": 5.228323699421965e-05,
+      "loss": 7.1211,
       "step": 3859
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.7369942196531792e-05,
-      "loss": 5.0185,
+      "learning_rate": 5.2196531791907507e-05,
+      "loss": 6.895,
       "step": 3860
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.7341040462427746e-05,
-      "loss": 5.07,
+      "learning_rate": 5.2109826589595374e-05,
+      "loss": 6.8715,
       "step": 3861
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.7312138728323703e-05,
-      "loss": 4.9797,
+      "learning_rate": 5.2023121387283234e-05,
+      "loss": 6.895,
       "step": 3862
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.7283236994219653e-05,
-      "loss": 4.7966,
+      "learning_rate": 5.1936416184971094e-05,
+      "loss": 6.8337,
       "step": 3863
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.725433526011561e-05,
-      "loss": 4.7589,
+      "learning_rate": 5.1849710982658955e-05,
+      "loss": 6.8061,
       "step": 3864
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.722543352601156e-05,
-      "loss": 4.6263,
+      "learning_rate": 5.176300578034682e-05,
+      "loss": 6.6984,
       "step": 3865
     },
     {
       "epoch": 4.33,
-      "learning_rate": 1.7196531791907516e-05,
-      "loss": 4.998,
+      "learning_rate": 5.1676300578034675e-05,
+      "loss": 7.1156,
       "step": 3866
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.716763005780347e-05,
-      "loss": 4.2131,
+      "learning_rate": 5.158959537572254e-05,
+      "loss": 6.727,
       "step": 3867
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.7138728323699423e-05,
-      "loss": 3.9453,
+      "learning_rate": 5.1502890173410396e-05,
+      "loss": 6.619,
       "step": 3868
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.7109826589595377e-05,
-      "loss": 6.7182,
+      "learning_rate": 5.141618497109826e-05,
+      "loss": 7.073,
       "step": 3869
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.708092485549133e-05,
-      "loss": 6.5425,
+      "learning_rate": 5.1329479768786124e-05,
+      "loss": 6.9398,
       "step": 3870
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.7052023121387284e-05,
-      "loss": 6.5493,
+      "learning_rate": 5.124277456647399e-05,
+      "loss": 6.9782,
       "step": 3871
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.702312138728324e-05,
-      "loss": 6.4775,
+      "learning_rate": 5.1156069364161844e-05,
+      "loss": 6.8288,
       "step": 3872
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.699421965317919e-05,
-      "loss": 6.4412,
+      "learning_rate": 5.106936416184971e-05,
+      "loss": 6.9827,
       "step": 3873
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.6965317919075147e-05,
-      "loss": 6.376,
+      "learning_rate": 5.0982658959537565e-05,
+      "loss": 6.9427,
       "step": 3874
     },
     {
       "epoch": 4.34,
-      "learning_rate": 1.6936416184971097e-05,
-      "loss": 6.466,
+      "learning_rate": 5.089595375722543e-05,
+      "loss": 7.0139,
       "step": 3875
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.6907514450867054e-05,
-      "loss": 6.4638,
+      "learning_rate": 5.080924855491329e-05,
+      "loss": 7.104,
       "step": 3876
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.6878612716763008e-05,
-      "loss": 6.1567,
+      "learning_rate": 5.072254335260115e-05,
+      "loss": 6.8795,
       "step": 3877
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.684971098265896e-05,
-      "loss": 6.0722,
+      "learning_rate": 5.063583815028901e-05,
+      "loss": 6.9282,
       "step": 3878
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.6820809248554915e-05,
-      "loss": 6.0263,
+      "learning_rate": 5.054913294797688e-05,
+      "loss": 6.8903,
       "step": 3879
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.6791907514450868e-05,
-      "loss": 6.1255,
+      "learning_rate": 5.0462427745664734e-05,
+      "loss": 7.1448,
       "step": 3880
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.676300578034682e-05,
-      "loss": 5.9037,
+      "learning_rate": 5.03757225433526e-05,
+      "loss": 6.9641,
       "step": 3881
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.673410404624278e-05,
-      "loss": 5.8278,
+      "learning_rate": 5.028901734104046e-05,
+      "loss": 6.8116,
       "step": 3882
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.670520231213873e-05,
-      "loss": 6.0987,
+      "learning_rate": 5.020231213872832e-05,
+      "loss": 6.9393,
       "step": 3883
     },
     {
       "epoch": 4.35,
-      "learning_rate": 1.6676300578034685e-05,
-      "loss": 6.239,
+      "learning_rate": 5.011560693641618e-05,
+      "loss": 6.8925,
       "step": 3884
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.6647398843930635e-05,
-      "loss": 6.2088,
+      "learning_rate": 5.002890173410405e-05,
+      "loss": 6.8705,
       "step": 3885
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.661849710982659e-05,
-      "loss": 6.2986,
+      "learning_rate": 4.99421965317919e-05,
+      "loss": 6.9766,
       "step": 3886
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.6589595375722546e-05,
-      "loss": 6.1682,
+      "learning_rate": 4.985549132947976e-05,
+      "loss": 6.9477,
       "step": 3887
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.6560693641618496e-05,
-      "loss": 6.0649,
+      "learning_rate": 4.976878612716762e-05,
+      "loss": 6.9509,
       "step": 3888
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.6531791907514452e-05,
-      "loss": 5.8488,
+      "learning_rate": 4.9682080924855484e-05,
+      "loss": 6.9704,
       "step": 3889
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.6502890173410403e-05,
-      "loss": 5.5933,
+      "learning_rate": 4.959537572254335e-05,
+      "loss": 6.8596,
       "step": 3890
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.647398843930636e-05,
-      "loss": 5.7102,
+      "learning_rate": 4.9508670520231204e-05,
+      "loss": 6.9811,
       "step": 3891
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.6445086705202313e-05,
-      "loss": 5.6103,
+      "learning_rate": 4.942196531791907e-05,
+      "loss": 6.9081,
       "step": 3892
     },
     {
       "epoch": 4.36,
-      "learning_rate": 1.6416184971098266e-05,
-      "loss": 5.6725,
+      "learning_rate": 4.9335260115606925e-05,
+      "loss": 6.9361,
       "step": 3893
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.638728323699422e-05,
-      "loss": 5.7005,
+      "learning_rate": 4.924855491329479e-05,
+      "loss": 7.0107,
       "step": 3894
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.6358381502890173e-05,
-      "loss": 5.6551,
+      "learning_rate": 4.916184971098265e-05,
+      "loss": 6.9704,
       "step": 3895
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.6329479768786127e-05,
-      "loss": 5.4875,
+      "learning_rate": 4.907514450867052e-05,
+      "loss": 6.8489,
       "step": 3896
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.6300578034682083e-05,
-      "loss": 5.4165,
+      "learning_rate": 4.898843930635837e-05,
+      "loss": 6.9059,
       "step": 3897
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.6271676300578033e-05,
-      "loss": 5.4071,
+      "learning_rate": 4.890173410404624e-05,
+      "loss": 6.9522,
       "step": 3898
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.624277456647399e-05,
-      "loss": 5.4265,
+      "learning_rate": 4.8815028901734094e-05,
+      "loss": 6.9117,
       "step": 3899
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.621387283236994e-05,
-      "loss": 5.3019,
+      "learning_rate": 4.872832369942196e-05,
+      "loss": 6.8779,
       "step": 3900
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.6184971098265897e-05,
-      "loss": 5.5302,
+      "learning_rate": 4.864161849710982e-05,
+      "loss": 6.9868,
       "step": 3901
     },
     {
       "epoch": 4.37,
-      "learning_rate": 1.615606936416185e-05,
-      "loss": 5.4324,
+      "learning_rate": 4.855491329479768e-05,
+      "loss": 6.9545,
       "step": 3902
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.6127167630057804e-05,
-      "loss": 5.5549,
+      "learning_rate": 4.846820809248554e-05,
+      "loss": 7.0444,
       "step": 3903
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.6098265895953758e-05,
-      "loss": 5.4056,
+      "learning_rate": 4.838150289017341e-05,
+      "loss": 6.9886,
       "step": 3904
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.606936416184971e-05,
-      "loss": 5.3873,
+      "learning_rate": 4.829479768786126e-05,
+      "loss": 7.0432,
       "step": 3905
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.6040462427745664e-05,
-      "loss": 5.436,
+      "learning_rate": 4.820809248554913e-05,
+      "loss": 7.0082,
       "step": 3906
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.601156069364162e-05,
-      "loss": 5.1523,
+      "learning_rate": 4.812138728323699e-05,
+      "loss": 6.8373,
       "step": 3907
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.598265895953757e-05,
-      "loss": 5.323,
+      "learning_rate": 4.803468208092485e-05,
+      "loss": 6.9962,
       "step": 3908
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.5953757225433528e-05,
-      "loss": 5.177,
+      "learning_rate": 4.794797687861271e-05,
+      "loss": 6.9665,
       "step": 3909
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.5924855491329478e-05,
-      "loss": 5.2651,
+      "learning_rate": 4.786127167630058e-05,
+      "loss": 7.062,
       "step": 3910
     },
     {
       "epoch": 4.38,
-      "learning_rate": 1.5895953757225435e-05,
-      "loss": 5.2872,
+      "learning_rate": 4.777456647398843e-05,
+      "loss": 6.9639,
       "step": 3911
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.586705202312139e-05,
-      "loss": 5.0614,
+      "learning_rate": 4.76878612716763e-05,
+      "loss": 6.9225,
       "step": 3912
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.5838150289017342e-05,
-      "loss": 4.6807,
+      "learning_rate": 4.760115606936415e-05,
+      "loss": 6.6616,
       "step": 3913
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.5809248554913295e-05,
-      "loss": 4.8296,
+      "learning_rate": 4.751445086705202e-05,
+      "loss": 6.9932,
       "step": 3914
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.578034682080925e-05,
-      "loss": 4.8219,
+      "learning_rate": 4.742774566473988e-05,
+      "loss": 7.0079,
       "step": 3915
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.5751445086705202e-05,
-      "loss": 4.5377,
+      "learning_rate": 4.734104046242774e-05,
+      "loss": 6.8155,
       "step": 3916
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.5722543352601156e-05,
-      "loss": 4.139,
+      "learning_rate": 4.72543352601156e-05,
+      "loss": 6.6108,
       "step": 3917
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.569364161849711e-05,
-      "loss": 3.8552,
+      "learning_rate": 4.716763005780347e-05,
+      "loss": 6.5826,
       "step": 3918
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.5664739884393066e-05,
-      "loss": 6.5087,
+      "learning_rate": 4.708092485549132e-05,
+      "loss": 7.0252,
       "step": 3919
     },
     {
       "epoch": 4.39,
-      "learning_rate": 1.5635838150289016e-05,
-      "loss": 6.3663,
+      "learning_rate": 4.699421965317919e-05,
+      "loss": 6.9098,
       "step": 3920
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.5606936416184973e-05,
-      "loss": 6.1723,
+      "learning_rate": 4.690751445086705e-05,
+      "loss": 6.7253,
       "step": 3921
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.5578034682080926e-05,
-      "loss": 6.4023,
+      "learning_rate": 4.682080924855491e-05,
+      "loss": 6.9311,
       "step": 3922
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.554913294797688e-05,
-      "loss": 6.3003,
+      "learning_rate": 4.673410404624277e-05,
+      "loss": 6.9691,
       "step": 3923
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.5520231213872833e-05,
-      "loss": 6.0945,
+      "learning_rate": 4.6647398843930636e-05,
+      "loss": 6.8952,
       "step": 3924
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.5491329479768787e-05,
-      "loss": 6.1131,
+      "learning_rate": 4.656069364161849e-05,
+      "loss": 6.9512,
       "step": 3925
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.546242774566474e-05,
-      "loss": 6.0517,
+      "learning_rate": 4.647398843930636e-05,
+      "loss": 7.05,
       "step": 3926
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.5433526011560694e-05,
-      "loss": 5.8191,
+      "learning_rate": 4.638728323699421e-05,
+      "loss": 6.8654,
       "step": 3927
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.5404624277456647e-05,
-      "loss": 5.8689,
+      "learning_rate": 4.630057803468208e-05,
+      "loss": 7.0599,
       "step": 3928
     },
     {
       "epoch": 4.4,
-      "learning_rate": 1.5375722543352604e-05,
-      "loss": 5.562,
+      "learning_rate": 4.621387283236994e-05,
+      "loss": 6.781,
       "step": 3929
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.5346820809248554e-05,
-      "loss": 5.6994,
+      "learning_rate": 4.61271676300578e-05,
+      "loss": 6.8743,
       "step": 3930
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.531791907514451e-05,
-      "loss": 5.8369,
+      "learning_rate": 4.604046242774566e-05,
+      "loss": 6.9778,
       "step": 3931
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.528901734104046e-05,
-      "loss": 5.8143,
+      "learning_rate": 4.5953757225433526e-05,
+      "loss": 6.8515,
       "step": 3932
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.5260115606936418e-05,
-      "loss": 5.8222,
+      "learning_rate": 4.586705202312138e-05,
+      "loss": 6.9158,
       "step": 3933
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.523121387283237e-05,
-      "loss": 5.9076,
+      "learning_rate": 4.5780346820809246e-05,
+      "loss": 6.9327,
       "step": 3934
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.5202312138728325e-05,
-      "loss": 5.811,
+      "learning_rate": 4.5693641618497107e-05,
+      "loss": 6.8504,
       "step": 3935
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.5173410404624278e-05,
-      "loss": 5.7064,
+      "learning_rate": 4.560693641618497e-05,
+      "loss": 6.8653,
       "step": 3936
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.5144508670520233e-05,
-      "loss": 5.5008,
+      "learning_rate": 4.552023121387283e-05,
+      "loss": 6.7296,
       "step": 3937
     },
     {
       "epoch": 4.41,
-      "learning_rate": 1.5115606936416185e-05,
-      "loss": 5.768,
+      "learning_rate": 4.5433526011560694e-05,
+      "loss": 7.0079,
       "step": 3938
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.508670520231214e-05,
-      "loss": 5.5831,
+      "learning_rate": 4.534682080924855e-05,
+      "loss": 6.9927,
       "step": 3939
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.5057803468208092e-05,
-      "loss": 5.4458,
+      "learning_rate": 4.5260115606936415e-05,
+      "loss": 6.941,
       "step": 3940
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.5028901734104049e-05,
-      "loss": 5.6271,
+      "learning_rate": 4.517341040462427e-05,
+      "loss": 7.0726,
       "step": 3941
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.5e-05,
-      "loss": 5.5425,
+      "learning_rate": 4.5086705202312136e-05,
+      "loss": 6.9614,
       "step": 3942
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.4971098265895956e-05,
-      "loss": 5.6004,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 7.0155,
       "step": 3943
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.4942196531791907e-05,
-      "loss": 5.4829,
+      "learning_rate": 4.491329479768786e-05,
+      "loss": 7.012,
       "step": 3944
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.4913294797687862e-05,
-      "loss": 5.4784,
+      "learning_rate": 4.482658959537572e-05,
+      "loss": 7.0444,
       "step": 3945
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.4884393063583816e-05,
-      "loss": 5.5535,
+      "learning_rate": 4.4739884393063584e-05,
+      "loss": 7.0343,
       "step": 3946
     },
     {
       "epoch": 4.42,
-      "learning_rate": 1.4855491329479771e-05,
-      "loss": 5.359,
+      "learning_rate": 4.465317919075144e-05,
+      "loss": 6.8973,
       "step": 3947
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.4826589595375723e-05,
-      "loss": 5.374,
+      "learning_rate": 4.4566473988439305e-05,
+      "loss": 6.8725,
       "step": 3948
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.4797687861271678e-05,
-      "loss": 5.5766,
+      "learning_rate": 4.4479768786127165e-05,
+      "loss": 6.9846,
       "step": 3949
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.476878612716763e-05,
-      "loss": 5.3083,
+      "learning_rate": 4.4393063583815025e-05,
+      "loss": 6.8283,
       "step": 3950
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.4739884393063585e-05,
-      "loss": 5.3545,
+      "learning_rate": 4.4306358381502886e-05,
+      "loss": 6.8858,
       "step": 3951
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.4710982658959538e-05,
-      "loss": 5.3207,
+      "learning_rate": 4.421965317919075e-05,
+      "loss": 6.8764,
       "step": 3952
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.4682080924855493e-05,
-      "loss": 5.3442,
+      "learning_rate": 4.4132947976878606e-05,
+      "loss": 7.0919,
       "step": 3953
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.4653179190751445e-05,
-      "loss": 5.4448,
+      "learning_rate": 4.404624277456647e-05,
+      "loss": 7.0646,
       "step": 3954
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.46242774566474e-05,
-      "loss": 5.4705,
+      "learning_rate": 4.395953757225433e-05,
+      "loss": 7.0863,
       "step": 3955
     },
     {
       "epoch": 4.43,
-      "learning_rate": 1.4595375722543354e-05,
-      "loss": 5.3715,
+      "learning_rate": 4.3872832369942194e-05,
+      "loss": 7.0719,
       "step": 3956
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.4566473988439309e-05,
-      "loss": 5.1269,
+      "learning_rate": 4.3786127167630054e-05,
+      "loss": 6.9073,
       "step": 3957
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.453757225433526e-05,
-      "loss": 5.1202,
+      "learning_rate": 4.369942196531792e-05,
+      "loss": 6.9129,
       "step": 3958
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.4508670520231216e-05,
-      "loss": 5.202,
+      "learning_rate": 4.3612716763005775e-05,
+      "loss": 6.9198,
       "step": 3959
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.4479768786127167e-05,
-      "loss": 5.2885,
+      "learning_rate": 4.352601156069364e-05,
+      "loss": 6.943,
       "step": 3960
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.4450867052023123e-05,
-      "loss": 5.0152,
+      "learning_rate": 4.3439306358381496e-05,
+      "loss": 6.7849,
       "step": 3961
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.4421965317919076e-05,
-      "loss": 5.1744,
+      "learning_rate": 4.335260115606936e-05,
+      "loss": 6.8842,
       "step": 3962
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.4393063583815031e-05,
-      "loss": 4.8562,
+      "learning_rate": 4.326589595375722e-05,
+      "loss": 6.9027,
       "step": 3963
     },
     {
       "epoch": 4.44,
-      "learning_rate": 1.4364161849710983e-05,
-      "loss": 4.7809,
+      "learning_rate": 4.3179190751445084e-05,
+      "loss": 6.7534,
       "step": 3964
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.4335260115606938e-05,
-      "loss": 4.6839,
+      "learning_rate": 4.3092485549132944e-05,
+      "loss": 6.882,
       "step": 3965
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.430635838150289e-05,
-      "loss": 4.3062,
+      "learning_rate": 4.300578034682081e-05,
+      "loss": 6.7657,
       "step": 3966
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.4277456647398845e-05,
-      "loss": 4.178,
+      "learning_rate": 4.2919075144508665e-05,
+      "loss": 6.7116,
       "step": 3967
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.4248554913294798e-05,
-      "loss": 3.9096,
+      "learning_rate": 4.283236994219653e-05,
+      "loss": 6.5509,
       "step": 3968
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.4219653179190754e-05,
-      "loss": 6.5728,
+      "learning_rate": 4.274566473988439e-05,
+      "loss": 6.9767,
       "step": 3969
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.4190751445086705e-05,
-      "loss": 6.557,
+      "learning_rate": 4.265895953757225e-05,
+      "loss": 6.9807,
       "step": 3970
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.416184971098266e-05,
-      "loss": 6.4671,
+      "learning_rate": 4.257225433526011e-05,
+      "loss": 6.9568,
       "step": 3971
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.4132947976878614e-05,
-      "loss": 6.2769,
+      "learning_rate": 4.248554913294798e-05,
+      "loss": 6.8426,
       "step": 3972
     },
     {
       "epoch": 4.45,
-      "learning_rate": 1.4104046242774569e-05,
-      "loss": 6.3879,
+      "learning_rate": 4.239884393063583e-05,
+      "loss": 7.0399,
       "step": 3973
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.407514450867052e-05,
-      "loss": 6.2884,
+      "learning_rate": 4.23121387283237e-05,
+      "loss": 6.9232,
       "step": 3974
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.4046242774566476e-05,
-      "loss": 6.2003,
+      "learning_rate": 4.2225433526011554e-05,
+      "loss": 6.8981,
       "step": 3975
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.4017341040462428e-05,
-      "loss": 6.1475,
+      "learning_rate": 4.213872832369942e-05,
+      "loss": 6.9327,
       "step": 3976
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.3988439306358383e-05,
-      "loss": 6.1828,
+      "learning_rate": 4.205202312138728e-05,
+      "loss": 6.9938,
       "step": 3977
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.3959537572254336e-05,
-      "loss": 6.0049,
+      "learning_rate": 4.196531791907514e-05,
+      "loss": 6.9433,
       "step": 3978
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.3930635838150291e-05,
-      "loss": 5.7772,
+      "learning_rate": 4.1878612716763e-05,
+      "loss": 6.6889,
       "step": 3979
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.3901734104046243e-05,
-      "loss": 5.9507,
+      "learning_rate": 4.179190751445087e-05,
+      "loss": 6.9164,
       "step": 3980
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.3872832369942197e-05,
-      "loss": 5.7756,
+      "learning_rate": 4.170520231213872e-05,
+      "loss": 6.8616,
       "step": 3981
     },
     {
       "epoch": 4.46,
-      "learning_rate": 1.3843930635838152e-05,
-      "loss": 5.7847,
+      "learning_rate": 4.161849710982658e-05,
+      "loss": 6.8743,
       "step": 3982
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3815028901734104e-05,
-      "loss": 5.8256,
+      "learning_rate": 4.153179190751445e-05,
+      "loss": 6.9648,
       "step": 3983
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3786127167630059e-05,
-      "loss": 5.7439,
+      "learning_rate": 4.1445086705202304e-05,
+      "loss": 6.8351,
       "step": 3984
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.375722543352601e-05,
-      "loss": 5.9492,
+      "learning_rate": 4.135838150289017e-05,
+      "loss": 6.8895,
       "step": 3985
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3728323699421966e-05,
-      "loss": 5.8044,
+      "learning_rate": 4.1271676300578025e-05,
+      "loss": 6.9309,
       "step": 3986
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3699421965317919e-05,
-      "loss": 5.8837,
+      "learning_rate": 4.118497109826589e-05,
+      "loss": 6.9754,
       "step": 3987
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3670520231213874e-05,
-      "loss": 5.8731,
+      "learning_rate": 4.109826589595375e-05,
+      "loss": 7.0924,
       "step": 3988
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3641618497109826e-05,
-      "loss": 5.6193,
+      "learning_rate": 4.101156069364161e-05,
+      "loss": 6.9607,
       "step": 3989
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3612716763005781e-05,
-      "loss": 5.6381,
+      "learning_rate": 4.092485549132947e-05,
+      "loss": 6.9748,
       "step": 3990
     },
     {
       "epoch": 4.47,
-      "learning_rate": 1.3583815028901733e-05,
-      "loss": 5.4619,
+      "learning_rate": 4.083815028901734e-05,
+      "loss": 6.8684,
       "step": 3991
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3554913294797688e-05,
-      "loss": 5.4587,
+      "learning_rate": 4.075144508670519e-05,
+      "loss": 6.8884,
       "step": 3992
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3526011560693641e-05,
-      "loss": 5.4906,
+      "learning_rate": 4.066473988439306e-05,
+      "loss": 6.8941,
       "step": 3993
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3497109826589597e-05,
-      "loss": 5.5844,
+      "learning_rate": 4.057803468208092e-05,
+      "loss": 6.9784,
       "step": 3994
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3468208092485548e-05,
-      "loss": 5.512,
+      "learning_rate": 4.049132947976878e-05,
+      "loss": 6.9531,
       "step": 3995
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3439306358381503e-05,
-      "loss": 5.5035,
+      "learning_rate": 4.040462427745664e-05,
+      "loss": 6.9605,
       "step": 3996
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3410404624277457e-05,
-      "loss": 5.3505,
+      "learning_rate": 4.031791907514451e-05,
+      "loss": 6.904,
       "step": 3997
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3381502890173412e-05,
-      "loss": 5.3836,
+      "learning_rate": 4.023121387283236e-05,
+      "loss": 6.9335,
       "step": 3998
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3352601156069364e-05,
-      "loss": 5.1755,
+      "learning_rate": 4.014450867052023e-05,
+      "loss": 6.7267,
       "step": 3999
     },
     {
       "epoch": 4.48,
-      "learning_rate": 1.3323699421965319e-05,
-      "loss": 5.5255,
+      "learning_rate": 4.005780346820808e-05,
+      "loss": 7.0027,
       "step": 4000
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.329479768786127e-05,
-      "loss": 5.4752,
+      "learning_rate": 3.997109826589595e-05,
+      "loss": 6.9353,
       "step": 4001
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.3265895953757226e-05,
-      "loss": 5.3746,
+      "learning_rate": 3.988439306358381e-05,
+      "loss": 6.8664,
       "step": 4002
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.323699421965318e-05,
-      "loss": 5.6287,
+      "learning_rate": 3.979768786127167e-05,
+      "loss": 7.0066,
       "step": 4003
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.3208092485549134e-05,
-      "loss": 5.2426,
+      "learning_rate": 3.971098265895953e-05,
+      "loss": 6.7601,
       "step": 4004
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.3179190751445086e-05,
-      "loss": 5.3168,
+      "learning_rate": 3.96242774566474e-05,
+      "loss": 6.9296,
       "step": 4005
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.3150289017341041e-05,
-      "loss": 5.1261,
+      "learning_rate": 3.953757225433525e-05,
+      "loss": 6.8955,
       "step": 4006
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.3121387283236993e-05,
-      "loss": 5.2652,
+      "learning_rate": 3.945086705202312e-05,
+      "loss": 7.0104,
       "step": 4007
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.3092485549132948e-05,
-      "loss": 5.2561,
+      "learning_rate": 3.936416184971098e-05,
+      "loss": 6.9958,
       "step": 4008
     },
     {
       "epoch": 4.49,
-      "learning_rate": 1.3063583815028902e-05,
-      "loss": 5.1966,
+      "learning_rate": 3.927745664739884e-05,
+      "loss": 7.0173,
       "step": 4009
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.3034682080924857e-05,
-      "loss": 4.8879,
+      "learning_rate": 3.91907514450867e-05,
+      "loss": 6.8247,
       "step": 4010
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.3005780346820809e-05,
-      "loss": 5.0079,
+      "learning_rate": 3.910404624277457e-05,
+      "loss": 6.8927,
       "step": 4011
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.2976878612716764e-05,
-      "loss": 5.0487,
+      "learning_rate": 3.901734104046242e-05,
+      "loss": 7.0302,
       "step": 4012
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.2947976878612717e-05,
-      "loss": 4.574,
+      "learning_rate": 3.893063583815029e-05,
+      "loss": 6.7211,
       "step": 4013
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.2919075144508672e-05,
-      "loss": 4.4593,
+      "learning_rate": 3.884393063583814e-05,
+      "loss": 6.6907,
       "step": 4014
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.2890173410404624e-05,
-      "loss": 4.3536,
+      "learning_rate": 3.875722543352601e-05,
+      "loss": 6.7469,
       "step": 4015
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.2861271676300579e-05,
-      "loss": 4.0989,
+      "learning_rate": 3.867052023121387e-05,
+      "loss": 6.7272,
       "step": 4016
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.2832369942196531e-05,
-      "loss": 3.8103,
+      "learning_rate": 3.858381502890173e-05,
+      "loss": 6.5714,
       "step": 4017
     },
     {
       "epoch": 4.5,
-      "learning_rate": 1.2803468208092486e-05,
-      "loss": 3.773,
+      "learning_rate": 3.849710982658959e-05,
+      "loss": 6.5729,
       "step": 4018
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.277456647398844e-05,
-      "loss": 6.4156,
+      "learning_rate": 3.8410404624277456e-05,
+      "loss": 6.9027,
       "step": 4019
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.2745664739884395e-05,
-      "loss": 6.4902,
+      "learning_rate": 3.832369942196531e-05,
+      "loss": 7.054,
       "step": 4020
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.2716763005780346e-05,
-      "loss": 6.5281,
+      "learning_rate": 3.823699421965318e-05,
+      "loss": 7.0606,
       "step": 4021
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.2687861271676302e-05,
-      "loss": 6.4017,
+      "learning_rate": 3.815028901734104e-05,
+      "loss": 6.9001,
       "step": 4022
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.2658959537572255e-05,
-      "loss": 6.3418,
+      "learning_rate": 3.80635838150289e-05,
+      "loss": 6.8978,
       "step": 4023
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.263005780346821e-05,
-      "loss": 6.3027,
+      "learning_rate": 3.797687861271676e-05,
+      "loss": 6.8662,
       "step": 4024
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.2601156069364162e-05,
-      "loss": 6.2963,
+      "learning_rate": 3.7890173410404625e-05,
+      "loss": 7.0456,
       "step": 4025
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.2572254335260117e-05,
-      "loss": 6.1172,
+      "learning_rate": 3.780346820809248e-05,
+      "loss": 6.971,
       "step": 4026
     },
     {
       "epoch": 4.51,
-      "learning_rate": 1.2543352601156069e-05,
-      "loss": 5.9734,
+      "learning_rate": 3.7716763005780346e-05,
+      "loss": 6.8992,
       "step": 4027
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.2514450867052024e-05,
-      "loss": 5.7865,
+      "learning_rate": 3.76300578034682e-05,
+      "loss": 6.9002,
       "step": 4028
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.2485549132947977e-05,
-      "loss": 5.6941,
+      "learning_rate": 3.7543352601156067e-05,
+      "loss": 6.8627,
       "step": 4029
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.245664739884393e-05,
-      "loss": 5.5558,
+      "learning_rate": 3.745664739884393e-05,
+      "loss": 6.7664,
       "step": 4030
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.2427745664739884e-05,
-      "loss": 5.8495,
+      "learning_rate": 3.736994219653179e-05,
+      "loss": 7.0375,
       "step": 4031
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.2398843930635838e-05,
-      "loss": 5.8519,
+      "learning_rate": 3.728323699421965e-05,
+      "loss": 7.0272,
       "step": 4032
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.2369942196531791e-05,
-      "loss": 5.8558,
+      "learning_rate": 3.719653179190751e-05,
+      "loss": 6.9641,
       "step": 4033
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.2341040462427746e-05,
-      "loss": 5.789,
+      "learning_rate": 3.710982658959537e-05,
+      "loss": 7.0071,
       "step": 4034
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.23121387283237e-05,
-      "loss": 5.6982,
+      "learning_rate": 3.702312138728323e-05,
+      "loss": 6.801,
       "step": 4035
     },
     {
       "epoch": 4.52,
-      "learning_rate": 1.2283236994219653e-05,
-      "loss": 5.8928,
+      "learning_rate": 3.6936416184971096e-05,
+      "loss": 6.9349,
       "step": 4036
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2254335260115607e-05,
-      "loss": 5.7963,
+      "learning_rate": 3.6849710982658956e-05,
+      "loss": 6.9221,
       "step": 4037
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.222543352601156e-05,
-      "loss": 5.6913,
+      "learning_rate": 3.6763005780346816e-05,
+      "loss": 6.8913,
       "step": 4038
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2196531791907515e-05,
-      "loss": 5.8401,
+      "learning_rate": 3.667630057803468e-05,
+      "loss": 7.0869,
       "step": 4039
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2167630057803469e-05,
-      "loss": 5.5354,
+      "learning_rate": 3.658959537572254e-05,
+      "loss": 6.8962,
       "step": 4040
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2138728323699422e-05,
-      "loss": 5.4835,
+      "learning_rate": 3.65028901734104e-05,
+      "loss": 6.8734,
       "step": 4041
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2109826589595376e-05,
-      "loss": 5.5481,
+      "learning_rate": 3.6416184971098265e-05,
+      "loss": 7.0143,
       "step": 4042
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2080924855491329e-05,
-      "loss": 5.4919,
+      "learning_rate": 3.6329479768786125e-05,
+      "loss": 6.9782,
       "step": 4043
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2052023121387284e-05,
-      "loss": 5.4883,
+      "learning_rate": 3.6242774566473985e-05,
+      "loss": 6.9124,
       "step": 4044
     },
     {
       "epoch": 4.53,
-      "learning_rate": 1.2023121387283238e-05,
-      "loss": 5.5114,
+      "learning_rate": 3.6156069364161846e-05,
+      "loss": 6.9528,
       "step": 4045
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.1994219653179191e-05,
-      "loss": 5.6038,
+      "learning_rate": 3.6069364161849706e-05,
+      "loss": 6.962,
       "step": 4046
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.1965317919075144e-05,
-      "loss": 5.4006,
+      "learning_rate": 3.5982658959537566e-05,
+      "loss": 6.9516,
       "step": 4047
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.1936416184971098e-05,
-      "loss": 5.5422,
+      "learning_rate": 3.5895953757225427e-05,
+      "loss": 7.0165,
       "step": 4048
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.1907514450867053e-05,
-      "loss": 5.5759,
+      "learning_rate": 3.5809248554913294e-05,
+      "loss": 7.0736,
       "step": 4049
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.1878612716763006e-05,
-      "loss": 5.2335,
+      "learning_rate": 3.5722543352601154e-05,
+      "loss": 6.9381,
       "step": 4050
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.184971098265896e-05,
-      "loss": 5.2275,
+      "learning_rate": 3.5635838150289014e-05,
+      "loss": 6.8931,
       "step": 4051
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.1820809248554913e-05,
-      "loss": 5.5213,
+      "learning_rate": 3.5549132947976875e-05,
+      "loss": 7.0175,
       "step": 4052
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.1791907514450867e-05,
-      "loss": 5.48,
+      "learning_rate": 3.5462427745664735e-05,
+      "loss": 7.0837,
       "step": 4053
     },
     {
       "epoch": 4.54,
-      "learning_rate": 1.176300578034682e-05,
-      "loss": 5.2681,
+      "learning_rate": 3.5375722543352595e-05,
+      "loss": 6.9377,
       "step": 4054
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1734104046242775e-05,
-      "loss": 5.4972,
+      "learning_rate": 3.5289017341040456e-05,
+      "loss": 7.1011,
       "step": 4055
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1705202312138729e-05,
-      "loss": 5.303,
+      "learning_rate": 3.520231213872832e-05,
+      "loss": 7.0245,
       "step": 4056
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1676300578034682e-05,
-      "loss": 4.9525,
+      "learning_rate": 3.511560693641618e-05,
+      "loss": 6.8254,
       "step": 4057
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1647398843930636e-05,
-      "loss": 5.2565,
+      "learning_rate": 3.5028901734104043e-05,
+      "loss": 7.0416,
       "step": 4058
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.161849710982659e-05,
-      "loss": 5.1572,
+      "learning_rate": 3.4942196531791904e-05,
+      "loss": 6.9918,
       "step": 4059
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1589595375722544e-05,
-      "loss": 4.9928,
+      "learning_rate": 3.4855491329479764e-05,
+      "loss": 6.82,
       "step": 4060
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1560693641618498e-05,
-      "loss": 5.1569,
+      "learning_rate": 3.4768786127167625e-05,
+      "loss": 7.03,
       "step": 4061
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1531791907514451e-05,
-      "loss": 4.9778,
+      "learning_rate": 3.4682080924855485e-05,
+      "loss": 6.9133,
       "step": 4062
     },
     {
       "epoch": 4.55,
-      "learning_rate": 1.1502890173410405e-05,
-      "loss": 5.0956,
+      "learning_rate": 3.459537572254335e-05,
+      "loss": 6.9957,
       "step": 4063
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.1473988439306358e-05,
-      "loss": 4.6968,
+      "learning_rate": 3.450867052023121e-05,
+      "loss": 6.8475,
       "step": 4064
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.1445086705202313e-05,
-      "loss": 4.6806,
+      "learning_rate": 3.442196531791907e-05,
+      "loss": 6.8852,
       "step": 4065
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.1416184971098267e-05,
-      "loss": 4.3921,
+      "learning_rate": 3.433526011560693e-05,
+      "loss": 6.8975,
       "step": 4066
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.138728323699422e-05,
-      "loss": 4.3549,
+      "learning_rate": 3.424855491329479e-05,
+      "loss": 6.8318,
       "step": 4067
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.1358381502890174e-05,
-      "loss": 3.6633,
+      "learning_rate": 3.4161849710982654e-05,
+      "loss": 6.4837,
       "step": 4068
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.1329479768786127e-05,
-      "loss": 6.5995,
+      "learning_rate": 3.4075144508670514e-05,
+      "loss": 7.0552,
       "step": 4069
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.1300578034682082e-05,
-      "loss": 6.3783,
+      "learning_rate": 3.398843930635838e-05,
+      "loss": 6.9675,
       "step": 4070
     },
     {
       "epoch": 4.56,
-      "learning_rate": 1.1271676300578036e-05,
-      "loss": 6.3674,
+      "learning_rate": 3.390173410404624e-05,
+      "loss": 7.0252,
       "step": 4071
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1242774566473989e-05,
-      "loss": 6.4261,
+      "learning_rate": 3.38150289017341e-05,
+      "loss": 6.9048,
       "step": 4072
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1213872832369943e-05,
-      "loss": 6.3743,
+      "learning_rate": 3.372832369942196e-05,
+      "loss": 6.976,
       "step": 4073
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1184971098265896e-05,
-      "loss": 6.3163,
+      "learning_rate": 3.364161849710982e-05,
+      "loss": 6.9116,
       "step": 4074
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.115606936416185e-05,
-      "loss": 6.2462,
+      "learning_rate": 3.355491329479768e-05,
+      "loss": 6.9677,
       "step": 4075
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1127167630057805e-05,
-      "loss": 6.1376,
+      "learning_rate": 3.346820809248554e-05,
+      "loss": 6.9165,
       "step": 4076
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1098265895953758e-05,
-      "loss": 6.2671,
+      "learning_rate": 3.338150289017341e-05,
+      "loss": 7.0684,
       "step": 4077
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1069364161849711e-05,
-      "loss": 6.0616,
+      "learning_rate": 3.329479768786127e-05,
+      "loss": 6.9026,
       "step": 4078
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1040462427745665e-05,
-      "loss": 5.9397,
+      "learning_rate": 3.320809248554913e-05,
+      "loss": 6.8761,
       "step": 4079
     },
     {
       "epoch": 4.57,
-      "learning_rate": 1.1011560693641618e-05,
-      "loss": 5.8794,
+      "learning_rate": 3.312138728323699e-05,
+      "loss": 6.8633,
       "step": 4080
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.0982658959537573e-05,
-      "loss": 5.9453,
+      "learning_rate": 3.303468208092485e-05,
+      "loss": 6.9874,
       "step": 4081
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.0953757225433527e-05,
-      "loss": 5.6563,
+      "learning_rate": 3.294797687861271e-05,
+      "loss": 6.8464,
       "step": 4082
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.092485549132948e-05,
-      "loss": 5.6894,
+      "learning_rate": 3.286127167630057e-05,
+      "loss": 6.9148,
       "step": 4083
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.0895953757225434e-05,
-      "loss": 5.7794,
+      "learning_rate": 3.277456647398844e-05,
+      "loss": 6.862,
       "step": 4084
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.0867052023121387e-05,
-      "loss": 5.9632,
+      "learning_rate": 3.26878612716763e-05,
+      "loss": 6.9831,
       "step": 4085
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.0838150289017342e-05,
-      "loss": 5.8421,
+      "learning_rate": 3.260115606936416e-05,
+      "loss": 6.909,
       "step": 4086
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.0809248554913296e-05,
-      "loss": 5.8056,
+      "learning_rate": 3.251445086705202e-05,
+      "loss": 6.8574,
       "step": 4087
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.078034682080925e-05,
-      "loss": 5.6457,
+      "learning_rate": 3.242774566473988e-05,
+      "loss": 6.8205,
       "step": 4088
     },
     {
       "epoch": 4.58,
-      "learning_rate": 1.0751445086705203e-05,
-      "loss": 5.8681,
+      "learning_rate": 3.234104046242774e-05,
+      "loss": 7.0914,
       "step": 4089
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0722543352601156e-05,
-      "loss": 5.7457,
+      "learning_rate": 3.22543352601156e-05,
+      "loss": 6.916,
       "step": 4090
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0693641618497111e-05,
-      "loss": 5.7093,
+      "learning_rate": 3.216763005780347e-05,
+      "loss": 6.9778,
       "step": 4091
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0664739884393065e-05,
-      "loss": 5.5667,
+      "learning_rate": 3.208092485549133e-05,
+      "loss": 7.0409,
       "step": 4092
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0635838150289018e-05,
-      "loss": 5.311,
+      "learning_rate": 3.199421965317919e-05,
+      "loss": 6.8453,
       "step": 4093
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0606936416184972e-05,
-      "loss": 5.39,
+      "learning_rate": 3.190751445086705e-05,
+      "loss": 6.9046,
       "step": 4094
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0578034682080925e-05,
-      "loss": 5.5124,
+      "learning_rate": 3.182080924855491e-05,
+      "loss": 6.9794,
       "step": 4095
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0549132947976879e-05,
-      "loss": 5.3304,
+      "learning_rate": 3.173410404624277e-05,
+      "loss": 6.8972,
       "step": 4096
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0520231213872834e-05,
-      "loss": 5.465,
+      "learning_rate": 3.164739884393063e-05,
+      "loss": 6.988,
       "step": 4097
     },
     {
       "epoch": 4.59,
-      "learning_rate": 1.0491329479768787e-05,
-      "loss": 5.4178,
+      "learning_rate": 3.15606936416185e-05,
+      "loss": 6.9718,
       "step": 4098
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.046242774566474e-05,
-      "loss": 5.4314,
+      "learning_rate": 3.147398843930636e-05,
+      "loss": 6.9707,
       "step": 4099
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.0433526011560694e-05,
-      "loss": 5.2056,
+      "learning_rate": 3.138728323699422e-05,
+      "loss": 6.8295,
       "step": 4100
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.0404624277456647e-05,
-      "loss": 5.1704,
+      "learning_rate": 3.130057803468208e-05,
+      "loss": 6.7752,
       "step": 4101
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.0375722543352603e-05,
-      "loss": 5.201,
+      "learning_rate": 3.121387283236994e-05,
+      "loss": 6.8705,
       "step": 4102
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.0346820809248556e-05,
-      "loss": 5.3255,
+      "learning_rate": 3.11271676300578e-05,
+      "loss": 6.9599,
       "step": 4103
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.031791907514451e-05,
-      "loss": 5.385,
+      "learning_rate": 3.1040462427745667e-05,
+      "loss": 6.8992,
       "step": 4104
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.0289017341040463e-05,
-      "loss": 5.1491,
+      "learning_rate": 3.095375722543353e-05,
+      "loss": 6.7947,
       "step": 4105
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.0260115606936416e-05,
-      "loss": 5.1507,
+      "learning_rate": 3.086705202312139e-05,
+      "loss": 6.7966,
       "step": 4106
     },
     {
       "epoch": 4.6,
-      "learning_rate": 1.0231213872832372e-05,
-      "loss": 5.2354,
+      "learning_rate": 3.078034682080925e-05,
+      "loss": 7.0383,
       "step": 4107
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.0202312138728325e-05,
-      "loss": 5.283,
+      "learning_rate": 3.069364161849711e-05,
+      "loss": 7.0103,
       "step": 4108
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.0173410404624278e-05,
-      "loss": 5.1928,
+      "learning_rate": 3.060693641618497e-05,
+      "loss": 7.0165,
       "step": 4109
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.0144508670520232e-05,
-      "loss": 5.2653,
+      "learning_rate": 3.052023121387283e-05,
+      "loss": 6.9284,
       "step": 4110
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.0115606936416185e-05,
-      "loss": 4.9158,
+      "learning_rate": 3.0433526011560692e-05,
+      "loss": 6.8621,
       "step": 4111
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.008670520231214e-05,
-      "loss": 4.8284,
+      "learning_rate": 3.0346820809248553e-05,
+      "loss": 6.8118,
       "step": 4112
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.0057803468208094e-05,
-      "loss": 4.5343,
+      "learning_rate": 3.0260115606936416e-05,
+      "loss": 6.6831,
       "step": 4113
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1.0028901734104047e-05,
-      "loss": 4.5748,
+      "learning_rate": 3.0173410404624277e-05,
+      "loss": 6.7733,
       "step": 4114
     },
     {
       "epoch": 4.61,
-      "learning_rate": 1e-05,
-      "loss": 4.4914,
+      "learning_rate": 3.0086705202312137e-05,
+      "loss": 6.8827,
       "step": 4115
     },
     {
       "epoch": 4.61,
-      "learning_rate": 9.971098265895954e-06,
-      "loss": 4.4291,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 6.8326,
       "step": 4116
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.94219653179191e-06,
-      "loss": 4.2506,
+      "learning_rate": 2.991329479768786e-05,
+      "loss": 6.8167,
       "step": 4117
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.913294797687863e-06,
-      "loss": 3.9041,
+      "learning_rate": 2.982658959537572e-05,
+      "loss": 6.5859,
       "step": 4118
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.884393063583816e-06,
-      "loss": 6.0075,
+      "learning_rate": 2.9739884393063582e-05,
+      "loss": 6.8456,
       "step": 4119
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.85549132947977e-06,
-      "loss": 6.2002,
+      "learning_rate": 2.9653179190751446e-05,
+      "loss": 6.9611,
       "step": 4120
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.826589595375723e-06,
-      "loss": 6.1697,
+      "learning_rate": 2.9566473988439306e-05,
+      "loss": 6.9496,
       "step": 4121
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.797687861271677e-06,
-      "loss": 6.1344,
+      "learning_rate": 2.9479768786127166e-05,
+      "loss": 6.9696,
       "step": 4122
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.768786127167632e-06,
-      "loss": 6.0714,
+      "learning_rate": 2.939306358381503e-05,
+      "loss": 6.9204,
       "step": 4123
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.739884393063585e-06,
-      "loss": 5.8699,
+      "learning_rate": 2.930635838150289e-05,
+      "loss": 6.8583,
       "step": 4124
     },
     {
       "epoch": 4.62,
-      "learning_rate": 9.710982658959537e-06,
-      "loss": 6.0691,
+      "learning_rate": 2.921965317919075e-05,
+      "loss": 6.8663,
       "step": 4125
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.68208092485549e-06,
-      "loss": 5.7163,
+      "learning_rate": 2.9132947976878608e-05,
+      "loss": 6.7986,
       "step": 4126
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.653179190751446e-06,
-      "loss": 5.9173,
+      "learning_rate": 2.9046242774566468e-05,
+      "loss": 7.1,
       "step": 4127
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.624277456647399e-06,
-      "loss": 5.6616,
+      "learning_rate": 2.895953757225433e-05,
+      "loss": 6.905,
       "step": 4128
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.595375722543352e-06,
-      "loss": 5.8049,
+      "learning_rate": 2.8872832369942192e-05,
+      "loss": 7.0603,
       "step": 4129
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.566473988439306e-06,
-      "loss": 5.6885,
+      "learning_rate": 2.8786127167630052e-05,
+      "loss": 7.0588,
       "step": 4130
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.53757225433526e-06,
-      "loss": 5.686,
+      "learning_rate": 2.8699421965317916e-05,
+      "loss": 7.0514,
       "step": 4131
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.508670520231214e-06,
-      "loss": 5.5683,
+      "learning_rate": 2.8612716763005776e-05,
+      "loss": 6.9145,
       "step": 4132
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.479768786127168e-06,
-      "loss": 5.5754,
+      "learning_rate": 2.8526011560693637e-05,
+      "loss": 6.8597,
       "step": 4133
     },
     {
       "epoch": 4.63,
-      "learning_rate": 9.450867052023121e-06,
-      "loss": 5.934,
+      "learning_rate": 2.8439306358381497e-05,
+      "loss": 7.1177,
       "step": 4134
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.421965317919075e-06,
-      "loss": 5.6857,
+      "learning_rate": 2.835260115606936e-05,
+      "loss": 6.9817,
       "step": 4135
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.393063583815028e-06,
-      "loss": 5.6438,
+      "learning_rate": 2.826589595375722e-05,
+      "loss": 6.9682,
       "step": 4136
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.364161849710983e-06,
-      "loss": 5.5081,
+      "learning_rate": 2.817919075144508e-05,
+      "loss": 6.8247,
       "step": 4137
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.335260115606937e-06,
-      "loss": 5.6237,
+      "learning_rate": 2.8092485549132945e-05,
+      "loss": 7.0097,
       "step": 4138
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.30635838150289e-06,
-      "loss": 5.4432,
+      "learning_rate": 2.8005780346820806e-05,
+      "loss": 6.9309,
       "step": 4139
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.277456647398844e-06,
-      "loss": 5.3205,
+      "learning_rate": 2.7919075144508666e-05,
+      "loss": 6.9546,
       "step": 4140
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.248554913294797e-06,
-      "loss": 5.1666,
+      "learning_rate": 2.783236994219653e-05,
+      "loss": 6.8596,
       "step": 4141
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.21965317919075e-06,
-      "loss": 5.2541,
+      "learning_rate": 2.774566473988439e-05,
+      "loss": 6.8443,
       "step": 4142
     },
     {
       "epoch": 4.64,
-      "learning_rate": 9.190751445086706e-06,
-      "loss": 5.56,
+      "learning_rate": 2.765895953757225e-05,
+      "loss": 7.0076,
       "step": 4143
     },
     {
       "epoch": 4.65,
-      "learning_rate": 9.16184971098266e-06,
-      "loss": 5.3525,
+      "learning_rate": 2.757225433526011e-05,
+      "loss": 6.8839,
       "step": 4144
     },
     {
       "epoch": 4.65,
-      "learning_rate": 9.132947976878613e-06,
-      "loss": 5.5597,
+      "learning_rate": 2.7485549132947974e-05,
+      "loss": 7.0486,
       "step": 4145
     },
     {
       "epoch": 4.65,
-      "learning_rate": 9.104046242774566e-06,
-      "loss": 5.4601,
+      "learning_rate": 2.7398843930635835e-05,
+      "loss": 7.0035,
       "step": 4146
     },
     {
       "epoch": 4.65,
-      "learning_rate": 9.07514450867052e-06,
-      "loss": 5.3062,
+      "learning_rate": 2.7312138728323695e-05,
+      "loss": 6.8325,
       "step": 4147
     },
     {
       "epoch": 4.65,
-      "learning_rate": 9.046242774566475e-06,
-      "loss": 5.2759,
+      "learning_rate": 2.722543352601156e-05,
+      "loss": 6.9135,
       "step": 4148
     },
     {
       "epoch": 4.65,
-      "learning_rate": 9.017341040462428e-06,
-      "loss": 5.2559,
+      "learning_rate": 2.713872832369942e-05,
+      "loss": 6.9171,
       "step": 4149
     },
     {
       "epoch": 4.65,
-      "learning_rate": 8.988439306358382e-06,
-      "loss": 5.3074,
+      "learning_rate": 2.705202312138728e-05,
+      "loss": 6.9315,
       "step": 4150
     },
     {
       "epoch": 4.65,
-      "learning_rate": 8.959537572254335e-06,
-      "loss": 5.2623,
+      "learning_rate": 2.696531791907514e-05,
+      "loss": 6.8852,
       "step": 4151
     },
     {
       "epoch": 4.65,
-      "learning_rate": 8.930635838150288e-06,
-      "loss": 5.1409,
+      "learning_rate": 2.6878612716763003e-05,
+      "loss": 6.8059,
       "step": 4152
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.901734104046244e-06,
-      "loss": 5.2475,
+      "learning_rate": 2.6791907514450864e-05,
+      "loss": 6.9575,
       "step": 4153
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.872832369942197e-06,
-      "loss": 5.2607,
+      "learning_rate": 2.6705202312138724e-05,
+      "loss": 6.943,
       "step": 4154
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.84393063583815e-06,
-      "loss": 5.0799,
+      "learning_rate": 2.6618497109826588e-05,
+      "loss": 6.8016,
       "step": 4155
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.815028901734104e-06,
-      "loss": 5.069,
+      "learning_rate": 2.6531791907514448e-05,
+      "loss": 6.8917,
       "step": 4156
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.786127167630057e-06,
-      "loss": 4.9814,
+      "learning_rate": 2.644508670520231e-05,
+      "loss": 6.8176,
       "step": 4157
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.757225433526013e-06,
-      "loss": 5.0874,
+      "learning_rate": 2.635838150289017e-05,
+      "loss": 6.9047,
       "step": 4158
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.728323699421966e-06,
-      "loss": 5.0339,
+      "learning_rate": 2.6271676300578033e-05,
+      "loss": 6.8892,
       "step": 4159
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.69942196531792e-06,
-      "loss": 5.1204,
+      "learning_rate": 2.6184971098265893e-05,
+      "loss": 6.8995,
       "step": 4160
     },
     {
       "epoch": 4.66,
-      "learning_rate": 8.670520231213873e-06,
-      "loss": 5.1418,
+      "learning_rate": 2.6098265895953753e-05,
+      "loss": 7.1038,
       "step": 4161
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.641618497109826e-06,
-      "loss": 4.9368,
+      "learning_rate": 2.6011560693641617e-05,
+      "loss": 6.918,
       "step": 4162
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.61271676300578e-06,
-      "loss": 4.9632,
+      "learning_rate": 2.5924855491329477e-05,
+      "loss": 7.0508,
       "step": 4163
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.583815028901735e-06,
-      "loss": 4.6644,
+      "learning_rate": 2.5838150289017338e-05,
+      "loss": 6.9488,
       "step": 4164
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.554913294797688e-06,
-      "loss": 4.8856,
+      "learning_rate": 2.5751445086705198e-05,
+      "loss": 7.0035,
       "step": 4165
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.526011560693642e-06,
-      "loss": 4.2233,
+      "learning_rate": 2.5664739884393062e-05,
+      "loss": 6.645,
       "step": 4166
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.497109826589595e-06,
-      "loss": 4.1746,
+      "learning_rate": 2.5578034682080922e-05,
+      "loss": 6.6652,
       "step": 4167
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.468208092485549e-06,
-      "loss": 4.2492,
+      "learning_rate": 2.5491329479768782e-05,
+      "loss": 6.7187,
       "step": 4168
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.439306358381504e-06,
-      "loss": 6.0684,
+      "learning_rate": 2.5404624277456646e-05,
+      "loss": 6.9024,
       "step": 4169
     },
     {
       "epoch": 4.67,
-      "learning_rate": 8.410404624277457e-06,
-      "loss": 5.94,
+      "learning_rate": 2.5317919075144507e-05,
+      "loss": 6.9495,
       "step": 4170
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.38150289017341e-06,
-      "loss": 6.0316,
+      "learning_rate": 2.5231213872832367e-05,
+      "loss": 7.0466,
       "step": 4171
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.352601156069364e-06,
-      "loss": 5.874,
+      "learning_rate": 2.514450867052023e-05,
+      "loss": 6.9663,
       "step": 4172
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.323699421965318e-06,
-      "loss": 5.7842,
+      "learning_rate": 2.505780346820809e-05,
+      "loss": 6.9636,
       "step": 4173
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.294797687861273e-06,
-      "loss": 5.75,
+      "learning_rate": 2.497109826589595e-05,
+      "loss": 6.9458,
       "step": 4174
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.265895953757226e-06,
-      "loss": 5.6436,
+      "learning_rate": 2.488439306358381e-05,
+      "loss": 7.015,
       "step": 4175
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.23699421965318e-06,
-      "loss": 5.6983,
+      "learning_rate": 2.4797687861271675e-05,
+      "loss": 7.046,
       "step": 4176
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.208092485549133e-06,
-      "loss": 5.5064,
+      "learning_rate": 2.4710982658959536e-05,
+      "loss": 6.9647,
       "step": 4177
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.179190751445087e-06,
-      "loss": 5.5451,
+      "learning_rate": 2.4624277456647396e-05,
+      "loss": 6.9196,
       "step": 4178
     },
     {
       "epoch": 4.68,
-      "learning_rate": 8.150289017341042e-06,
-      "loss": 5.4936,
+      "learning_rate": 2.453757225433526e-05,
+      "loss": 6.9567,
       "step": 4179
     },
     {
       "epoch": 4.69,
-      "learning_rate": 8.121387283236995e-06,
-      "loss": 5.5739,
+      "learning_rate": 2.445086705202312e-05,
+      "loss": 6.9303,
       "step": 4180
     },
     {
       "epoch": 4.69,
-      "learning_rate": 8.092485549132949e-06,
-      "loss": 5.5792,
+      "learning_rate": 2.436416184971098e-05,
+      "loss": 6.9048,
       "step": 4181
     },
     {
       "epoch": 4.69,
-      "learning_rate": 8.063583815028902e-06,
-      "loss": 5.6132,
+      "learning_rate": 2.427745664739884e-05,
+      "loss": 6.9593,
       "step": 4182
     },
     {
       "epoch": 4.69,
-      "learning_rate": 8.034682080924856e-06,
-      "loss": 5.5796,
+      "learning_rate": 2.4190751445086704e-05,
+      "loss": 6.936,
       "step": 4183
     },
     {
       "epoch": 4.69,
-      "learning_rate": 8.00578034682081e-06,
-      "loss": 5.5946,
+      "learning_rate": 2.4104046242774565e-05,
+      "loss": 7.0035,
       "step": 4184
     },
     {
       "epoch": 4.69,
-      "learning_rate": 7.976878612716764e-06,
-      "loss": 5.403,
+      "learning_rate": 2.4017341040462425e-05,
+      "loss": 6.8994,
       "step": 4185
     },
     {
       "epoch": 4.69,
-      "learning_rate": 7.947976878612718e-06,
-      "loss": 5.2869,
+      "learning_rate": 2.393063583815029e-05,
+      "loss": 6.8942,
       "step": 4186
     },
     {
       "epoch": 4.69,
-      "learning_rate": 7.919075144508671e-06,
-      "loss": 5.4534,
+      "learning_rate": 2.384393063583815e-05,
+      "loss": 6.9649,
       "step": 4187
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.890173410404624e-06,
-      "loss": 5.5861,
+      "learning_rate": 2.375722543352601e-05,
+      "loss": 7.1306,
       "step": 4188
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.861271676300578e-06,
-      "loss": 5.3014,
+      "learning_rate": 2.367052023121387e-05,
+      "loss": 6.8744,
       "step": 4189
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.832369942196533e-06,
-      "loss": 5.5538,
+      "learning_rate": 2.3583815028901734e-05,
+      "loss": 7.0015,
       "step": 4190
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.803468208092486e-06,
-      "loss": 5.4302,
+      "learning_rate": 2.3497109826589594e-05,
+      "loss": 6.8961,
       "step": 4191
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.77456647398844e-06,
-      "loss": 5.3928,
+      "learning_rate": 2.3410404624277454e-05,
+      "loss": 6.9084,
       "step": 4192
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.745664739884393e-06,
-      "loss": 5.4786,
+      "learning_rate": 2.3323699421965318e-05,
+      "loss": 7.0515,
       "step": 4193
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.716763005780347e-06,
-      "loss": 5.377,
+      "learning_rate": 2.323699421965318e-05,
+      "loss": 6.8986,
       "step": 4194
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.687861271676302e-06,
-      "loss": 5.55,
+      "learning_rate": 2.315028901734104e-05,
+      "loss": 7.1274,
       "step": 4195
     },
     {
       "epoch": 4.7,
-      "learning_rate": 7.658959537572255e-06,
-      "loss": 5.2516,
+      "learning_rate": 2.30635838150289e-05,
+      "loss": 6.8865,
       "step": 4196
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.630057803468209e-06,
-      "loss": 5.4221,
+      "learning_rate": 2.2976878612716763e-05,
+      "loss": 6.9769,
       "step": 4197
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.601156069364162e-06,
-      "loss": 5.4687,
+      "learning_rate": 2.2890173410404623e-05,
+      "loss": 7.0439,
       "step": 4198
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.5722543352601166e-06,
-      "loss": 5.3625,
+      "learning_rate": 2.2803468208092483e-05,
+      "loss": 6.9815,
       "step": 4199
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.54335260115607e-06,
-      "loss": 5.4341,
+      "learning_rate": 2.2716763005780347e-05,
+      "loss": 7.0395,
       "step": 4200
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.514450867052024e-06,
-      "loss": 5.1456,
+      "learning_rate": 2.2630057803468208e-05,
+      "loss": 6.834,
       "step": 4201
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.485549132947978e-06,
-      "loss": 5.3496,
+      "learning_rate": 2.2543352601156068e-05,
+      "loss": 6.9697,
       "step": 4202
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.456647398843931e-06,
-      "loss": 5.1668,
+      "learning_rate": 2.245664739884393e-05,
+      "loss": 6.8685,
       "step": 4203
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.4277456647398855e-06,
-      "loss": 5.0742,
+      "learning_rate": 2.2369942196531792e-05,
+      "loss": 6.9343,
       "step": 4204
     },
     {
       "epoch": 4.71,
-      "learning_rate": 7.398843930635839e-06,
-      "loss": 5.0064,
+      "learning_rate": 2.2283236994219652e-05,
+      "loss": 6.8214,
       "step": 4205
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.369942196531792e-06,
-      "loss": 5.2777,
+      "learning_rate": 2.2196531791907513e-05,
+      "loss": 7.0009,
       "step": 4206
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.341040462427747e-06,
-      "loss": 5.1695,
+      "learning_rate": 2.2109826589595376e-05,
+      "loss": 6.9761,
       "step": 4207
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.3121387283237e-06,
-      "loss": 4.9523,
+      "learning_rate": 2.2023121387283237e-05,
+      "loss": 6.8327,
       "step": 4208
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.2832369942196544e-06,
-      "loss": 5.1878,
+      "learning_rate": 2.1936416184971097e-05,
+      "loss": 7.0641,
       "step": 4209
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.254335260115608e-06,
-      "loss": 5.2217,
+      "learning_rate": 2.184971098265896e-05,
+      "loss": 7.1142,
       "step": 4210
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.225433526011561e-06,
-      "loss": 4.7771,
+      "learning_rate": 2.176300578034682e-05,
+      "loss": 6.81,
       "step": 4211
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.196531791907516e-06,
-      "loss": 4.857,
+      "learning_rate": 2.167630057803468e-05,
+      "loss": 6.8699,
       "step": 4212
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.167630057803469e-06,
-      "loss": 4.6437,
+      "learning_rate": 2.1589595375722542e-05,
+      "loss": 6.7522,
       "step": 4213
     },
     {
       "epoch": 4.72,
-      "learning_rate": 7.1387283236994225e-06,
-      "loss": 4.766,
+      "learning_rate": 2.1502890173410405e-05,
+      "loss": 6.8971,
       "step": 4214
     },
     {
       "epoch": 4.73,
-      "learning_rate": 7.109826589595377e-06,
-      "loss": 4.3298,
+      "learning_rate": 2.1416184971098266e-05,
+      "loss": 6.8525,
       "step": 4215
     },
     {
       "epoch": 4.73,
-      "learning_rate": 7.08092485549133e-06,
-      "loss": 4.3057,
+      "learning_rate": 2.1329479768786126e-05,
+      "loss": 6.7815,
       "step": 4216
     },
     {
       "epoch": 4.73,
-      "learning_rate": 7.0520231213872845e-06,
-      "loss": 4.3092,
+      "learning_rate": 2.124277456647399e-05,
+      "loss": 6.7988,
       "step": 4217
     },
     {
       "epoch": 4.73,
-      "learning_rate": 7.023121387283238e-06,
-      "loss": 3.715,
+      "learning_rate": 2.115606936416185e-05,
+      "loss": 6.6496,
       "step": 4218
     },
     {
       "epoch": 4.73,
-      "learning_rate": 6.9942196531791914e-06,
-      "loss": 6.4174,
+      "learning_rate": 2.106936416184971e-05,
+      "loss": 7.0981,
       "step": 4219
     },
     {
       "epoch": 4.73,
-      "learning_rate": 6.965317919075146e-06,
-      "loss": 6.2261,
+      "learning_rate": 2.098265895953757e-05,
+      "loss": 7.0139,
       "step": 4220
     },
     {
       "epoch": 4.73,
-      "learning_rate": 6.936416184971098e-06,
-      "loss": 6.2294,
+      "learning_rate": 2.0895953757225435e-05,
+      "loss": 7.0817,
       "step": 4221
     },
     {
       "epoch": 4.73,
-      "learning_rate": 6.907514450867052e-06,
-      "loss": 6.224,
+      "learning_rate": 2.080924855491329e-05,
+      "loss": 7.1166,
       "step": 4222
     },
     {
       "epoch": 4.73,
-      "learning_rate": 6.878612716763005e-06,
-      "loss": 6.0364,
+      "learning_rate": 2.0722543352601152e-05,
+      "loss": 6.8537,
       "step": 4223
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.8497109826589595e-06,
-      "loss": 5.8933,
+      "learning_rate": 2.0635838150289012e-05,
+      "loss": 6.7851,
       "step": 4224
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.820809248554913e-06,
-      "loss": 5.9875,
+      "learning_rate": 2.0549132947976876e-05,
+      "loss": 7.0243,
       "step": 4225
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.791907514450866e-06,
-      "loss": 5.8701,
+      "learning_rate": 2.0462427745664736e-05,
+      "loss": 6.9191,
       "step": 4226
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.763005780346821e-06,
-      "loss": 5.8085,
+      "learning_rate": 2.0375722543352597e-05,
+      "loss": 6.9656,
       "step": 4227
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.734104046242774e-06,
-      "loss": 5.639,
+      "learning_rate": 2.028901734104046e-05,
+      "loss": 6.9592,
       "step": 4228
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.7052023121387284e-06,
-      "loss": 5.5634,
+      "learning_rate": 2.020231213872832e-05,
+      "loss": 6.9574,
       "step": 4229
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.676300578034682e-06,
-      "loss": 5.5456,
+      "learning_rate": 2.011560693641618e-05,
+      "loss": 6.915,
       "step": 4230
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.647398843930635e-06,
-      "loss": 5.514,
+      "learning_rate": 2.002890173410404e-05,
+      "loss": 6.8419,
       "step": 4231
     },
     {
       "epoch": 4.74,
-      "learning_rate": 6.61849710982659e-06,
-      "loss": 5.4247,
+      "learning_rate": 1.9942196531791905e-05,
+      "loss": 6.8581,
       "step": 4232
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.589595375722543e-06,
-      "loss": 5.4829,
+      "learning_rate": 1.9855491329479765e-05,
+      "loss": 6.8334,
       "step": 4233
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.5606936416184965e-06,
-      "loss": 5.5631,
+      "learning_rate": 1.9768786127167626e-05,
+      "loss": 6.9677,
       "step": 4234
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.531791907514451e-06,
-      "loss": 5.5608,
+      "learning_rate": 1.968208092485549e-05,
+      "loss": 6.911,
       "step": 4235
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.502890173410404e-06,
-      "loss": 5.5731,
+      "learning_rate": 1.959537572254335e-05,
+      "loss": 6.9197,
       "step": 4236
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.4739884393063585e-06,
-      "loss": 5.5844,
+      "learning_rate": 1.950867052023121e-05,
+      "loss": 6.9141,
       "step": 4237
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.445086705202312e-06,
-      "loss": 5.5307,
+      "learning_rate": 1.942196531791907e-05,
+      "loss": 6.9379,
       "step": 4238
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.4161849710982654e-06,
-      "loss": 5.4756,
+      "learning_rate": 1.9335260115606934e-05,
+      "loss": 6.8733,
       "step": 4239
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.38728323699422e-06,
-      "loss": 5.3833,
+      "learning_rate": 1.9248554913294795e-05,
+      "loss": 6.9716,
       "step": 4240
     },
     {
       "epoch": 4.75,
-      "learning_rate": 6.358381502890173e-06,
-      "loss": 5.059,
+      "learning_rate": 1.9161849710982655e-05,
+      "loss": 6.7101,
       "step": 4241
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.3294797687861275e-06,
-      "loss": 5.3328,
+      "learning_rate": 1.907514450867052e-05,
+      "loss": 6.9088,
       "step": 4242
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.300578034682081e-06,
-      "loss": 5.3949,
+      "learning_rate": 1.898843930635838e-05,
+      "loss": 6.9192,
       "step": 4243
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.271676300578034e-06,
-      "loss": 5.3645,
+      "learning_rate": 1.890173410404624e-05,
+      "loss": 7.0275,
       "step": 4244
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.242774566473989e-06,
-      "loss": 5.3693,
+      "learning_rate": 1.88150289017341e-05,
+      "loss": 7.0108,
       "step": 4245
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.213872832369942e-06,
-      "loss": 5.373,
+      "learning_rate": 1.8728323699421963e-05,
+      "loss": 6.9408,
       "step": 4246
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.1849710982658956e-06,
-      "loss": 5.3452,
+      "learning_rate": 1.8641618497109824e-05,
+      "loss": 7.0375,
       "step": 4247
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.15606936416185e-06,
-      "loss": 5.1939,
+      "learning_rate": 1.8554913294797684e-05,
+      "loss": 6.9193,
       "step": 4248
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.127167630057803e-06,
-      "loss": 5.1361,
+      "learning_rate": 1.8468208092485548e-05,
+      "loss": 6.747,
       "step": 4249
     },
     {
       "epoch": 4.76,
-      "learning_rate": 6.098265895953758e-06,
-      "loss": 5.229,
+      "learning_rate": 1.8381502890173408e-05,
+      "loss": 6.9606,
       "step": 4250
     },
     {
       "epoch": 4.77,
-      "learning_rate": 6.069364161849711e-06,
-      "loss": 5.3744,
+      "learning_rate": 1.829479768786127e-05,
+      "loss": 6.9733,
       "step": 4251
     },
     {
       "epoch": 4.77,
-      "learning_rate": 6.0404624277456645e-06,
-      "loss": 5.3393,
+      "learning_rate": 1.8208092485549132e-05,
+      "loss": 6.9911,
       "step": 4252
     },
     {
       "epoch": 4.77,
-      "learning_rate": 6.011560693641619e-06,
-      "loss": 5.3941,
+      "learning_rate": 1.8121387283236993e-05,
+      "loss": 6.9785,
       "step": 4253
     },
     {
       "epoch": 4.77,
-      "learning_rate": 5.982658959537572e-06,
-      "loss": 5.1939,
+      "learning_rate": 1.8034682080924853e-05,
+      "loss": 6.7716,
       "step": 4254
     },
     {
       "epoch": 4.77,
-      "learning_rate": 5.9537572254335265e-06,
-      "loss": 5.2072,
+      "learning_rate": 1.7947976878612713e-05,
+      "loss": 6.9537,
       "step": 4255
     },
     {
       "epoch": 4.77,
-      "learning_rate": 5.92485549132948e-06,
-      "loss": 5.1388,
+      "learning_rate": 1.7861271676300577e-05,
+      "loss": 6.9196,
       "step": 4256
     },
     {
       "epoch": 4.77,
-      "learning_rate": 5.895953757225433e-06,
-      "loss": 4.9826,
+      "learning_rate": 1.7774566473988437e-05,
+      "loss": 6.7952,
       "step": 4257
     },
     {
       "epoch": 4.77,
-      "learning_rate": 5.867052023121388e-06,
-      "loss": 5.0052,
+      "learning_rate": 1.7687861271676298e-05,
+      "loss": 6.8975,
       "step": 4258
     },
     {
       "epoch": 4.77,
-      "learning_rate": 5.838150289017341e-06,
-      "loss": 5.1712,
+      "learning_rate": 1.760115606936416e-05,
+      "loss": 7.0171,
       "step": 4259
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.809248554913295e-06,
-      "loss": 4.8303,
+      "learning_rate": 1.7514450867052022e-05,
+      "loss": 6.8101,
       "step": 4260
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.780346820809249e-06,
-      "loss": 5.0595,
+      "learning_rate": 1.7427745664739882e-05,
+      "loss": 7.0577,
       "step": 4261
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.751445086705202e-06,
-      "loss": 4.5839,
+      "learning_rate": 1.7341040462427742e-05,
+      "loss": 6.7525,
       "step": 4262
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.722543352601157e-06,
-      "loss": 4.5375,
+      "learning_rate": 1.7254335260115606e-05,
+      "loss": 6.7595,
       "step": 4263
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.69364161849711e-06,
-      "loss": 4.3971,
+      "learning_rate": 1.7167630057803466e-05,
+      "loss": 6.7598,
       "step": 4264
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.6647398843930635e-06,
-      "loss": 4.4333,
+      "learning_rate": 1.7080924855491327e-05,
+      "loss": 6.7106,
       "step": 4265
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.635838150289018e-06,
-      "loss": 4.2823,
+      "learning_rate": 1.699421965317919e-05,
+      "loss": 6.7523,
       "step": 4266
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.606936416184971e-06,
-      "loss": 4.1607,
+      "learning_rate": 1.690751445086705e-05,
+      "loss": 6.7534,
       "step": 4267
     },
     {
       "epoch": 4.78,
-      "learning_rate": 5.578034682080925e-06,
-      "loss": 3.4364,
+      "learning_rate": 1.682080924855491e-05,
+      "loss": 6.4369,
       "step": 4268
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.549132947976879e-06,
-      "loss": 6.305,
+      "learning_rate": 1.673410404624277e-05,
+      "loss": 6.9857,
       "step": 4269
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.5202312138728324e-06,
-      "loss": 6.3239,
+      "learning_rate": 1.6647398843930635e-05,
+      "loss": 7.0151,
       "step": 4270
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.491329479768787e-06,
-      "loss": 6.1988,
+      "learning_rate": 1.6560693641618496e-05,
+      "loss": 6.9181,
       "step": 4271
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.46242774566474e-06,
-      "loss": 5.9985,
+      "learning_rate": 1.6473988439306356e-05,
+      "loss": 6.7758,
       "step": 4272
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.433526011560694e-06,
-      "loss": 6.207,
+      "learning_rate": 1.638728323699422e-05,
+      "loss": 6.9219,
       "step": 4273
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.404624277456648e-06,
-      "loss": 6.2508,
+      "learning_rate": 1.630057803468208e-05,
+      "loss": 7.0054,
       "step": 4274
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.375722543352601e-06,
-      "loss": 6.1254,
+      "learning_rate": 1.621387283236994e-05,
+      "loss": 7.0063,
       "step": 4275
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.346820809248556e-06,
-      "loss": 6.081,
+      "learning_rate": 1.61271676300578e-05,
+      "loss": 7.0255,
       "step": 4276
     },
     {
       "epoch": 4.79,
-      "learning_rate": 5.317919075144509e-06,
-      "loss": 5.8837,
+      "learning_rate": 1.6040462427745664e-05,
+      "loss": 6.8821,
       "step": 4277
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.2890173410404626e-06,
-      "loss": 5.7216,
+      "learning_rate": 1.5953757225433525e-05,
+      "loss": 6.8539,
       "step": 4278
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.260115606936417e-06,
-      "loss": 5.5981,
+      "learning_rate": 1.5867052023121385e-05,
+      "loss": 6.8853,
       "step": 4279
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.23121387283237e-06,
-      "loss": 5.8126,
+      "learning_rate": 1.578034682080925e-05,
+      "loss": 7.0145,
       "step": 4280
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.202312138728324e-06,
-      "loss": 5.5098,
+      "learning_rate": 1.569364161849711e-05,
+      "loss": 6.9408,
       "step": 4281
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.173410404624278e-06,
-      "loss": 5.5075,
+      "learning_rate": 1.560693641618497e-05,
+      "loss": 6.8978,
       "step": 4282
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.1445086705202315e-06,
-      "loss": 5.7925,
+      "learning_rate": 1.5520231213872833e-05,
+      "loss": 7.1277,
       "step": 4283
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.115606936416186e-06,
-      "loss": 5.4596,
+      "learning_rate": 1.5433526011560694e-05,
+      "loss": 6.8905,
       "step": 4284
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.086705202312139e-06,
-      "loss": 5.6335,
+      "learning_rate": 1.5346820809248554e-05,
+      "loss": 7.01,
       "step": 4285
     },
     {
       "epoch": 4.8,
-      "learning_rate": 5.057803468208093e-06,
-      "loss": 5.3928,
+      "learning_rate": 1.5260115606936414e-05,
+      "loss": 6.8554,
       "step": 4286
     },
     {
       "epoch": 4.81,
-      "learning_rate": 5.028901734104047e-06,
-      "loss": 5.5797,
+      "learning_rate": 1.5173410404624276e-05,
+      "loss": 6.9597,
       "step": 4287
     },
     {
       "epoch": 4.81,
-      "learning_rate": 5e-06,
-      "loss": 5.5949,
+      "learning_rate": 1.5086705202312138e-05,
+      "loss": 6.9864,
       "step": 4288
     },
     {
       "epoch": 4.81,
-      "learning_rate": 4.971098265895955e-06,
-      "loss": 5.562,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 6.9583,
       "step": 4289
     },
     {
       "epoch": 4.81,
-      "learning_rate": 4.942196531791908e-06,
-      "loss": 5.5725,
+      "learning_rate": 1.491329479768786e-05,
+      "loss": 6.9182,
       "step": 4290
     },
     {
       "epoch": 4.81,
-      "learning_rate": 4.913294797687862e-06,
-      "loss": 5.4066,
+      "learning_rate": 1.4826589595375723e-05,
+      "loss": 6.8201,
       "step": 4291
     },
     {
       "epoch": 4.81,
-      "learning_rate": 4.884393063583816e-06,
-      "loss": 5.4024,
+      "learning_rate": 1.4739884393063583e-05,
+      "loss": 6.8999,
       "step": 4292
     },
     {
       "epoch": 4.81,
-      "learning_rate": 4.8554913294797685e-06,
-      "loss": 5.388,
+      "learning_rate": 1.4653179190751445e-05,
+      "loss": 6.9465,
       "step": 4293
     },
     {
       "epoch": 4.81,
-      "learning_rate": 4.826589595375723e-06,
-      "loss": 5.2656,
+      "learning_rate": 1.4566473988439304e-05,
+      "loss": 6.8985,
       "step": 4294
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.797687861271676e-06,
-      "loss": 5.1213,
+      "learning_rate": 1.4479768786127166e-05,
+      "loss": 6.8349,
       "step": 4295
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.76878612716763e-06,
-      "loss": 5.4919,
+      "learning_rate": 1.4393063583815026e-05,
+      "loss": 7.0739,
       "step": 4296
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.739884393063584e-06,
-      "loss": 5.2063,
+      "learning_rate": 1.4306358381502888e-05,
+      "loss": 6.9107,
       "step": 4297
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.710982658959537e-06,
-      "loss": 5.3047,
+      "learning_rate": 1.4219653179190749e-05,
+      "loss": 7.0469,
       "step": 4298
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.682080924855492e-06,
-      "loss": 5.3144,
+      "learning_rate": 1.413294797687861e-05,
+      "loss": 6.925,
       "step": 4299
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.653179190751445e-06,
-      "loss": 5.0782,
+      "learning_rate": 1.4046242774566473e-05,
+      "loss": 6.8243,
       "step": 4300
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.624277456647399e-06,
-      "loss": 5.2284,
+      "learning_rate": 1.3959537572254333e-05,
+      "loss": 6.9001,
       "step": 4301
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.595375722543353e-06,
-      "loss": 5.1857,
+      "learning_rate": 1.3872832369942195e-05,
+      "loss": 6.88,
       "step": 4302
     },
     {
       "epoch": 4.82,
-      "learning_rate": 4.566473988439306e-06,
-      "loss": 5.347,
+      "learning_rate": 1.3786127167630055e-05,
+      "loss": 7.0367,
       "step": 4303
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.53757225433526e-06,
-      "loss": 5.0277,
+      "learning_rate": 1.3699421965317917e-05,
+      "loss": 6.7954,
       "step": 4304
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.508670520231214e-06,
-      "loss": 5.1388,
+      "learning_rate": 1.361271676300578e-05,
+      "loss": 6.8722,
       "step": 4305
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.4797687861271675e-06,
-      "loss": 5.1877,
+      "learning_rate": 1.352601156069364e-05,
+      "loss": 7.0166,
       "step": 4306
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.450867052023122e-06,
-      "loss": 5.0038,
+      "learning_rate": 1.3439306358381502e-05,
+      "loss": 6.7862,
       "step": 4307
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.421965317919075e-06,
-      "loss": 5.0988,
+      "learning_rate": 1.3352601156069362e-05,
+      "loss": 6.9026,
       "step": 4308
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.393063583815029e-06,
-      "loss": 4.9769,
+      "learning_rate": 1.3265895953757224e-05,
+      "loss": 6.8792,
       "step": 4309
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.364161849710983e-06,
-      "loss": 5.0264,
+      "learning_rate": 1.3179190751445084e-05,
+      "loss": 6.8808,
       "step": 4310
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.3352601156069365e-06,
-      "loss": 4.8607,
+      "learning_rate": 1.3092485549132946e-05,
+      "loss": 6.8781,
       "step": 4311
     },
     {
       "epoch": 4.83,
-      "learning_rate": 4.30635838150289e-06,
-      "loss": 4.8676,
+      "learning_rate": 1.3005780346820809e-05,
+      "loss": 6.8583,
       "step": 4312
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.277456647398844e-06,
-      "loss": 4.773,
+      "learning_rate": 1.2919075144508669e-05,
+      "loss": 6.8026,
       "step": 4313
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.248554913294798e-06,
-      "loss": 4.8571,
+      "learning_rate": 1.2832369942196531e-05,
+      "loss": 6.9904,
       "step": 4314
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.219653179190752e-06,
-      "loss": 4.3389,
+      "learning_rate": 1.2745664739884391e-05,
+      "loss": 6.6062,
       "step": 4315
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.190751445086705e-06,
-      "loss": 4.442,
+      "learning_rate": 1.2658959537572253e-05,
+      "loss": 6.7661,
       "step": 4316
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.161849710982659e-06,
-      "loss": 4.075,
+      "learning_rate": 1.2572254335260115e-05,
+      "loss": 6.6967,
       "step": 4317
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.132947976878613e-06,
-      "loss": 3.7851,
+      "learning_rate": 1.2485549132947976e-05,
+      "loss": 6.6814,
       "step": 4318
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.1040462427745666e-06,
-      "loss": 6.3274,
+      "learning_rate": 1.2398843930635838e-05,
+      "loss": 7.0375,
       "step": 4319
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.075144508670521e-06,
-      "loss": 6.2208,
+      "learning_rate": 1.2312138728323698e-05,
+      "loss": 6.8326,
       "step": 4320
     },
     {
       "epoch": 4.84,
-      "learning_rate": 4.046242774566474e-06,
-      "loss": 6.2574,
+      "learning_rate": 1.222543352601156e-05,
+      "loss": 6.976,
       "step": 4321
     },
     {
       "epoch": 4.85,
-      "learning_rate": 4.017341040462428e-06,
-      "loss": 6.37,
+      "learning_rate": 1.213872832369942e-05,
+      "loss": 7.0123,
       "step": 4322
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.988439306358382e-06,
-      "loss": 6.1325,
+      "learning_rate": 1.2052023121387282e-05,
+      "loss": 6.9386,
       "step": 4323
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.9595375722543355e-06,
-      "loss": 6.1964,
+      "learning_rate": 1.1965317919075144e-05,
+      "loss": 7.0615,
       "step": 4324
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.930635838150289e-06,
-      "loss": 6.2439,
+      "learning_rate": 1.1878612716763005e-05,
+      "loss": 7.0096,
       "step": 4325
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.901734104046243e-06,
-      "loss": 6.0703,
+      "learning_rate": 1.1791907514450867e-05,
+      "loss": 6.9045,
       "step": 4326
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.872832369942197e-06,
-      "loss": 6.1186,
+      "learning_rate": 1.1705202312138727e-05,
+      "loss": 6.9573,
       "step": 4327
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.843930635838151e-06,
-      "loss": 5.9281,
+      "learning_rate": 1.161849710982659e-05,
+      "loss": 6.9085,
       "step": 4328
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.815028901734104e-06,
-      "loss": 5.885,
+      "learning_rate": 1.153179190751445e-05,
+      "loss": 7.0225,
       "step": 4329
     },
     {
       "epoch": 4.85,
-      "learning_rate": 3.7861271676300583e-06,
-      "loss": 5.8815,
+      "learning_rate": 1.1445086705202312e-05,
+      "loss": 6.9884,
       "step": 4330
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.757225433526012e-06,
-      "loss": 5.5378,
+      "learning_rate": 1.1358381502890174e-05,
+      "loss": 6.8637,
       "step": 4331
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.7283236994219656e-06,
-      "loss": 5.6502,
+      "learning_rate": 1.1271676300578034e-05,
+      "loss": 7.0758,
       "step": 4332
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.6994219653179195e-06,
-      "loss": 5.6646,
+      "learning_rate": 1.1184971098265896e-05,
+      "loss": 7.0392,
       "step": 4333
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.6705202312138733e-06,
-      "loss": 5.3647,
+      "learning_rate": 1.1098265895953756e-05,
+      "loss": 6.8839,
       "step": 4334
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.6416184971098272e-06,
-      "loss": 5.4427,
+      "learning_rate": 1.1011560693641618e-05,
+      "loss": 6.9149,
       "step": 4335
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.6127167630057807e-06,
-      "loss": 5.283,
+      "learning_rate": 1.092485549132948e-05,
+      "loss": 6.87,
       "step": 4336
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.5838150289017345e-06,
-      "loss": 5.4797,
+      "learning_rate": 1.083815028901734e-05,
+      "loss": 7.0021,
       "step": 4337
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.5549132947976884e-06,
-      "loss": 5.6146,
+      "learning_rate": 1.0751445086705203e-05,
+      "loss": 7.055,
       "step": 4338
     },
     {
       "epoch": 4.86,
-      "learning_rate": 3.5260115606936423e-06,
-      "loss": 5.3316,
+      "learning_rate": 1.0664739884393063e-05,
+      "loss": 6.8057,
       "step": 4339
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.4971098265895957e-06,
-      "loss": 5.6296,
+      "learning_rate": 1.0578034682080925e-05,
+      "loss": 6.9673,
       "step": 4340
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.468208092485549e-06,
-      "loss": 5.464,
+      "learning_rate": 1.0491329479768785e-05,
+      "loss": 6.8945,
       "step": 4341
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.4393063583815026e-06,
-      "loss": 5.6087,
+      "learning_rate": 1.0404624277456646e-05,
+      "loss": 6.9544,
       "step": 4342
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.4104046242774565e-06,
-      "loss": 5.5159,
+      "learning_rate": 1.0317919075144506e-05,
+      "loss": 6.9317,
       "step": 4343
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.3815028901734103e-06,
-      "loss": 5.3658,
+      "learning_rate": 1.0231213872832368e-05,
+      "loss": 6.8782,
       "step": 4344
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.3526011560693642e-06,
-      "loss": 5.7776,
+      "learning_rate": 1.014450867052023e-05,
+      "loss": 7.0921,
       "step": 4345
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.3236994219653177e-06,
-      "loss": 5.5267,
+      "learning_rate": 1.005780346820809e-05,
+      "loss": 7.038,
       "step": 4346
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.2947976878612715e-06,
-      "loss": 5.4011,
+      "learning_rate": 9.971098265895953e-06,
+      "loss": 6.961,
       "step": 4347
     },
     {
       "epoch": 4.87,
-      "learning_rate": 3.2658959537572254e-06,
-      "loss": 5.2775,
+      "learning_rate": 9.884393063583813e-06,
+      "loss": 6.9794,
       "step": 4348
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.2369942196531793e-06,
-      "loss": 5.4673,
+      "learning_rate": 9.797687861271675e-06,
+      "loss": 7.08,
       "step": 4349
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.2080924855491327e-06,
-      "loss": 5.3458,
+      "learning_rate": 9.710982658959535e-06,
+      "loss": 7.0134,
       "step": 4350
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.1791907514450866e-06,
-      "loss": 5.3345,
+      "learning_rate": 9.624277456647397e-06,
+      "loss": 6.9623,
       "step": 4351
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.1502890173410405e-06,
-      "loss": 5.2402,
+      "learning_rate": 9.53757225433526e-06,
+      "loss": 6.9728,
       "step": 4352
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.1213872832369943e-06,
-      "loss": 5.0652,
+      "learning_rate": 9.45086705202312e-06,
+      "loss": 6.8967,
       "step": 4353
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.0924855491329478e-06,
-      "loss": 5.1677,
+      "learning_rate": 9.364161849710982e-06,
+      "loss": 6.9406,
       "step": 4354
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.0635838150289016e-06,
-      "loss": 5.2843,
+      "learning_rate": 9.277456647398842e-06,
+      "loss": 7.03,
       "step": 4355
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.0346820809248555e-06,
-      "loss": 5.1184,
+      "learning_rate": 9.190751445086704e-06,
+      "loss": 6.9053,
       "step": 4356
     },
     {
       "epoch": 4.88,
-      "learning_rate": 3.0057803468208094e-06,
-      "loss": 5.1615,
+      "learning_rate": 9.104046242774566e-06,
+      "loss": 7.0074,
       "step": 4357
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.9768786127167633e-06,
-      "loss": 4.959,
+      "learning_rate": 9.017341040462426e-06,
+      "loss": 6.9067,
       "step": 4358
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.9479768786127167e-06,
-      "loss": 4.7911,
+      "learning_rate": 8.930635838150288e-06,
+      "loss": 6.8997,
       "step": 4359
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.9190751445086706e-06,
-      "loss": 4.9612,
+      "learning_rate": 8.843930635838149e-06,
+      "loss": 6.9879,
       "step": 4360
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.8901734104046244e-06,
-      "loss": 4.7392,
+      "learning_rate": 8.757225433526011e-06,
+      "loss": 6.7785,
       "step": 4361
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.8612716763005783e-06,
-      "loss": 4.4631,
+      "learning_rate": 8.670520231213871e-06,
+      "loss": 6.7391,
       "step": 4362
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.8323699421965318e-06,
-      "loss": 4.5451,
+      "learning_rate": 8.583815028901733e-06,
+      "loss": 6.7808,
       "step": 4363
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.8034682080924856e-06,
-      "loss": 4.4182,
+      "learning_rate": 8.497109826589595e-06,
+      "loss": 6.7403,
       "step": 4364
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.7745664739884395e-06,
-      "loss": 4.7127,
+      "learning_rate": 8.410404624277456e-06,
+      "loss": 6.9322,
       "step": 4365
     },
     {
       "epoch": 4.89,
-      "learning_rate": 2.7456647398843934e-06,
-      "loss": 4.1891,
+      "learning_rate": 8.323699421965318e-06,
+      "loss": 6.6861,
       "step": 4366
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.716763005780347e-06,
-      "loss": 4.2624,
+      "learning_rate": 8.236994219653178e-06,
+      "loss": 6.743,
       "step": 4367
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.6878612716763007e-06,
-      "loss": 3.7581,
+      "learning_rate": 8.15028901734104e-06,
+      "loss": 6.5922,
       "step": 4368
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.6589595375722546e-06,
-      "loss": 6.1743,
+      "learning_rate": 8.0635838150289e-06,
+      "loss": 7.0225,
       "step": 4369
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.6300578034682084e-06,
-      "loss": 6.0782,
+      "learning_rate": 7.976878612716762e-06,
+      "loss": 6.9654,
       "step": 4370
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.601156069364162e-06,
-      "loss": 6.0032,
+      "learning_rate": 7.890173410404624e-06,
+      "loss": 6.8714,
       "step": 4371
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.5722543352601157e-06,
-      "loss": 6.075,
+      "learning_rate": 7.803468208092485e-06,
+      "loss": 6.9203,
       "step": 4372
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.5433526011560696e-06,
-      "loss": 5.9823,
+      "learning_rate": 7.716763005780347e-06,
+      "loss": 6.9038,
       "step": 4373
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.5144508670520235e-06,
-      "loss": 6.0397,
+      "learning_rate": 7.630057803468207e-06,
+      "loss": 6.9594,
       "step": 4374
     },
     {
       "epoch": 4.9,
-      "learning_rate": 2.4855491329479774e-06,
-      "loss": 6.1663,
+      "learning_rate": 7.543352601156069e-06,
+      "loss": 7.0463,
       "step": 4375
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.456647398843931e-06,
-      "loss": 5.9713,
+      "learning_rate": 7.45664739884393e-06,
+      "loss": 6.8771,
       "step": 4376
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.4277456647398842e-06,
-      "loss": 6.0608,
+      "learning_rate": 7.3699421965317915e-06,
+      "loss": 6.9704,
       "step": 4377
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.398843930635838e-06,
-      "loss": 5.8153,
+      "learning_rate": 7.283236994219652e-06,
+      "loss": 6.9203,
       "step": 4378
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.369942196531792e-06,
-      "loss": 5.9472,
+      "learning_rate": 7.196531791907513e-06,
+      "loss": 6.9814,
       "step": 4379
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.369942196531792e-06,
-      "loss": 5.6277,
+      "learning_rate": 7.109826589595374e-06,
+      "loss": 6.8491,
       "step": 4380
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.341040462427746e-06,
-      "loss": 5.6023,
+      "learning_rate": 7.023121387283236e-06,
+      "loss": 6.8582,
       "step": 4381
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.3121387283236993e-06,
-      "loss": 5.7118,
+      "learning_rate": 6.9364161849710975e-06,
+      "loss": 6.9613,
       "step": 4382
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.283236994219653e-06,
-      "loss": 5.6646,
+      "learning_rate": 6.849710982658959e-06,
+      "loss": 7.0067,
       "step": 4383
     },
     {
       "epoch": 4.91,
-      "learning_rate": 2.254335260115607e-06,
-      "loss": 5.7964,
+      "learning_rate": 6.76300578034682e-06,
+      "loss": 7.1529,
       "step": 4384
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.225433526011561e-06,
-      "loss": 5.6142,
+      "learning_rate": 6.676300578034681e-06,
+      "loss": 7.0049,
       "step": 4385
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.1965317919075144e-06,
-      "loss": 5.4314,
+      "learning_rate": 6.589595375722542e-06,
+      "loss": 6.9255,
       "step": 4386
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.1676300578034682e-06,
-      "loss": 5.4338,
+      "learning_rate": 6.502890173410404e-06,
+      "loss": 6.9192,
       "step": 4387
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.138728323699422e-06,
-      "loss": 5.3144,
+      "learning_rate": 6.4161849710982654e-06,
+      "loss": 6.9239,
       "step": 4388
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.109826589595376e-06,
-      "loss": 5.1765,
+      "learning_rate": 6.329479768786127e-06,
+      "loss": 6.8484,
       "step": 4389
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.0809248554913294e-06,
-      "loss": 5.2393,
+      "learning_rate": 6.242774566473988e-06,
+      "loss": 6.8918,
       "step": 4390
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.0520231213872833e-06,
-      "loss": 5.2039,
+      "learning_rate": 6.156069364161849e-06,
+      "loss": 6.9134,
       "step": 4391
     },
     {
       "epoch": 4.92,
-      "learning_rate": 2.023121387283237e-06,
-      "loss": 5.4529,
+      "learning_rate": 6.06936416184971e-06,
+      "loss": 7.0699,
       "step": 4392
     },
     {
       "epoch": 4.92,
-      "learning_rate": 1.994219653179191e-06,
-      "loss": 5.2208,
+      "learning_rate": 5.982658959537572e-06,
+      "loss": 6.8655,
       "step": 4393
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.9653179190751445e-06,
-      "loss": 5.2383,
+      "learning_rate": 5.895953757225433e-06,
+      "loss": 6.9172,
       "step": 4394
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.9364161849710983e-06,
-      "loss": 5.4608,
+      "learning_rate": 5.809248554913295e-06,
+      "loss": 7.0221,
       "step": 4395
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.907514450867052e-06,
-      "loss": 5.3538,
+      "learning_rate": 5.722543352601156e-06,
+      "loss": 7.0247,
       "step": 4396
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.878612716763006e-06,
-      "loss": 5.1917,
+      "learning_rate": 5.635838150289017e-06,
+      "loss": 6.7723,
       "step": 4397
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.8497109826589597e-06,
-      "loss": 5.2691,
+      "learning_rate": 5.549132947976878e-06,
+      "loss": 6.9188,
       "step": 4398
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.8208092485549136e-06,
-      "loss": 5.279,
+      "learning_rate": 5.46242774566474e-06,
+      "loss": 6.9314,
       "step": 4399
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.7919075144508673e-06,
-      "loss": 5.4194,
+      "learning_rate": 5.375722543352601e-06,
+      "loss": 7.0194,
       "step": 4400
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.7630057803468211e-06,
-      "loss": 5.3064,
+      "learning_rate": 5.2890173410404626e-06,
+      "loss": 6.9021,
       "step": 4401
     },
     {
       "epoch": 4.93,
-      "learning_rate": 1.7341040462427746e-06,
-      "loss": 5.1039,
+      "learning_rate": 5.202312138728323e-06,
+      "loss": 6.8366,
       "step": 4402
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.7052023121387282e-06,
-      "loss": 5.1588,
+      "learning_rate": 5.115606936416184e-06,
+      "loss": 6.8947,
       "step": 4403
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.6763005780346821e-06,
-      "loss": 5.1006,
+      "learning_rate": 5.028901734104045e-06,
+      "loss": 6.8538,
       "step": 4404
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.6473988439306358e-06,
-      "loss": 5.1432,
+      "learning_rate": 4.9421965317919065e-06,
+      "loss": 6.9384,
       "step": 4405
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.6184971098265896e-06,
-      "loss": 5.1155,
+      "learning_rate": 4.855491329479768e-06,
+      "loss": 6.9591,
       "step": 4406
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.5895953757225433e-06,
-      "loss": 5.212,
+      "learning_rate": 4.76878612716763e-06,
+      "loss": 7.0154,
       "step": 4407
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.5606936416184972e-06,
-      "loss": 5.1042,
+      "learning_rate": 4.682080924855491e-06,
+      "loss": 6.9845,
       "step": 4408
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.5317919075144508e-06,
-      "loss": 5.0511,
+      "learning_rate": 4.595375722543352e-06,
+      "loss": 6.9611,
       "step": 4409
     },
     {
       "epoch": 4.94,
-      "learning_rate": 1.5028901734104047e-06,
-      "loss": 4.8784,
+      "learning_rate": 4.508670520231213e-06,
+      "loss": 6.7967,
       "step": 4410
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.4739884393063584e-06,
-      "loss": 4.9736,
+      "learning_rate": 4.421965317919074e-06,
+      "loss": 6.9197,
       "step": 4411
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.4450867052023122e-06,
-      "loss": 4.7221,
+      "learning_rate": 4.335260115606936e-06,
+      "loss": 6.8209,
       "step": 4412
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.4161849710982659e-06,
-      "loss": 4.6951,
+      "learning_rate": 4.248554913294798e-06,
+      "loss": 6.8278,
       "step": 4413
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.3872832369942197e-06,
-      "loss": 4.7227,
+      "learning_rate": 4.161849710982659e-06,
+      "loss": 6.9714,
       "step": 4414
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.3583815028901734e-06,
-      "loss": 4.2708,
+      "learning_rate": 4.07514450867052e-06,
+      "loss": 6.7777,
       "step": 4415
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.3294797687861273e-06,
-      "loss": 4.2728,
+      "learning_rate": 3.988439306358381e-06,
+      "loss": 6.8396,
       "step": 4416
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.300578034682081e-06,
-      "loss": 4.1787,
+      "learning_rate": 3.901734104046242e-06,
+      "loss": 6.6862,
       "step": 4417
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.2716763005780348e-06,
-      "loss": 3.8656,
+      "learning_rate": 3.8150289017341036e-06,
+      "loss": 6.702,
       "step": 4418
     },
     {
       "epoch": 4.95,
-      "learning_rate": 1.2427745664739887e-06,
-      "loss": 5.8873,
+      "learning_rate": 3.728323699421965e-06,
+      "loss": 6.866,
       "step": 4419
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.2138728323699421e-06,
-      "loss": 5.8597,
+      "learning_rate": 3.641618497109826e-06,
+      "loss": 6.8809,
       "step": 4420
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.184971098265896e-06,
-      "loss": 6.0573,
+      "learning_rate": 3.554913294797687e-06,
+      "loss": 6.9834,
       "step": 4421
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.1560693641618497e-06,
-      "loss": 5.9539,
+      "learning_rate": 3.4682080924855487e-06,
+      "loss": 6.938,
       "step": 4422
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.1271676300578035e-06,
-      "loss": 5.8813,
+      "learning_rate": 3.38150289017341e-06,
+      "loss": 6.8548,
       "step": 4423
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.0982658959537572e-06,
-      "loss": 5.9151,
+      "learning_rate": 3.294797687861271e-06,
+      "loss": 6.9352,
       "step": 4424
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.069364161849711e-06,
-      "loss": 5.8753,
+      "learning_rate": 3.2080924855491327e-06,
+      "loss": 6.8297,
       "step": 4425
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.0404624277456647e-06,
-      "loss": 5.928,
+      "learning_rate": 3.121387283236994e-06,
+      "loss": 6.9057,
       "step": 4426
     },
     {
       "epoch": 4.96,
-      "learning_rate": 1.0115606936416186e-06,
-      "loss": 5.9403,
+      "learning_rate": 3.034682080924855e-06,
+      "loss": 6.9311,
       "step": 4427
     },
     {
       "epoch": 4.96,
-      "learning_rate": 9.826589595375722e-07,
-      "loss": 5.7143,
+      "learning_rate": 2.9479768786127167e-06,
+      "loss": 6.8329,
       "step": 4428
     },
     {
       "epoch": 4.97,
-      "learning_rate": 9.53757225433526e-07,
-      "loss": 5.8091,
+      "learning_rate": 2.861271676300578e-06,
+      "loss": 6.9484,
       "step": 4429
     },
     {
       "epoch": 4.97,
-      "learning_rate": 9.248554913294799e-07,
-      "loss": 5.7823,
+      "learning_rate": 2.774566473988439e-06,
+      "loss": 6.9079,
       "step": 4430
     },
     {
       "epoch": 4.97,
-      "learning_rate": 8.959537572254336e-07,
-      "loss": 5.8684,
+      "learning_rate": 2.6878612716763007e-06,
+      "loss": 7.0077,
       "step": 4431
     },
     {
       "epoch": 4.97,
-      "learning_rate": 8.670520231213873e-07,
-      "loss": 5.6768,
+      "learning_rate": 2.6011560693641614e-06,
+      "loss": 6.895,
       "step": 4432
     },
     {
       "epoch": 4.97,
-      "learning_rate": 8.381502890173411e-07,
-      "loss": 5.7361,
+      "learning_rate": 2.5144508670520226e-06,
+      "loss": 6.9868,
       "step": 4433
     },
     {
       "epoch": 4.97,
-      "learning_rate": 8.092485549132948e-07,
-      "loss": 5.6718,
+      "learning_rate": 2.427745664739884e-06,
+      "loss": 6.9323,
       "step": 4434
     },
     {
       "epoch": 4.97,
-      "learning_rate": 7.803468208092486e-07,
-      "loss": 5.6932,
+      "learning_rate": 2.3410404624277454e-06,
+      "loss": 6.9826,
       "step": 4435
     },
     {
       "epoch": 4.97,
-      "learning_rate": 7.514450867052023e-07,
-      "loss": 5.5666,
+      "learning_rate": 2.2543352601156066e-06,
+      "loss": 6.9629,
       "step": 4436
     },
     {
       "epoch": 4.97,
-      "learning_rate": 7.225433526011561e-07,
-      "loss": 5.4353,
+      "learning_rate": 2.167630057803468e-06,
+      "loss": 6.8359,
       "step": 4437
     },
     {
       "epoch": 4.98,
-      "learning_rate": 6.936416184971099e-07,
-      "loss": 5.5441,
+      "learning_rate": 2.0809248554913294e-06,
+      "loss": 7.0235,
       "step": 4438
     },
     {
       "epoch": 4.98,
-      "learning_rate": 6.647398843930636e-07,
-      "loss": 5.4843,
+      "learning_rate": 1.9942196531791906e-06,
+      "loss": 6.9299,
       "step": 4439
     },
     {
       "epoch": 4.98,
-      "learning_rate": 6.358381502890174e-07,
-      "loss": 5.5614,
+      "learning_rate": 1.9075144508670518e-06,
+      "loss": 6.9898,
       "step": 4440
     },
     {
       "epoch": 4.98,
-      "learning_rate": 6.069364161849711e-07,
-      "loss": 5.3419,
+      "learning_rate": 1.820809248554913e-06,
+      "loss": 7.0156,
       "step": 4441
     },
     {
       "epoch": 4.98,
-      "learning_rate": 5.780346820809248e-07,
-      "loss": 5.5625,
+      "learning_rate": 1.7341040462427744e-06,
+      "loss": 7.0065,
       "step": 4442
     },
     {
       "epoch": 4.98,
-      "learning_rate": 5.491329479768786e-07,
-      "loss": 5.4491,
+      "learning_rate": 1.6473988439306356e-06,
+      "loss": 7.051,
       "step": 4443
     },
     {
       "epoch": 4.98,
-      "learning_rate": 5.202312138728324e-07,
-      "loss": 5.2859,
+      "learning_rate": 1.560693641618497e-06,
+      "loss": 6.9725,
       "step": 4444
     },
     {
       "epoch": 4.98,
-      "learning_rate": 4.913294797687861e-07,
-      "loss": 5.1455,
+      "learning_rate": 1.4739884393063584e-06,
+      "loss": 6.8502,
       "step": 4445
     },
     {
       "epoch": 4.98,
-      "learning_rate": 4.6242774566473993e-07,
-      "loss": 5.238,
+      "learning_rate": 1.3872832369942195e-06,
+      "loss": 6.9732,
       "step": 4446
     },
     {
       "epoch": 4.99,
-      "learning_rate": 4.3352601156069365e-07,
-      "loss": 5.2683,
+      "learning_rate": 1.3005780346820807e-06,
+      "loss": 7.0538,
       "step": 4447
     },
     {
       "epoch": 4.99,
-      "learning_rate": 4.046242774566474e-07,
-      "loss": 5.4307,
+      "learning_rate": 1.213872832369942e-06,
+      "loss": 7.04,
       "step": 4448
     },
     {
       "epoch": 4.99,
-      "learning_rate": 3.7572254335260117e-07,
-      "loss": 5.0885,
+      "learning_rate": 1.1271676300578033e-06,
+      "loss": 6.8582,
       "step": 4449
     },
     {
       "epoch": 4.99,
-      "learning_rate": 3.4682080924855494e-07,
-      "loss": 5.1458,
+      "learning_rate": 1.0404624277456647e-06,
+      "loss": 6.9343,
       "step": 4450
     },
     {
       "epoch": 4.99,
-      "learning_rate": 3.179190751445087e-07,
-      "loss": 5.0388,
+      "learning_rate": 9.537572254335259e-07,
+      "loss": 6.8676,
       "step": 4451
     },
     {
       "epoch": 4.99,
-      "learning_rate": 2.890173410404624e-07,
-      "loss": 5.1946,
+      "learning_rate": 8.670520231213872e-07,
+      "loss": 6.9336,
       "step": 4452
     },
     {
       "epoch": 4.99,
-      "learning_rate": 2.601156069364162e-07,
-      "loss": 5.1829,
+      "learning_rate": 7.803468208092485e-07,
+      "loss": 7.0925,
       "step": 4453
     },
     {
       "epoch": 4.99,
-      "learning_rate": 2.3121387283236997e-07,
-      "loss": 5.0437,
+      "learning_rate": 6.936416184971098e-07,
+      "loss": 6.9586,
       "step": 4454
     },
     {
       "epoch": 4.99,
-      "learning_rate": 2.023121387283237e-07,
-      "loss": 4.8257,
+      "learning_rate": 6.06936416184971e-07,
+      "loss": 6.8471,
       "step": 4455
     },
     {
       "epoch": 5.0,
-      "learning_rate": 1.7341040462427747e-07,
-      "loss": 4.8584,
+      "learning_rate": 5.202312138728324e-07,
+      "loss": 6.9077,
       "step": 4456
     },
     {
       "epoch": 5.0,
-      "learning_rate": 1.445086705202312e-07,
-      "loss": 4.6384,
+      "learning_rate": 4.335260115606936e-07,
+      "loss": 6.8428,
       "step": 4457
     },
     {
       "epoch": 5.0,
-      "learning_rate": 1.1560693641618498e-07,
-      "loss": 4.2346,
+      "learning_rate": 3.468208092485549e-07,
+      "loss": 6.7026,
       "step": 4458
     },
     {
       "epoch": 5.0,
-      "learning_rate": 8.670520231213873e-08,
-      "loss": 4.6415,
+      "learning_rate": 2.601156069364162e-07,
+      "loss": 6.93,
       "step": 4459
     },
     {
       "epoch": 5.0,
-      "learning_rate": 5.780346820809249e-08,
-      "loss": 3.5204,
+      "learning_rate": 1.7341040462427744e-07,
+      "loss": 6.3299,
       "step": 4460
     },
     {
       "epoch": 5.0,
       "step": 4460,
       "total_flos": 0.0,
-      "train_loss": 6.2593836770463955,
-      "train_runtime": 29013.0161,
-      "train_samples_per_second": 4.918,
-      "train_steps_per_second": 0.154
+      "train_loss": 6.994354631227228,
+      "train_runtime": 29094.1939,
+      "train_samples_per_second": 4.904,
+      "train_steps_per_second": 0.153
     }
   ],
   "max_steps": 4460,