diff --git "a/checkpoint-4582/trainer_state.json" "b/checkpoint-4582/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-4582/trainer_state.json"
@@ -0,0 +1,27517 @@
+{
+  "best_metric": 0.9528456330299377,
+  "best_model_checkpoint": "bird_sounds_classification/checkpoint-4582",
+  "epoch": 1.0,
+  "global_step": 4582,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-08,
+      "loss": 1.3016,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-07,
+      "loss": 1.4236,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8e-07,
+      "loss": 1.031,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4000000000000003e-07,
+      "loss": 2.4174,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 0.961,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.6e-07,
+      "loss": 1.1798,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.2000000000000006e-07,
+      "loss": 1.1328,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.800000000000001e-07,
+      "loss": 1.5953,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.4e-07,
+      "loss": 0.7631,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 1.5506,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.6e-07,
+      "loss": 0.3344,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.2e-07,
+      "loss": 0.1082,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8e-07,
+      "loss": 1.4726,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.400000000000001e-07,
+      "loss": 0.0654,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9e-07,
+      "loss": 1.3277,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.600000000000001e-07,
+      "loss": 1.5969,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0200000000000002e-06,
+      "loss": 2.6363,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.08e-06,
+      "loss": 0.8425,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.14e-06,
+      "loss": 1.1371,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 1.1291,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.26e-06,
+      "loss": 1.7922,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.32e-06,
+      "loss": 0.7448,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3800000000000001e-06,
+      "loss": 0.7833,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.44e-06,
+      "loss": 0.9916,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5e-06,
+      "loss": 1.6177,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.56e-06,
+      "loss": 0.8715,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6200000000000002e-06,
+      "loss": 1.1374,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6800000000000002e-06,
+      "loss": 0.6923,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7399999999999999e-06,
+      "loss": 0.7537,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8e-06,
+      "loss": 0.6535,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.86e-06,
+      "loss": 1.5392,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9200000000000003e-06,
+      "loss": 0.7652,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.98e-06,
+      "loss": 0.8269,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0400000000000004e-06,
+      "loss": 1.9389,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1e-06,
+      "loss": 0.914,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.16e-06,
+      "loss": 4.276,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.22e-06,
+      "loss": 0.3002,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.28e-06,
+      "loss": 2.1748,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.34e-06,
+      "loss": 0.1181,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 0.543,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4599999999999997e-06,
+      "loss": 0.9683,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.52e-06,
+      "loss": 1.3163,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.58e-06,
+      "loss": 0.8825,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.64e-06,
+      "loss": 1.6164,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7e-06,
+      "loss": 1.6605,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7600000000000003e-06,
+      "loss": 0.7514,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.82e-06,
+      "loss": 2.4526,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.88e-06,
+      "loss": 1.8677,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9400000000000002e-06,
+      "loss": 1.051,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 0.3322,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.999338040600177e-06,
+      "loss": 1.6537,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.998676081200353e-06,
+      "loss": 1.2124,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.99801412180053e-06,
+      "loss": 0.8706,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.997352162400706e-06,
+      "loss": 0.4436,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.996690203000883e-06,
+      "loss": 0.116,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.996028243601059e-06,
+      "loss": 0.3678,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.995366284201236e-06,
+      "loss": 0.7264,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9947043248014125e-06,
+      "loss": 1.9573,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.994042365401589e-06,
+      "loss": 1.657,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.993380406001765e-06,
+      "loss": 1.1507,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.992718446601942e-06,
+      "loss": 0.6664,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9920564872021185e-06,
+      "loss": 0.903,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.991394527802295e-06,
+      "loss": 0.6758,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9907325684024715e-06,
+      "loss": 1.904,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.990070609002648e-06,
+      "loss": 0.6578,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9894086496028245e-06,
+      "loss": 2.2447,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.988746690203001e-06,
+      "loss": 1.7526,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9880847308031775e-06,
+      "loss": 1.4045,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9874227714033542e-06,
+      "loss": 1.895,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9867608120035305e-06,
+      "loss": 0.7105,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.986098852603707e-06,
+      "loss": 2.446,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9854368932038835e-06,
+      "loss": 1.734,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9847749338040602e-06,
+      "loss": 3.6684,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9841129744042365e-06,
+      "loss": 1.2918,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9834510150044132e-06,
+      "loss": 1.2678,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9827890556045895e-06,
+      "loss": 0.4808,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9821270962047662e-06,
+      "loss": 1.1776,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9814651368049425e-06,
+      "loss": 1.8976,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9808031774051192e-06,
+      "loss": 3.0657,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.980141218005296e-06,
+      "loss": 0.2343,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9794792586054722e-06,
+      "loss": 0.9544,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9788172992056485e-06,
+      "loss": 2.1713,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9781553398058252e-06,
+      "loss": 1.9033,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.977493380406002e-06,
+      "loss": 0.0902,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9768314210061782e-06,
+      "loss": 0.7747,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.976169461606355e-06,
+      "loss": 0.7256,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9755075022065317e-06,
+      "loss": 0.091,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.974845542806708e-06,
+      "loss": 0.5707,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9741835834068842e-06,
+      "loss": 1.4559,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.973521624007061e-06,
+      "loss": 4.3329,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9728596646072377e-06,
+      "loss": 1.457,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9721977052074144e-06,
+      "loss": 1.7977,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9715357458075907e-06,
+      "loss": 3.2639,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.970873786407767e-06,
+      "loss": 1.7403,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9702118270079437e-06,
+      "loss": 1.3468,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.96954986760812e-06,
+      "loss": 0.4561,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9688879082082967e-06,
+      "loss": 1.0686,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9682259488084734e-06,
+      "loss": 0.9515,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9675639894086497e-06,
+      "loss": 0.0696,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.966902030008826e-06,
+      "loss": 2.4785,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9662400706090027e-06,
+      "loss": 0.9754,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9655781112091794e-06,
+      "loss": 0.1931,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9649161518093557e-06,
+      "loss": 0.2905,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9642541924095324e-06,
+      "loss": 1.7158,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9635922330097087e-06,
+      "loss": 0.2768,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9629302736098854e-06,
+      "loss": 0.2711,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9622683142100617e-06,
+      "loss": 2.2652,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9616063548102384e-06,
+      "loss": 0.7651,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.960944395410415e-06,
+      "loss": 2.1651,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9602824360105914e-06,
+      "loss": 0.7463,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9596204766107677e-06,
+      "loss": 1.4135,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9589585172109444e-06,
+      "loss": 1.3253,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.958296557811121e-06,
+      "loss": 2.207,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.9576345984112974e-06,
+      "loss": 1.888,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.956972639011474e-06,
+      "loss": 1.3512,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9563106796116504e-06,
+      "loss": 1.3246,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.955648720211827e-06,
+      "loss": 0.7346,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9549867608120034e-06,
+      "loss": 1.0946,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.95432480141218e-06,
+      "loss": 1.6785,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.953662842012357e-06,
+      "loss": 0.804,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9530008826125335e-06,
+      "loss": 1.7811,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9523389232127094e-06,
+      "loss": 0.7188,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.951676963812886e-06,
+      "loss": 0.6387,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.951015004413063e-06,
+      "loss": 1.1566,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.950353045013239e-06,
+      "loss": 0.6625,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.949691085613416e-06,
+      "loss": 0.8042,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9490291262135925e-06,
+      "loss": 0.8896,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.948367166813769e-06,
+      "loss": 1.2047,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.947705207413945e-06,
+      "loss": 1.0578,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.947043248014122e-06,
+      "loss": 1.1998,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9463812886142985e-06,
+      "loss": 2.5471,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9457193292144752e-06,
+      "loss": 0.9397,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9450573698146515e-06,
+      "loss": 3.0599,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.944395410414828e-06,
+      "loss": 0.9478,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9437334510150045e-06,
+      "loss": 1.8745,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.943071491615181e-06,
+      "loss": 1.5782,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9424095322153575e-06,
+      "loss": 1.537,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9417475728155342e-06,
+      "loss": 1.9965,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9410856134157105e-06,
+      "loss": 1.1491,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.940423654015887e-06,
+      "loss": 2.5614,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9397616946160635e-06,
+      "loss": 0.8887,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9390997352162403e-06,
+      "loss": 0.8997,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.938437775816417e-06,
+      "loss": 1.5019,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9377758164165933e-06,
+      "loss": 1.7895,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9371138570167695e-06,
+      "loss": 1.0543,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9364518976169463e-06,
+      "loss": 0.6248,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9357899382171225e-06,
+      "loss": 0.7041,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9351279788172993e-06,
+      "loss": 0.7234,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.934466019417476e-06,
+      "loss": 0.5685,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9338040600176523e-06,
+      "loss": 1.542,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9331421006178285e-06,
+      "loss": 1.5487,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9324801412180053e-06,
+      "loss": 0.8602,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.931818181818182e-06,
+      "loss": 1.4722,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9311562224183587e-06,
+      "loss": 0.7952,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.930494263018535e-06,
+      "loss": 1.3398,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9298323036187113e-06,
+      "loss": 1.8289,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.929170344218888e-06,
+      "loss": 1.0958,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9285083848190643e-06,
+      "loss": 1.3023,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.927846425419241e-06,
+      "loss": 1.3858,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.9271844660194177e-06,
+      "loss": 2.0207,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9265225066195944e-06,
+      "loss": 2.3563,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9258605472197707e-06,
+      "loss": 0.1415,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.925198587819947e-06,
+      "loss": 0.5377,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9245366284201237e-06,
+      "loss": 0.7538,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9238746690203004e-06,
+      "loss": 1.9679,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9232127096204767e-06,
+      "loss": 1.1835,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9225507502206534e-06,
+      "loss": 1.611,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9218887908208297e-06,
+      "loss": 0.6546,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.921226831421006e-06,
+      "loss": 1.0037,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9205648720211827e-06,
+      "loss": 0.1967,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9199029126213594e-06,
+      "loss": 1.3726,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.919240953221536e-06,
+      "loss": 0.8055,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9185789938217124e-06,
+      "loss": 1.6406,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9179170344218887e-06,
+      "loss": 0.8978,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9172550750220654e-06,
+      "loss": 1.7348,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.916593115622242e-06,
+      "loss": 0.6842,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9159311562224184e-06,
+      "loss": 1.3107,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.915269196822595e-06,
+      "loss": 0.0946,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9146072374227714e-06,
+      "loss": 1.5981,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9139452780229477e-06,
+      "loss": 0.9081,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9132833186231244e-06,
+      "loss": 0.1906,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.912621359223301e-06,
+      "loss": 1.0728,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.911959399823478e-06,
+      "loss": 0.3097,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.911297440423654e-06,
+      "loss": 0.9772,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9106354810238304e-06,
+      "loss": 0.5447,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.909973521624007e-06,
+      "loss": 0.5134,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.909311562224184e-06,
+      "loss": 0.7792,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.90864960282436e-06,
+      "loss": 1.2554,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.907987643424537e-06,
+      "loss": 2.5754,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.907325684024713e-06,
+      "loss": 1.0341,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9066637246248894e-06,
+      "loss": 0.4566,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.906001765225066e-06,
+      "loss": 1.3424,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.905339805825243e-06,
+      "loss": 1.9094,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9046778464254196e-06,
+      "loss": 1.1521,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.904015887025596e-06,
+      "loss": 1.6677,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.903353927625772e-06,
+      "loss": 2.4829,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.902691968225949e-06,
+      "loss": 0.532,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9020300088261256e-06,
+      "loss": 0.4668,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.901368049426302e-06,
+      "loss": 0.8762,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9007060900264786e-06,
+      "loss": 2.4188,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9000441306266553e-06,
+      "loss": 1.4375,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8993821712268316e-06,
+      "loss": 0.809,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.898720211827008e-06,
+      "loss": 0.8304,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8980582524271846e-06,
+      "loss": 3.1678,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8973962930273613e-06,
+      "loss": 2.0071,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8967343336275376e-06,
+      "loss": 1.3518,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8960723742277143e-06,
+      "loss": 2.1771,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8954104148278906e-06,
+      "loss": 0.8628,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.894748455428067e-06,
+      "loss": 0.1327,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8940864960282436e-06,
+      "loss": 1.2521,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8934245366284203e-06,
+      "loss": 1.8163,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.892762577228597e-06,
+      "loss": 0.4327,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8921006178287733e-06,
+      "loss": 1.0353,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8914386584289496e-06,
+      "loss": 3.0186,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8907766990291263e-06,
+      "loss": 1.3149,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.890114739629303e-06,
+      "loss": 0.778,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8894527802294793e-06,
+      "loss": 1.4993,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.888790820829656e-06,
+      "loss": 0.121,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8881288614298323e-06,
+      "loss": 2.0121,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8874669020300086e-06,
+      "loss": 1.7496,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8868049426301853e-06,
+      "loss": 0.2019,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.886142983230362e-06,
+      "loss": 0.4892,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8854810238305387e-06,
+      "loss": 1.5198,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.884819064430715e-06,
+      "loss": 2.0653,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8841571050308913e-06,
+      "loss": 1.7733,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.883495145631068e-06,
+      "loss": 1.8564,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8828331862312447e-06,
+      "loss": 0.7633,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.882171226831421e-06,
+      "loss": 0.7536,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8815092674315977e-06,
+      "loss": 1.1069,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.880847308031774e-06,
+      "loss": 0.9144,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8801853486319503e-06,
+      "loss": 2.2805,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.879523389232127e-06,
+      "loss": 1.8066,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8788614298323037e-06,
+      "loss": 2.8787,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8781994704324804e-06,
+      "loss": 1.2,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8775375110326567e-06,
+      "loss": 1.6133,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8768755516328334e-06,
+      "loss": 0.9984,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8762135922330097e-06,
+      "loss": 1.6893,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8755516328331864e-06,
+      "loss": 0.9049,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8748896734333627e-06,
+      "loss": 1.4849,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8742277140335394e-06,
+      "loss": 0.747,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.873565754633716e-06,
+      "loss": 0.6171,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8729037952338924e-06,
+      "loss": 0.3417,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8722418358340687e-06,
+      "loss": 0.8099,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8715798764342454e-06,
+      "loss": 1.6341,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.870917917034422e-06,
+      "loss": 1.1256,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8702559576345984e-06,
+      "loss": 1.3505,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.869593998234775e-06,
+      "loss": 1.164,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8689320388349514e-06,
+      "loss": 0.9342,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.868270079435128e-06,
+      "loss": 0.2415,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.8676081200353044e-06,
+      "loss": 1.47,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.866946160635481e-06,
+      "loss": 0.7119,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 2.866284201235658e-06,
+      "loss": 2.2288,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.865622241835834e-06,
+      "loss": 0.693,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8649602824360104e-06,
+      "loss": 2.452,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.864298323036187e-06,
+      "loss": 1.1401,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.863636363636364e-06,
+      "loss": 1.007,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.86297440423654e-06,
+      "loss": 1.2347,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.862312444836717e-06,
+      "loss": 0.0888,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.861650485436893e-06,
+      "loss": 0.6648,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.86098852603707e-06,
+      "loss": 2.0792,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.860326566637246e-06,
+      "loss": 0.8639,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.859664607237423e-06,
+      "loss": 1.2011,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8590026478375996e-06,
+      "loss": 0.7192,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.858340688437776e-06,
+      "loss": 2.0624,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.857678729037952e-06,
+      "loss": 0.3914,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.857016769638129e-06,
+      "loss": 1.1975,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8563548102383056e-06,
+      "loss": 0.2286,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.855692850838482e-06,
+      "loss": 1.3226,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8550308914386586e-06,
+      "loss": 0.3202,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8543689320388353e-06,
+      "loss": 0.3663,
+      "step": 270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8537069726390116e-06,
+      "loss": 1.766,
+      "step": 271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.853045013239188e-06,
+      "loss": 1.3911,
+      "step": 272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8523830538393646e-06,
+      "loss": 1.4815,
+      "step": 273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8517210944395413e-06,
+      "loss": 1.6067,
+      "step": 274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8510591350397176e-06,
+      "loss": 0.7553,
+      "step": 275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8503971756398943e-06,
+      "loss": 0.0856,
+      "step": 276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8497352162400706e-06,
+      "loss": 1.8525,
+      "step": 277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8490732568402473e-06,
+      "loss": 4.1202,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8484112974404236e-06,
+      "loss": 1.1763,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8477493380406003e-06,
+      "loss": 2.6881,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.847087378640777e-06,
+      "loss": 1.1994,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8464254192409533e-06,
+      "loss": 1.7149,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8457634598411296e-06,
+      "loss": 1.404,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8451015004413063e-06,
+      "loss": 1.1393,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.844439541041483e-06,
+      "loss": 0.2008,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8437775816416593e-06,
+      "loss": 2.6246,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.843115622241836e-06,
+      "loss": 1.6099,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8424536628420123e-06,
+      "loss": 1.724,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.841791703442189e-06,
+      "loss": 0.1002,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8411297440423653e-06,
+      "loss": 2.6718,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.840467784642542e-06,
+      "loss": 2.8884,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8398058252427187e-06,
+      "loss": 1.312,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.839143865842895e-06,
+      "loss": 0.1082,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8384819064430713e-06,
+      "loss": 0.9424,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.837819947043248e-06,
+      "loss": 0.4412,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.8371579876434247e-06,
+      "loss": 0.2138,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 2.836496028243601e-06,
+      "loss": 0.8596,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8358340688437777e-06,
+      "loss": 2.1219,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.835172109443954e-06,
+      "loss": 0.8733,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8345101500441307e-06,
+      "loss": 0.9414,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.833848190644307e-06,
+      "loss": 1.3569,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8331862312444837e-06,
+      "loss": 1.1386,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8325242718446605e-06,
+      "loss": 0.9315,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.831862312444837e-06,
+      "loss": 0.7443,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.831200353045013e-06,
+      "loss": 0.8524,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8305383936451897e-06,
+      "loss": 0.9231,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8298764342453665e-06,
+      "loss": 1.861,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8292144748455427e-06,
+      "loss": 0.2756,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8285525154457195e-06,
+      "loss": 0.8732,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.827890556045896e-06,
+      "loss": 1.2429,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8272285966460725e-06,
+      "loss": 1.4515,
+      "step": 311
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8265666372462487e-06,
+      "loss": 1.0918,
+      "step": 312
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8259046778464255e-06,
+      "loss": 1.1565,
+      "step": 313
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.825242718446602e-06,
+      "loss": 1.9076,
+      "step": 314
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8245807590467785e-06,
+      "loss": 1.0669,
+      "step": 315
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.823918799646955e-06,
+      "loss": 1.5951,
+      "step": 316
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8232568402471315e-06,
+      "loss": 0.7125,
+      "step": 317
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.822594880847308e-06,
+      "loss": 1.1355,
+      "step": 318
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8219329214474845e-06,
+      "loss": 0.3368,
+      "step": 319
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.821270962047661e-06,
+      "loss": 0.7743,
+      "step": 320
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.820609002647838e-06,
+      "loss": 1.2419,
+      "step": 321
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.819947043248014e-06,
+      "loss": 1.4227,
+      "step": 322
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8192850838481905e-06,
+      "loss": 0.73,
+      "step": 323
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.818623124448367e-06,
+      "loss": 2.2101,
+      "step": 324
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.817961165048544e-06,
+      "loss": 0.7819,
+      "step": 325
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.81729920564872e-06,
+      "loss": 0.6599,
+      "step": 326
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.816637246248897e-06,
+      "loss": 2.5375,
+      "step": 327
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.815975286849073e-06,
+      "loss": 0.9892,
+      "step": 328
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.81531332744925e-06,
+      "loss": 1.0312,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.814651368049426e-06,
+      "loss": 2.0213,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.813989408649603e-06,
+      "loss": 2.0023,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8133274492497796e-06,
+      "loss": 0.71,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.812665489849956e-06,
+      "loss": 0.2649,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.812003530450132e-06,
+      "loss": 1.2631,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.811341571050309e-06,
+      "loss": 0.5885,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8106796116504856e-06,
+      "loss": 1.8286,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.810017652250662e-06,
+      "loss": 0.2251,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8093556928508386e-06,
+      "loss": 0.8245,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.808693733451015e-06,
+      "loss": 2.4984,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8080317740511916e-06,
+      "loss": 1.0062,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.807369814651368e-06,
+      "loss": 0.5792,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8067078552515446e-06,
+      "loss": 0.9694,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 2.8060458958517213e-06,
+      "loss": 1.372,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.805383936451898e-06,
+      "loss": 1.6435,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8047219770520743e-06,
+      "loss": 1.8618,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8040600176522506e-06,
+      "loss": 1.1243,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8033980582524273e-06,
+      "loss": 0.6017,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8027360988526036e-06,
+      "loss": 0.1267,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8020741394527803e-06,
+      "loss": 1.4159,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.801412180052957e-06,
+      "loss": 2.1181,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8007502206531333e-06,
+      "loss": 1.3465,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.8000882612533096e-06,
+      "loss": 1.1341,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7994263018534863e-06,
+      "loss": 0.7212,
+      "step": 353
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.798764342453663e-06,
+      "loss": 0.8334,
+      "step": 354
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7981023830538398e-06,
+      "loss": 0.8057,
+      "step": 355
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.797440423654016e-06,
+      "loss": 1.7514,
+      "step": 356
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7967784642541923e-06,
+      "loss": 2.1322,
+      "step": 357
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.796116504854369e-06,
+      "loss": 0.7938,
+      "step": 358
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7954545454545453e-06,
+      "loss": 1.502,
+      "step": 359
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.794792586054722e-06,
+      "loss": 1.8222,
+      "step": 360
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7941306266548988e-06,
+      "loss": 1.0533,
+      "step": 361
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.793468667255075e-06,
+      "loss": 2.6728,
+      "step": 362
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7928067078552513e-06,
+      "loss": 0.753,
+      "step": 363
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.792144748455428e-06,
+      "loss": 0.4562,
+      "step": 364
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7914827890556048e-06,
+      "loss": 0.196,
+      "step": 365
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7908208296557815e-06,
+      "loss": 0.1674,
+      "step": 366
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7901588702559578e-06,
+      "loss": 1.073,
+      "step": 367
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.789496910856134e-06,
+      "loss": 0.7876,
+      "step": 368
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7888349514563108e-06,
+      "loss": 1.583,
+      "step": 369
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.788172992056487e-06,
+      "loss": 0.2252,
+      "step": 370
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7875110326566638e-06,
+      "loss": 1.3968,
+      "step": 371
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7868490732568405e-06,
+      "loss": 2.7114,
+      "step": 372
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7861871138570168e-06,
+      "loss": 0.6831,
+      "step": 373
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.785525154457193e-06,
+      "loss": 0.6266,
+      "step": 374
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7848631950573698e-06,
+      "loss": 0.0707,
+      "step": 375
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7842012356575465e-06,
+      "loss": 0.8268,
+      "step": 376
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.783539276257723e-06,
+      "loss": 1.2988,
+      "step": 377
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7828773168578995e-06,
+      "loss": 0.4914,
+      "step": 378
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.782215357458076e-06,
+      "loss": 1.5653,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7815533980582525e-06,
+      "loss": 0.5606,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7808914386584288e-06,
+      "loss": 0.3238,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7802294792586055e-06,
+      "loss": 1.7976,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.779567519858782e-06,
+      "loss": 0.8534,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.778905560458959e-06,
+      "loss": 0.7806,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.778243601059135e-06,
+      "loss": 0.9355,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.7775816416593115e-06,
+      "loss": 1.1393,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.776919682259488e-06,
+      "loss": 0.7252,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.776257722859665e-06,
+      "loss": 1.6334,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 2.775595763459841e-06,
+      "loss": 1.2011,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.774933804060018e-06,
+      "loss": 1.3402,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.774271844660194e-06,
+      "loss": 0.5156,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7736098852603705e-06,
+      "loss": 1.0406,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.772947925860547e-06,
+      "loss": 1.5114,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.772285966460724e-06,
+      "loss": 0.5736,
+      "step": 394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7716240070609006e-06,
+      "loss": 1.0089,
+      "step": 395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.770962047661077e-06,
+      "loss": 1.8862,
+      "step": 396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.770300088261253e-06,
+      "loss": 0.6021,
+      "step": 397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.76963812886143e-06,
+      "loss": 0.9081,
+      "step": 398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7689761694616066e-06,
+      "loss": 1.4438,
+      "step": 399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.768314210061783e-06,
+      "loss": 0.3868,
+      "step": 400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7676522506619596e-06,
+      "loss": 0.3121,
+      "step": 401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.766990291262136e-06,
+      "loss": 0.7848,
+      "step": 402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.766328331862312e-06,
+      "loss": 0.9484,
+      "step": 403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.765666372462489e-06,
+      "loss": 1.1321,
+      "step": 404
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7650044130626656e-06,
+      "loss": 1.4395,
+      "step": 405
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7643424536628423e-06,
+      "loss": 0.7045,
+      "step": 406
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7636804942630186e-06,
+      "loss": 1.6461,
+      "step": 407
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.763018534863195e-06,
+      "loss": 1.5361,
+      "step": 408
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7623565754633716e-06,
+      "loss": 0.7872,
+      "step": 409
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7616946160635483e-06,
+      "loss": 1.283,
+      "step": 410
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7610326566637246e-06,
+      "loss": 2.5007,
+      "step": 411
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7603706972639013e-06,
+      "loss": 0.7097,
+      "step": 412
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7597087378640776e-06,
+      "loss": 1.2965,
+      "step": 413
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.759046778464254e-06,
+      "loss": 2.3918,
+      "step": 414
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7583848190644306e-06,
+      "loss": 1.6002,
+      "step": 415
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7577228596646074e-06,
+      "loss": 3.7935,
+      "step": 416
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.757060900264784e-06,
+      "loss": 2.0966,
+      "step": 417
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7563989408649604e-06,
+      "loss": 0.0628,
+      "step": 418
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.755736981465137e-06,
+      "loss": 1.5312,
+      "step": 419
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7550750220653134e-06,
+      "loss": 1.3472,
+      "step": 420
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7544130626654896e-06,
+      "loss": 2.1064,
+      "step": 421
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7537511032656664e-06,
+      "loss": 0.8596,
+      "step": 422
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.753089143865843e-06,
+      "loss": 0.2261,
+      "step": 423
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7524271844660198e-06,
+      "loss": 0.8981,
+      "step": 424
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.751765225066196e-06,
+      "loss": 1.3047,
+      "step": 425
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7511032656663724e-06,
+      "loss": 1.5549,
+      "step": 426
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.750441306266549e-06,
+      "loss": 0.5695,
+      "step": 427
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7497793468667258e-06,
+      "loss": 1.1486,
+      "step": 428
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.749117387466902e-06,
+      "loss": 0.7093,
+      "step": 429
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7484554280670788e-06,
+      "loss": 0.1725,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.747793468667255e-06,
+      "loss": 1.0246,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7471315092674314e-06,
+      "loss": 0.5287,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.746469549867608e-06,
+      "loss": 0.9601,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7458075904677848e-06,
+      "loss": 2.0528,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2.7451456310679615e-06,
+      "loss": 0.8316,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7444836716681378e-06,
+      "loss": 0.1389,
+      "step": 436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.743821712268314e-06,
+      "loss": 0.9909,
+      "step": 437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7431597528684908e-06,
+      "loss": 0.9701,
+      "step": 438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7424977934686675e-06,
+      "loss": 0.8017,
+      "step": 439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.741835834068844e-06,
+      "loss": 0.298,
+      "step": 440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7411738746690205e-06,
+      "loss": 0.992,
+      "step": 441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.740511915269197e-06,
+      "loss": 2.8481,
+      "step": 442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.739849955869373e-06,
+      "loss": 0.9085,
+      "step": 443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.73918799646955e-06,
+      "loss": 1.1185,
+      "step": 444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7385260370697265e-06,
+      "loss": 2.8875,
+      "step": 445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7378640776699032e-06,
+      "loss": 2.6061,
+      "step": 446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7372021182700795e-06,
+      "loss": 1.3725,
+      "step": 447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.736540158870256e-06,
+      "loss": 0.798,
+      "step": 448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7358781994704325e-06,
+      "loss": 0.5447,
+      "step": 449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7352162400706092e-06,
+      "loss": 1.1723,
+      "step": 450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7345542806707855e-06,
+      "loss": 0.6111,
+      "step": 451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7338923212709622e-06,
+      "loss": 1.0693,
+      "step": 452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.733230361871139e-06,
+      "loss": 0.9672,
+      "step": 453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.732568402471315e-06,
+      "loss": 2.904,
+      "step": 454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7319064430714915e-06,
+      "loss": 0.199,
+      "step": 455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7312444836716682e-06,
+      "loss": 1.1961,
+      "step": 456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.730582524271845e-06,
+      "loss": 1.8706,
+      "step": 457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7299205648720212e-06,
+      "loss": 0.636,
+      "step": 458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.729258605472198e-06,
+      "loss": 1.0129,
+      "step": 459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7285966460723742e-06,
+      "loss": 0.545,
+      "step": 460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.727934686672551e-06,
+      "loss": 1.0732,
+      "step": 461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7272727272727272e-06,
+      "loss": 2.1969,
+      "step": 462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.726610767872904e-06,
+      "loss": 2.4366,
+      "step": 463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7259488084730807e-06,
+      "loss": 1.0323,
+      "step": 464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.725286849073257e-06,
+      "loss": 1.6548,
+      "step": 465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7246248896734332e-06,
+      "loss": 1.5419,
+      "step": 466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.72396293027361e-06,
+      "loss": 0.9024,
+      "step": 467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7233009708737867e-06,
+      "loss": 1.9403,
+      "step": 468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.722639011473963e-06,
+      "loss": 1.0786,
+      "step": 469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7219770520741397e-06,
+      "loss": 0.289,
+      "step": 470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.721315092674316e-06,
+      "loss": 2.5279,
+      "step": 471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7206531332744927e-06,
+      "loss": 2.0734,
+      "step": 472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.719991173874669e-06,
+      "loss": 1.0211,
+      "step": 473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7193292144748457e-06,
+      "loss": 0.9226,
+      "step": 474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7186672550750224e-06,
+      "loss": 1.3909,
+      "step": 475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7180052956751987e-06,
+      "loss": 1.4398,
+      "step": 476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.717343336275375e-06,
+      "loss": 1.5346,
+      "step": 477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7166813768755517e-06,
+      "loss": 1.3987,
+      "step": 478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7160194174757284e-06,
+      "loss": 0.7194,
+      "step": 479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7153574580759047e-06,
+      "loss": 2.4612,
+      "step": 480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 2.7146954986760814e-06,
+      "loss": 0.6474,
+      "step": 481
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7140335392762577e-06,
+      "loss": 2.5414,
+      "step": 482
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7133715798764344e-06,
+      "loss": 1.0413,
+      "step": 483
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7127096204766107e-06,
+      "loss": 1.9079,
+      "step": 484
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7120476610767874e-06,
+      "loss": 1.0119,
+      "step": 485
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.711385701676964e-06,
+      "loss": 0.7787,
+      "step": 486
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7107237422771404e-06,
+      "loss": 0.5946,
+      "step": 487
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7100617828773167e-06,
+      "loss": 2.53,
+      "step": 488
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7093998234774934e-06,
+      "loss": 0.9402,
+      "step": 489
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.70873786407767e-06,
+      "loss": 1.4443,
+      "step": 490
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7080759046778464e-06,
+      "loss": 0.881,
+      "step": 491
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.707413945278023e-06,
+      "loss": 1.4583,
+      "step": 492
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7067519858782e-06,
+      "loss": 0.674,
+      "step": 493
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.706090026478376e-06,
+      "loss": 0.7792,
+      "step": 494
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.7054280670785524e-06,
+      "loss": 0.5651,
+      "step": 495
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.704766107678729e-06,
+      "loss": 2.0574,
+      "step": 496
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.704104148278906e-06,
+      "loss": 0.0998,
+      "step": 497
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.703442188879082e-06,
+      "loss": 1.9255,
+      "step": 498
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.702780229479259e-06,
+      "loss": 0.4631,
+      "step": 499
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.702118270079435e-06,
+      "loss": 2.9048,
+      "step": 500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.701456310679612e-06,
+      "loss": 0.8063,
+      "step": 501
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.700794351279788e-06,
+      "loss": 0.2929,
+      "step": 502
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.700132391879965e-06,
+      "loss": 1.594,
+      "step": 503
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6994704324801415e-06,
+      "loss": 1.4014,
+      "step": 504
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.698808473080318e-06,
+      "loss": 2.5849,
+      "step": 505
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.698146513680494e-06,
+      "loss": 2.3135,
+      "step": 506
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.697484554280671e-06,
+      "loss": 1.0497,
+      "step": 507
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6968225948808475e-06,
+      "loss": 2.0599,
+      "step": 508
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.696160635481024e-06,
+      "loss": 0.6362,
+      "step": 509
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6954986760812005e-06,
+      "loss": 0.1837,
+      "step": 510
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.694836716681377e-06,
+      "loss": 1.9243,
+      "step": 511
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6941747572815535e-06,
+      "loss": 0.101,
+      "step": 512
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.69351279788173e-06,
+      "loss": 0.0848,
+      "step": 513
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6928508384819065e-06,
+      "loss": 1.1513,
+      "step": 514
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6921888790820832e-06,
+      "loss": 1.4008,
+      "step": 515
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6915269196822595e-06,
+      "loss": 0.9722,
+      "step": 516
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.690864960282436e-06,
+      "loss": 0.9034,
+      "step": 517
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6902030008826125e-06,
+      "loss": 1.3024,
+      "step": 518
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6895410414827892e-06,
+      "loss": 0.9437,
+      "step": 519
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6888790820829655e-06,
+      "loss": 0.9226,
+      "step": 520
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6882171226831422e-06,
+      "loss": 0.1121,
+      "step": 521
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6875551632833185e-06,
+      "loss": 0.1445,
+      "step": 522
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6868932038834952e-06,
+      "loss": 0.6045,
+      "step": 523
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6862312444836715e-06,
+      "loss": 0.5035,
+      "step": 524
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.6855692850838482e-06,
+      "loss": 0.8099,
+      "step": 525
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 2.684907325684025e-06,
+      "loss": 1.6568,
+      "step": 526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6842453662842012e-06,
+      "loss": 0.409,
+      "step": 527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.683583406884378e-06,
+      "loss": 1.7489,
+      "step": 528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6829214474845543e-06,
+      "loss": 2.1626,
+      "step": 529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.682259488084731e-06,
+      "loss": 0.746,
+      "step": 530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6815975286849073e-06,
+      "loss": 1.1719,
+      "step": 531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.680935569285084e-06,
+      "loss": 1.2424,
+      "step": 532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6802736098852607e-06,
+      "loss": 0.8405,
+      "step": 533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.679611650485437e-06,
+      "loss": 1.1679,
+      "step": 534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6789496910856133e-06,
+      "loss": 0.7852,
+      "step": 535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.67828773168579e-06,
+      "loss": 1.0502,
+      "step": 536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6776257722859667e-06,
+      "loss": 1.0163,
+      "step": 537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.676963812886143e-06,
+      "loss": 1.3728,
+      "step": 538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6763018534863197e-06,
+      "loss": 0.5199,
+      "step": 539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.675639894086496e-06,
+      "loss": 0.741,
+      "step": 540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6749779346866727e-06,
+      "loss": 0.5359,
+      "step": 541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.674315975286849e-06,
+      "loss": 1.4422,
+      "step": 542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6736540158870257e-06,
+      "loss": 0.1341,
+      "step": 543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6729920564872024e-06,
+      "loss": 1.8856,
+      "step": 544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6723300970873787e-06,
+      "loss": 0.3185,
+      "step": 545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.671668137687555e-06,
+      "loss": 0.3681,
+      "step": 546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6710061782877317e-06,
+      "loss": 1.3595,
+      "step": 547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6703442188879084e-06,
+      "loss": 1.003,
+      "step": 548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6696822594880847e-06,
+      "loss": 0.2269,
+      "step": 549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6690203000882614e-06,
+      "loss": 0.0641,
+      "step": 550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6683583406884377e-06,
+      "loss": 0.0587,
+      "step": 551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6676963812886144e-06,
+      "loss": 1.8584,
+      "step": 552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6670344218887907e-06,
+      "loss": 1.5839,
+      "step": 553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6663724624889674e-06,
+      "loss": 0.5675,
+      "step": 554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.665710503089144e-06,
+      "loss": 1.3251,
+      "step": 555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6650485436893204e-06,
+      "loss": 2.1321,
+      "step": 556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6643865842894967e-06,
+      "loss": 1.5222,
+      "step": 557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6637246248896734e-06,
+      "loss": 0.8931,
+      "step": 558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.66306266548985e-06,
+      "loss": 1.2496,
+      "step": 559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6624007060900264e-06,
+      "loss": 0.5612,
+      "step": 560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.661738746690203e-06,
+      "loss": 2.8023,
+      "step": 561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.66107678729038e-06,
+      "loss": 3.0269,
+      "step": 562
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.660414827890556e-06,
+      "loss": 2.2442,
+      "step": 563
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6597528684907324e-06,
+      "loss": 1.5636,
+      "step": 564
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.659090909090909e-06,
+      "loss": 0.9445,
+      "step": 565
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.658428949691086e-06,
+      "loss": 1.276,
+      "step": 566
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.6577669902912625e-06,
+      "loss": 1.321,
+      "step": 567
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.657105030891439e-06,
+      "loss": 1.4154,
+      "step": 568
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.656443071491615e-06,
+      "loss": 0.1928,
+      "step": 569
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.655781112091792e-06,
+      "loss": 1.3474,
+      "step": 570
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.655119152691968e-06,
+      "loss": 0.6926,
+      "step": 571
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 2.654457193292145e-06,
+      "loss": 0.973,
+      "step": 572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6537952338923215e-06,
+      "loss": 0.5534,
+      "step": 573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.653133274492498e-06,
+      "loss": 0.4546,
+      "step": 574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.652471315092674e-06,
+      "loss": 0.0699,
+      "step": 575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.651809355692851e-06,
+      "loss": 2.4839,
+      "step": 576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6511473962930276e-06,
+      "loss": 2.0125,
+      "step": 577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6504854368932043e-06,
+      "loss": 0.6736,
+      "step": 578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6498234774933806e-06,
+      "loss": 1.0993,
+      "step": 579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.649161518093557e-06,
+      "loss": 1.7612,
+      "step": 580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6484995586937336e-06,
+      "loss": 0.5727,
+      "step": 581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.64783759929391e-06,
+      "loss": 1.5075,
+      "step": 582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6471756398940866e-06,
+      "loss": 0.4696,
+      "step": 583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6465136804942633e-06,
+      "loss": 1.1583,
+      "step": 584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6458517210944396e-06,
+      "loss": 0.642,
+      "step": 585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.645189761694616e-06,
+      "loss": 1.0197,
+      "step": 586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6445278022947926e-06,
+      "loss": 0.7517,
+      "step": 587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6438658428949693e-06,
+      "loss": 1.2439,
+      "step": 588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.643203883495146e-06,
+      "loss": 1.6651,
+      "step": 589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6425419240953223e-06,
+      "loss": 2.0884,
+      "step": 590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6418799646954986e-06,
+      "loss": 1.0551,
+      "step": 591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6412180052956753e-06,
+      "loss": 1.1451,
+      "step": 592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6405560458958516e-06,
+      "loss": 0.7045,
+      "step": 593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6398940864960283e-06,
+      "loss": 1.399,
+      "step": 594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.639232127096205e-06,
+      "loss": 0.6659,
+      "step": 595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6385701676963813e-06,
+      "loss": 0.7988,
+      "step": 596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6379082082965576e-06,
+      "loss": 1.9442,
+      "step": 597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6372462488967343e-06,
+      "loss": 1.0665,
+      "step": 598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.636584289496911e-06,
+      "loss": 0.1161,
+      "step": 599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6359223300970877e-06,
+      "loss": 2.5353,
+      "step": 600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.635260370697264e-06,
+      "loss": 1.254,
+      "step": 601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6345984112974407e-06,
+      "loss": 0.9719,
+      "step": 602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.633936451897617e-06,
+      "loss": 0.4651,
+      "step": 603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6332744924977933e-06,
+      "loss": 1.9163,
+      "step": 604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.63261253309797e-06,
+      "loss": 0.3701,
+      "step": 605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6319505736981467e-06,
+      "loss": 0.7101,
+      "step": 606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6312886142983234e-06,
+      "loss": 1.6476,
+      "step": 607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6306266548984997e-06,
+      "loss": 1.4427,
+      "step": 608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.629964695498676e-06,
+      "loss": 0.8799,
+      "step": 609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6293027360988527e-06,
+      "loss": 0.5758,
+      "step": 610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6286407766990294e-06,
+      "loss": 0.0758,
+      "step": 611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6279788172992057e-06,
+      "loss": 2.1632,
+      "step": 612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6273168578993824e-06,
+      "loss": 1.3978,
+      "step": 613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6266548984995587e-06,
+      "loss": 2.0253,
+      "step": 614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.625992939099735e-06,
+      "loss": 1.2376,
+      "step": 615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6253309796999117e-06,
+      "loss": 0.4546,
+      "step": 616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.6246690203000884e-06,
+      "loss": 1.2011,
+      "step": 617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 2.624007060900265e-06,
+      "loss": 0.591,
+      "step": 618
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6233451015004414e-06,
+      "loss": 0.4679,
+      "step": 619
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6226831421006177e-06,
+      "loss": 2.05,
+      "step": 620
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6220211827007944e-06,
+      "loss": 1.0264,
+      "step": 621
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6213592233009707e-06,
+      "loss": 2.0592,
+      "step": 622
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6206972639011474e-06,
+      "loss": 1.0214,
+      "step": 623
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.620035304501324e-06,
+      "loss": 2.2301,
+      "step": 624
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6193733451015004e-06,
+      "loss": 0.1584,
+      "step": 625
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6187113857016767e-06,
+      "loss": 1.3521,
+      "step": 626
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6180494263018534e-06,
+      "loss": 0.9634,
+      "step": 627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.61738746690203e-06,
+      "loss": 0.0402,
+      "step": 628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.616725507502207e-06,
+      "loss": 0.1623,
+      "step": 629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.616063548102383e-06,
+      "loss": 0.9603,
+      "step": 630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6154015887025594e-06,
+      "loss": 0.8258,
+      "step": 631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.614739629302736e-06,
+      "loss": 1.8274,
+      "step": 632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6140776699029124e-06,
+      "loss": 1.1922,
+      "step": 633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.613415710503089e-06,
+      "loss": 0.3074,
+      "step": 634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.612753751103266e-06,
+      "loss": 0.6449,
+      "step": 635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6120917917034426e-06,
+      "loss": 1.047,
+      "step": 636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.611429832303619e-06,
+      "loss": 1.2687,
+      "step": 637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.610767872903795e-06,
+      "loss": 0.2015,
+      "step": 638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.610105913503972e-06,
+      "loss": 1.1085,
+      "step": 639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6094439541041486e-06,
+      "loss": 0.894,
+      "step": 640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.608781994704325e-06,
+      "loss": 1.9293,
+      "step": 641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6081200353045016e-06,
+      "loss": 1.1553,
+      "step": 642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.607458075904678e-06,
+      "loss": 1.1425,
+      "step": 643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.606796116504854e-06,
+      "loss": 1.4722,
+      "step": 644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.606134157105031e-06,
+      "loss": 0.3439,
+      "step": 645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6054721977052076e-06,
+      "loss": 0.5239,
+      "step": 646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6048102383053843e-06,
+      "loss": 1.0273,
+      "step": 647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6041482789055606e-06,
+      "loss": 2.7936,
+      "step": 648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.603486319505737e-06,
+      "loss": 0.7881,
+      "step": 649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6028243601059136e-06,
+      "loss": 0.9943,
+      "step": 650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6021624007060903e-06,
+      "loss": 0.6663,
+      "step": 651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6015004413062666e-06,
+      "loss": 1.1183,
+      "step": 652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6008384819064433e-06,
+      "loss": 1.3629,
+      "step": 653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.6001765225066196e-06,
+      "loss": 0.5319,
+      "step": 654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.599514563106796e-06,
+      "loss": 0.5271,
+      "step": 655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5988526037069726e-06,
+      "loss": 1.0053,
+      "step": 656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5981906443071493e-06,
+      "loss": 0.4956,
+      "step": 657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.597528684907326e-06,
+      "loss": 0.2197,
+      "step": 658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5968667255075023e-06,
+      "loss": 1.4173,
+      "step": 659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5962047661076786e-06,
+      "loss": 0.9023,
+      "step": 660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5955428067078553e-06,
+      "loss": 1.2508,
+      "step": 661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.594880847308032e-06,
+      "loss": 0.2107,
+      "step": 662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.5942188879082083e-06,
+      "loss": 0.5635,
+      "step": 663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 2.593556928508385e-06,
+      "loss": 0.6623,
+      "step": 664
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5928949691085613e-06,
+      "loss": 0.1273,
+      "step": 665
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5922330097087376e-06,
+      "loss": 0.4109,
+      "step": 666
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5915710503089143e-06,
+      "loss": 1.6751,
+      "step": 667
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.590909090909091e-06,
+      "loss": 0.303,
+      "step": 668
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5902471315092677e-06,
+      "loss": 1.2963,
+      "step": 669
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.589585172109444e-06,
+      "loss": 0.41,
+      "step": 670
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5889232127096203e-06,
+      "loss": 2.1439,
+      "step": 671
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.588261253309797e-06,
+      "loss": 0.894,
+      "step": 672
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5875992939099737e-06,
+      "loss": 2.1684,
+      "step": 673
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.58693733451015e-06,
+      "loss": 1.0262,
+      "step": 674
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5862753751103267e-06,
+      "loss": 1.185,
+      "step": 675
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5856134157105034e-06,
+      "loss": 2.3432,
+      "step": 676
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5849514563106797e-06,
+      "loss": 2.4532,
+      "step": 677
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.584289496910856e-06,
+      "loss": 0.7202,
+      "step": 678
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5836275375110327e-06,
+      "loss": 0.8162,
+      "step": 679
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5829655781112094e-06,
+      "loss": 1.6767,
+      "step": 680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5823036187113857e-06,
+      "loss": 1.3829,
+      "step": 681
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5816416593115624e-06,
+      "loss": 0.6021,
+      "step": 682
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5809796999117387e-06,
+      "loss": 0.8107,
+      "step": 683
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5803177405119154e-06,
+      "loss": 0.7807,
+      "step": 684
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5796557811120917e-06,
+      "loss": 0.879,
+      "step": 685
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5789938217122684e-06,
+      "loss": 0.1041,
+      "step": 686
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.578331862312445e-06,
+      "loss": 1.79,
+      "step": 687
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5776699029126214e-06,
+      "loss": 1.2915,
+      "step": 688
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5770079435127977e-06,
+      "loss": 0.2782,
+      "step": 689
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5763459841129745e-06,
+      "loss": 0.7021,
+      "step": 690
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.575684024713151e-06,
+      "loss": 1.081,
+      "step": 691
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5750220653133275e-06,
+      "loss": 0.8291,
+      "step": 692
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.574360105913504e-06,
+      "loss": 1.0834,
+      "step": 693
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5736981465136805e-06,
+      "loss": 1.3888,
+      "step": 694
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.573036187113857e-06,
+      "loss": 0.1967,
+      "step": 695
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5723742277140335e-06,
+      "loss": 0.2497,
+      "step": 696
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.57171226831421e-06,
+      "loss": 0.1041,
+      "step": 697
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.571050308914387e-06,
+      "loss": 1.0215,
+      "step": 698
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.570388349514563e-06,
+      "loss": 0.8172,
+      "step": 699
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5697263901147395e-06,
+      "loss": 0.4391,
+      "step": 700
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.569064430714916e-06,
+      "loss": 1.199,
+      "step": 701
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.568402471315093e-06,
+      "loss": 1.3511,
+      "step": 702
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.567740511915269e-06,
+      "loss": 3.2437,
+      "step": 703
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.567078552515446e-06,
+      "loss": 0.7999,
+      "step": 704
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.566416593115622e-06,
+      "loss": 2.2192,
+      "step": 705
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.565754633715799e-06,
+      "loss": 0.622,
+      "step": 706
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.565092674315975e-06,
+      "loss": 0.8225,
+      "step": 707
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.564430714916152e-06,
+      "loss": 1.6994,
+      "step": 708
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.5637687555163286e-06,
+      "loss": 2.5639,
+      "step": 709
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2.563106796116505e-06,
+      "loss": 1.8114,
+      "step": 710
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5624448367166816e-06,
+      "loss": 1.8849,
+      "step": 711
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.561782877316858e-06,
+      "loss": 2.0165,
+      "step": 712
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5611209179170346e-06,
+      "loss": 0.909,
+      "step": 713
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.560458958517211e-06,
+      "loss": 0.859,
+      "step": 714
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5597969991173876e-06,
+      "loss": 0.8894,
+      "step": 715
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5591350397175643e-06,
+      "loss": 0.4486,
+      "step": 716
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5584730803177406e-06,
+      "loss": 0.1926,
+      "step": 717
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.557811120917917e-06,
+      "loss": 0.8173,
+      "step": 718
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5571491615180936e-06,
+      "loss": 0.4745,
+      "step": 719
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5564872021182703e-06,
+      "loss": 1.0939,
+      "step": 720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5558252427184466e-06,
+      "loss": 2.9005,
+      "step": 721
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5551632833186233e-06,
+      "loss": 1.395,
+      "step": 722
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5545013239187996e-06,
+      "loss": 0.2539,
+      "step": 723
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5538393645189763e-06,
+      "loss": 1.1554,
+      "step": 724
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5531774051191526e-06,
+      "loss": 1.23,
+      "step": 725
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5525154457193293e-06,
+      "loss": 0.5048,
+      "step": 726
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.551853486319506e-06,
+      "loss": 0.6983,
+      "step": 727
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5511915269196823e-06,
+      "loss": 0.7375,
+      "step": 728
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5505295675198586e-06,
+      "loss": 0.2857,
+      "step": 729
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5498676081200353e-06,
+      "loss": 0.981,
+      "step": 730
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.549205648720212e-06,
+      "loss": 1.0804,
+      "step": 731
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5485436893203883e-06,
+      "loss": 1.1327,
+      "step": 732
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.547881729920565e-06,
+      "loss": 1.6768,
+      "step": 733
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5472197705207413e-06,
+      "loss": 0.0648,
+      "step": 734
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.546557811120918e-06,
+      "loss": 1.2321,
+      "step": 735
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5458958517210943e-06,
+      "loss": 1.647,
+      "step": 736
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.545233892321271e-06,
+      "loss": 0.1523,
+      "step": 737
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5445719329214478e-06,
+      "loss": 1.019,
+      "step": 738
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.543909973521624e-06,
+      "loss": 2.4386,
+      "step": 739
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5432480141218003e-06,
+      "loss": 0.3032,
+      "step": 740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.542586054721977e-06,
+      "loss": 0.1317,
+      "step": 741
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5419240953221538e-06,
+      "loss": 0.9406,
+      "step": 742
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.54126213592233e-06,
+      "loss": 0.6177,
+      "step": 743
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5406001765225068e-06,
+      "loss": 1.0765,
+      "step": 744
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5399382171226835e-06,
+      "loss": 1.2841,
+      "step": 745
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5392762577228598e-06,
+      "loss": 0.6561,
+      "step": 746
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.538614298323036e-06,
+      "loss": 0.7119,
+      "step": 747
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5379523389232128e-06,
+      "loss": 0.3232,
+      "step": 748
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5372903795233895e-06,
+      "loss": 2.8058,
+      "step": 749
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5366284201235658e-06,
+      "loss": 1.686,
+      "step": 750
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5359664607237425e-06,
+      "loss": 1.3288,
+      "step": 751
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5353045013239188e-06,
+      "loss": 0.2097,
+      "step": 752
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5346425419240955e-06,
+      "loss": 0.9064,
+      "step": 753
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5339805825242718e-06,
+      "loss": 1.1192,
+      "step": 754
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.5333186231244485e-06,
+      "loss": 0.1681,
+      "step": 755
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 2.532656663724625e-06,
+      "loss": 2.528,
+      "step": 756
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5319947043248015e-06,
+      "loss": 0.5382,
+      "step": 757
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5313327449249778e-06,
+      "loss": 0.7806,
+      "step": 758
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5306707855251545e-06,
+      "loss": 0.0929,
+      "step": 759
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.530008826125331e-06,
+      "loss": 1.8178,
+      "step": 760
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5293468667255075e-06,
+      "loss": 0.6157,
+      "step": 761
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.528684907325684e-06,
+      "loss": 1.3923,
+      "step": 762
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5280229479258605e-06,
+      "loss": 2.0824,
+      "step": 763
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.527360988526037e-06,
+      "loss": 0.6211,
+      "step": 764
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5266990291262135e-06,
+      "loss": 1.7016,
+      "step": 765
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.52603706972639e-06,
+      "loss": 0.9594,
+      "step": 766
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.525375110326567e-06,
+      "loss": 2.2957,
+      "step": 767
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.524713150926743e-06,
+      "loss": 1.1851,
+      "step": 768
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5240511915269195e-06,
+      "loss": 1.1074,
+      "step": 769
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.523389232127096e-06,
+      "loss": 1.0417,
+      "step": 770
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.522727272727273e-06,
+      "loss": 0.8845,
+      "step": 771
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.522065313327449e-06,
+      "loss": 0.9798,
+      "step": 772
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.521403353927626e-06,
+      "loss": 0.492,
+      "step": 773
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.520741394527802e-06,
+      "loss": 1.3264,
+      "step": 774
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.520079435127979e-06,
+      "loss": 1.103,
+      "step": 775
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.519417475728155e-06,
+      "loss": 2.0999,
+      "step": 776
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.518755516328332e-06,
+      "loss": 2.4523,
+      "step": 777
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5180935569285086e-06,
+      "loss": 1.8496,
+      "step": 778
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5174315975286853e-06,
+      "loss": 3.0668,
+      "step": 779
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.516769638128861e-06,
+      "loss": 0.7713,
+      "step": 780
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.516107678729038e-06,
+      "loss": 0.4579,
+      "step": 781
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5154457193292146e-06,
+      "loss": 0.637,
+      "step": 782
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.514783759929391e-06,
+      "loss": 0.0456,
+      "step": 783
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5141218005295676e-06,
+      "loss": 0.1862,
+      "step": 784
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5134598411297443e-06,
+      "loss": 1.1454,
+      "step": 785
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5127978817299206e-06,
+      "loss": 0.6294,
+      "step": 786
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.512135922330097e-06,
+      "loss": 1.4723,
+      "step": 787
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5114739629302736e-06,
+      "loss": 1.4087,
+      "step": 788
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5108120035304503e-06,
+      "loss": 0.6076,
+      "step": 789
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.510150044130627e-06,
+      "loss": 0.2782,
+      "step": 790
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5094880847308033e-06,
+      "loss": 1.2718,
+      "step": 791
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5088261253309796e-06,
+      "loss": 1.7089,
+      "step": 792
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5081641659311563e-06,
+      "loss": 1.1451,
+      "step": 793
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5075022065313326e-06,
+      "loss": 0.3009,
+      "step": 794
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5068402471315093e-06,
+      "loss": 0.0989,
+      "step": 795
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.506178287731686e-06,
+      "loss": 0.3226,
+      "step": 796
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5055163283318623e-06,
+      "loss": 3.0457,
+      "step": 797
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5048543689320386e-06,
+      "loss": 2.0645,
+      "step": 798
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5041924095322153e-06,
+      "loss": 1.5387,
+      "step": 799
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.503530450132392e-06,
+      "loss": 2.3692,
+      "step": 800
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 2.5028684907325688e-06,
+      "loss": 0.2053,
+      "step": 801
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.502206531332745e-06,
+      "loss": 0.8138,
+      "step": 802
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.5015445719329214e-06,
+      "loss": 1.0654,
+      "step": 803
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.500882612533098e-06,
+      "loss": 0.0989,
+      "step": 804
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.5002206531332744e-06,
+      "loss": 1.48,
+      "step": 805
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.499558693733451e-06,
+      "loss": 1.0412,
+      "step": 806
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4988967343336278e-06,
+      "loss": 1.1162,
+      "step": 807
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.498234774933804e-06,
+      "loss": 0.3169,
+      "step": 808
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4975728155339804e-06,
+      "loss": 0.3925,
+      "step": 809
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.496910856134157e-06,
+      "loss": 0.8684,
+      "step": 810
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4962488967343338e-06,
+      "loss": 1.3073,
+      "step": 811
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4955869373345105e-06,
+      "loss": 1.5974,
+      "step": 812
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4949249779346868e-06,
+      "loss": 1.2167,
+      "step": 813
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.494263018534863e-06,
+      "loss": 1.2787,
+      "step": 814
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4936010591350398e-06,
+      "loss": 0.6249,
+      "step": 815
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.492939099735216e-06,
+      "loss": 0.5861,
+      "step": 816
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4922771403353928e-06,
+      "loss": 0.8151,
+      "step": 817
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4916151809355695e-06,
+      "loss": 1.3265,
+      "step": 818
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.490953221535746e-06,
+      "loss": 1.9672,
+      "step": 819
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4902912621359225e-06,
+      "loss": 1.5002,
+      "step": 820
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4896293027360988e-06,
+      "loss": 1.4345,
+      "step": 821
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4889673433362755e-06,
+      "loss": 1.1378,
+      "step": 822
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4883053839364518e-06,
+      "loss": 3.6811,
+      "step": 823
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4876434245366285e-06,
+      "loss": 0.1133,
+      "step": 824
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.486981465136805e-06,
+      "loss": 1.0283,
+      "step": 825
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4863195057369815e-06,
+      "loss": 0.1097,
+      "step": 826
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.485657546337158e-06,
+      "loss": 1.5502,
+      "step": 827
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4849955869373345e-06,
+      "loss": 1.3106,
+      "step": 828
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4843336275375112e-06,
+      "loss": 0.4003,
+      "step": 829
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.483671668137688e-06,
+      "loss": 1.6758,
+      "step": 830
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4830097087378642e-06,
+      "loss": 0.5829,
+      "step": 831
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4823477493380405e-06,
+      "loss": 2.5275,
+      "step": 832
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4816857899382172e-06,
+      "loss": 1.1031,
+      "step": 833
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4810238305383935e-06,
+      "loss": 1.3413,
+      "step": 834
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4803618711385702e-06,
+      "loss": 0.5866,
+      "step": 835
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.479699911738747e-06,
+      "loss": 1.5072,
+      "step": 836
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4790379523389232e-06,
+      "loss": 1.1889,
+      "step": 837
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4783759929390995e-06,
+      "loss": 0.09,
+      "step": 838
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4777140335392762e-06,
+      "loss": 0.2014,
+      "step": 839
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.477052074139453e-06,
+      "loss": 1.4686,
+      "step": 840
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4763901147396296e-06,
+      "loss": 0.706,
+      "step": 841
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.475728155339806e-06,
+      "loss": 0.6746,
+      "step": 842
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4750661959399822e-06,
+      "loss": 0.9049,
+      "step": 843
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.474404236540159e-06,
+      "loss": 0.245,
+      "step": 844
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4737422771403352e-06,
+      "loss": 0.9767,
+      "step": 845
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.473080317740512e-06,
+      "loss": 0.5281,
+      "step": 846
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 2.4724183583406886e-06,
+      "loss": 0.762,
+      "step": 847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.471756398940865e-06,
+      "loss": 0.4017,
+      "step": 848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4710944395410412e-06,
+      "loss": 0.3682,
+      "step": 849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.470432480141218e-06,
+      "loss": 1.5117,
+      "step": 850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4697705207413947e-06,
+      "loss": 1.0559,
+      "step": 851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4691085613415714e-06,
+      "loss": 1.1651,
+      "step": 852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4684466019417477e-06,
+      "loss": 2.7313,
+      "step": 853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.467784642541924e-06,
+      "loss": 1.5127,
+      "step": 854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4671226831421007e-06,
+      "loss": 0.5441,
+      "step": 855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.466460723742277e-06,
+      "loss": 1.7171,
+      "step": 856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4657987643424537e-06,
+      "loss": 1.5418,
+      "step": 857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4651368049426304e-06,
+      "loss": 1.6214,
+      "step": 858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.464474845542807e-06,
+      "loss": 0.6628,
+      "step": 859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4638128861429834e-06,
+      "loss": 0.9531,
+      "step": 860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4631509267431597e-06,
+      "loss": 0.3604,
+      "step": 861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4624889673433364e-06,
+      "loss": 0.2266,
+      "step": 862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.461827007943513e-06,
+      "loss": 2.4302,
+      "step": 863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4611650485436894e-06,
+      "loss": 0.1211,
+      "step": 864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.460503089143866e-06,
+      "loss": 1.825,
+      "step": 865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4598411297440424e-06,
+      "loss": 0.1049,
+      "step": 866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4591791703442187e-06,
+      "loss": 1.8816,
+      "step": 867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4585172109443954e-06,
+      "loss": 0.747,
+      "step": 868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.457855251544572e-06,
+      "loss": 2.0755,
+      "step": 869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.457193292144749e-06,
+      "loss": 0.0414,
+      "step": 870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.456531332744925e-06,
+      "loss": 0.8314,
+      "step": 871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4558693733451014e-06,
+      "loss": 0.0626,
+      "step": 872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.455207413945278e-06,
+      "loss": 0.1701,
+      "step": 873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.454545454545455e-06,
+      "loss": 0.6999,
+      "step": 874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.453883495145631e-06,
+      "loss": 1.729,
+      "step": 875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.453221535745808e-06,
+      "loss": 0.5297,
+      "step": 876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.452559576345984e-06,
+      "loss": 2.2079,
+      "step": 877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4518976169461604e-06,
+      "loss": 0.9482,
+      "step": 878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.451235657546337e-06,
+      "loss": 1.7824,
+      "step": 879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.450573698146514e-06,
+      "loss": 1.1335,
+      "step": 880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4499117387466905e-06,
+      "loss": 1.1474,
+      "step": 881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.449249779346867e-06,
+      "loss": 0.6342,
+      "step": 882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.448587819947043e-06,
+      "loss": 0.4623,
+      "step": 883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.44792586054722e-06,
+      "loss": 1.6726,
+      "step": 884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4472639011473965e-06,
+      "loss": 0.1754,
+      "step": 885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.446601941747573e-06,
+      "loss": 0.8097,
+      "step": 886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4459399823477495e-06,
+      "loss": 2.5175,
+      "step": 887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.445278022947926e-06,
+      "loss": 1.3636,
+      "step": 888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.444616063548102e-06,
+      "loss": 1.8395,
+      "step": 889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.443954104148279e-06,
+      "loss": 1.6586,
+      "step": 890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4432921447484555e-06,
+      "loss": 0.5902,
+      "step": 891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4426301853486322e-06,
+      "loss": 1.7275,
+      "step": 892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 2.4419682259488085e-06,
+      "loss": 0.0595,
+      "step": 893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4413062665489852e-06,
+      "loss": 0.6944,
+      "step": 894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4406443071491615e-06,
+      "loss": 1.2097,
+      "step": 895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4399823477493382e-06,
+      "loss": 1.7712,
+      "step": 896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4393203883495145e-06,
+      "loss": 0.7844,
+      "step": 897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4386584289496912e-06,
+      "loss": 1.7314,
+      "step": 898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.437996469549868e-06,
+      "loss": 2.009,
+      "step": 899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4373345101500442e-06,
+      "loss": 0.5925,
+      "step": 900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4366725507502205e-06,
+      "loss": 0.0549,
+      "step": 901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4360105913503972e-06,
+      "loss": 0.8093,
+      "step": 902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.435348631950574e-06,
+      "loss": 0.8618,
+      "step": 903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4346866725507502e-06,
+      "loss": 1.3672,
+      "step": 904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.434024713150927e-06,
+      "loss": 0.8082,
+      "step": 905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4333627537511032e-06,
+      "loss": 1.065,
+      "step": 906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.43270079435128e-06,
+      "loss": 0.3672,
+      "step": 907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4320388349514562e-06,
+      "loss": 1.1721,
+      "step": 908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.431376875551633e-06,
+      "loss": 2.1635,
+      "step": 909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4307149161518097e-06,
+      "loss": 0.2381,
+      "step": 910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.430052956751986e-06,
+      "loss": 0.0846,
+      "step": 911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4293909973521622e-06,
+      "loss": 0.2723,
+      "step": 912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.428729037952339e-06,
+      "loss": 1.4849,
+      "step": 913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4280670785525157e-06,
+      "loss": 1.5598,
+      "step": 914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.427405119152692e-06,
+      "loss": 1.2086,
+      "step": 915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4267431597528687e-06,
+      "loss": 1.5953,
+      "step": 916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.426081200353045e-06,
+      "loss": 2.0734,
+      "step": 917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4254192409532217e-06,
+      "loss": 1.2601,
+      "step": 918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.424757281553398e-06,
+      "loss": 1.4235,
+      "step": 919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4240953221535747e-06,
+      "loss": 0.8032,
+      "step": 920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4234333627537514e-06,
+      "loss": 0.9443,
+      "step": 921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4227714033539277e-06,
+      "loss": 1.2397,
+      "step": 922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.422109443954104e-06,
+      "loss": 2.5231,
+      "step": 923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4214474845542807e-06,
+      "loss": 0.7133,
+      "step": 924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4207855251544574e-06,
+      "loss": 1.3869,
+      "step": 925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4201235657546337e-06,
+      "loss": 1.3454,
+      "step": 926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4194616063548104e-06,
+      "loss": 0.0403,
+      "step": 927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.418799646954987e-06,
+      "loss": 2.2712,
+      "step": 928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.418137687555163e-06,
+      "loss": 0.8235,
+      "step": 929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4174757281553397e-06,
+      "loss": 0.6487,
+      "step": 930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4168137687555164e-06,
+      "loss": 1.6011,
+      "step": 931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.416151809355693e-06,
+      "loss": 1.7588,
+      "step": 932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4154898499558694e-06,
+      "loss": 2.1308,
+      "step": 933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.414827890556046e-06,
+      "loss": 0.9926,
+      "step": 934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4141659311562224e-06,
+      "loss": 1.2653,
+      "step": 935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.413503971756399e-06,
+      "loss": 1.0868,
+      "step": 936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.4128420123565754e-06,
+      "loss": 1.0,
+      "step": 937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.412180052956752e-06,
+      "loss": 1.9976,
+      "step": 938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 2.411518093556929e-06,
+      "loss": 1.4658,
+      "step": 939
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.410856134157105e-06,
+      "loss": 0.3957,
+      "step": 940
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4101941747572814e-06,
+      "loss": 1.307,
+      "step": 941
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.409532215357458e-06,
+      "loss": 1.4888,
+      "step": 942
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.408870255957635e-06,
+      "loss": 0.8204,
+      "step": 943
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.408208296557811e-06,
+      "loss": 1.1709,
+      "step": 944
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.407546337157988e-06,
+      "loss": 1.4177,
+      "step": 945
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.406884377758164e-06,
+      "loss": 0.6246,
+      "step": 946
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.406222418358341e-06,
+      "loss": 0.6263,
+      "step": 947
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.405560458958517e-06,
+      "loss": 0.9751,
+      "step": 948
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.404898499558694e-06,
+      "loss": 0.4972,
+      "step": 949
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4042365401588705e-06,
+      "loss": 1.6012,
+      "step": 950
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.403574580759047e-06,
+      "loss": 2.295,
+      "step": 951
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.402912621359223e-06,
+      "loss": 2.1827,
+      "step": 952
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4022506619594e-06,
+      "loss": 0.3798,
+      "step": 953
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4015887025595765e-06,
+      "loss": 1.5645,
+      "step": 954
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.400926743159753e-06,
+      "loss": 1.7,
+      "step": 955
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.4002647837599295e-06,
+      "loss": 1.382,
+      "step": 956
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.399602824360106e-06,
+      "loss": 0.5618,
+      "step": 957
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3989408649602825e-06,
+      "loss": 0.0595,
+      "step": 958
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.398278905560459e-06,
+      "loss": 0.8514,
+      "step": 959
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3976169461606355e-06,
+      "loss": 2.3494,
+      "step": 960
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3969549867608123e-06,
+      "loss": 1.655,
+      "step": 961
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3962930273609885e-06,
+      "loss": 1.8356,
+      "step": 962
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.395631067961165e-06,
+      "loss": 0.8003,
+      "step": 963
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3949691085613416e-06,
+      "loss": 1.0166,
+      "step": 964
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3943071491615183e-06,
+      "loss": 0.5021,
+      "step": 965
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3936451897616946e-06,
+      "loss": 0.1234,
+      "step": 966
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3929832303618713e-06,
+      "loss": 1.9733,
+      "step": 967
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.392321270962048e-06,
+      "loss": 1.4809,
+      "step": 968
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3916593115622243e-06,
+      "loss": 1.1871,
+      "step": 969
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3909973521624006e-06,
+      "loss": 1.3816,
+      "step": 970
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3903353927625773e-06,
+      "loss": 1.547,
+      "step": 971
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.389673433362754e-06,
+      "loss": 0.856,
+      "step": 972
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3890114739629303e-06,
+      "loss": 1.5963,
+      "step": 973
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.388349514563107e-06,
+      "loss": 0.6391,
+      "step": 974
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3876875551632833e-06,
+      "loss": 2.9088,
+      "step": 975
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.38702559576346e-06,
+      "loss": 3.068,
+      "step": 976
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3863636363636363e-06,
+      "loss": 0.3605,
+      "step": 977
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.385701676963813e-06,
+      "loss": 0.2021,
+      "step": 978
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3850397175639897e-06,
+      "loss": 2.6538,
+      "step": 979
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.384377758164166e-06,
+      "loss": 1.2805,
+      "step": 980
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3837157987643423e-06,
+      "loss": 1.345,
+      "step": 981
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.383053839364519e-06,
+      "loss": 1.0493,
+      "step": 982
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3823918799646957e-06,
+      "loss": 1.142,
+      "step": 983
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.381729920564872e-06,
+      "loss": 0.1235,
+      "step": 984
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 2.3810679611650487e-06,
+      "loss": 1.5243,
+      "step": 985
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.380406001765225e-06,
+      "loss": 1.0012,
+      "step": 986
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3797440423654017e-06,
+      "loss": 0.3797,
+      "step": 987
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.379082082965578e-06,
+      "loss": 2.4248,
+      "step": 988
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3784201235657547e-06,
+      "loss": 1.9234,
+      "step": 989
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3777581641659314e-06,
+      "loss": 0.6551,
+      "step": 990
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3770962047661077e-06,
+      "loss": 0.3501,
+      "step": 991
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.376434245366284e-06,
+      "loss": 0.1297,
+      "step": 992
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3757722859664607e-06,
+      "loss": 1.7199,
+      "step": 993
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3751103265666374e-06,
+      "loss": 2.8429,
+      "step": 994
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3744483671668137e-06,
+      "loss": 0.9577,
+      "step": 995
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3737864077669904e-06,
+      "loss": 1.9818,
+      "step": 996
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3731244483671667e-06,
+      "loss": 1.8716,
+      "step": 997
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3724624889673434e-06,
+      "loss": 1.0864,
+      "step": 998
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3718005295675197e-06,
+      "loss": 2.076,
+      "step": 999
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3711385701676964e-06,
+      "loss": 0.1759,
+      "step": 1000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.370476610767873e-06,
+      "loss": 1.0113,
+      "step": 1001
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.36981465136805e-06,
+      "loss": 1.0163,
+      "step": 1002
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.369152691968226e-06,
+      "loss": 1.4597,
+      "step": 1003
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3684907325684024e-06,
+      "loss": 1.3338,
+      "step": 1004
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.367828773168579e-06,
+      "loss": 0.8591,
+      "step": 1005
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3671668137687554e-06,
+      "loss": 2.2625,
+      "step": 1006
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.366504854368932e-06,
+      "loss": 0.5425,
+      "step": 1007
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.365842894969109e-06,
+      "loss": 1.0667,
+      "step": 1008
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.365180935569285e-06,
+      "loss": 0.9535,
+      "step": 1009
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3645189761694614e-06,
+      "loss": 1.1017,
+      "step": 1010
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.363857016769638e-06,
+      "loss": 2.8528,
+      "step": 1011
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.363195057369815e-06,
+      "loss": 0.8467,
+      "step": 1012
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3625330979699916e-06,
+      "loss": 1.1308,
+      "step": 1013
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.361871138570168e-06,
+      "loss": 0.6081,
+      "step": 1014
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.361209179170344e-06,
+      "loss": 2.0134,
+      "step": 1015
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.360547219770521e-06,
+      "loss": 2.1985,
+      "step": 1016
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.359885260370697e-06,
+      "loss": 0.7821,
+      "step": 1017
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.359223300970874e-06,
+      "loss": 0.5173,
+      "step": 1018
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3585613415710506e-06,
+      "loss": 0.5112,
+      "step": 1019
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.357899382171227e-06,
+      "loss": 0.459,
+      "step": 1020
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.357237422771403e-06,
+      "loss": 3.0591,
+      "step": 1021
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.35657546337158e-06,
+      "loss": 1.7059,
+      "step": 1022
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3559135039717566e-06,
+      "loss": 0.58,
+      "step": 1023
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3552515445719333e-06,
+      "loss": 0.8042,
+      "step": 1024
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3545895851721096e-06,
+      "loss": 0.5115,
+      "step": 1025
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.353927625772286e-06,
+      "loss": 0.896,
+      "step": 1026
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3532656663724626e-06,
+      "loss": 1.935,
+      "step": 1027
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.352603706972639e-06,
+      "loss": 0.2494,
+      "step": 1028
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3519417475728156e-06,
+      "loss": 1.8401,
+      "step": 1029
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 2.3512797881729923e-06,
+      "loss": 0.7291,
+      "step": 1030
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3506178287731686e-06,
+      "loss": 1.7235,
+      "step": 1031
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.349955869373345e-06,
+      "loss": 0.6721,
+      "step": 1032
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3492939099735216e-06,
+      "loss": 1.9305,
+      "step": 1033
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3486319505736983e-06,
+      "loss": 1.1273,
+      "step": 1034
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3479699911738746e-06,
+      "loss": 3.5497,
+      "step": 1035
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3473080317740513e-06,
+      "loss": 0.4523,
+      "step": 1036
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.346646072374228e-06,
+      "loss": 1.5539,
+      "step": 1037
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3459841129744043e-06,
+      "loss": 0.7792,
+      "step": 1038
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3453221535745806e-06,
+      "loss": 0.3358,
+      "step": 1039
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3446601941747573e-06,
+      "loss": 1.4174,
+      "step": 1040
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.343998234774934e-06,
+      "loss": 1.3287,
+      "step": 1041
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3433362753751107e-06,
+      "loss": 0.144,
+      "step": 1042
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.342674315975287e-06,
+      "loss": 0.4235,
+      "step": 1043
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3420123565754633e-06,
+      "loss": 0.8184,
+      "step": 1044
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.34135039717564e-06,
+      "loss": 0.4436,
+      "step": 1045
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3406884377758163e-06,
+      "loss": 0.7096,
+      "step": 1046
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.340026478375993e-06,
+      "loss": 1.7555,
+      "step": 1047
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3393645189761697e-06,
+      "loss": 1.4419,
+      "step": 1048
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.338702559576346e-06,
+      "loss": 2.2095,
+      "step": 1049
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3380406001765223e-06,
+      "loss": 0.8105,
+      "step": 1050
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.337378640776699e-06,
+      "loss": 0.7081,
+      "step": 1051
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3367166813768757e-06,
+      "loss": 0.5014,
+      "step": 1052
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3360547219770524e-06,
+      "loss": 0.0589,
+      "step": 1053
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3353927625772287e-06,
+      "loss": 0.5765,
+      "step": 1054
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.334730803177405e-06,
+      "loss": 1.5679,
+      "step": 1055
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3340688437775817e-06,
+      "loss": 1.0278,
+      "step": 1056
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.333406884377758e-06,
+      "loss": 2.1162,
+      "step": 1057
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3327449249779347e-06,
+      "loss": 1.8578,
+      "step": 1058
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3320829655781114e-06,
+      "loss": 1.0907,
+      "step": 1059
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3314210061782877e-06,
+      "loss": 1.5145,
+      "step": 1060
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.330759046778464e-06,
+      "loss": 1.6952,
+      "step": 1061
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3300970873786407e-06,
+      "loss": 0.7172,
+      "step": 1062
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3294351279788174e-06,
+      "loss": 1.9138,
+      "step": 1063
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.328773168578994e-06,
+      "loss": 1.8443,
+      "step": 1064
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3281112091791704e-06,
+      "loss": 0.6902,
+      "step": 1065
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3274492497793467e-06,
+      "loss": 0.0721,
+      "step": 1066
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3267872903795234e-06,
+      "loss": 2.8791,
+      "step": 1067
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3261253309796997e-06,
+      "loss": 0.8078,
+      "step": 1068
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3254633715798764e-06,
+      "loss": 0.77,
+      "step": 1069
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.324801412180053e-06,
+      "loss": 1.4864,
+      "step": 1070
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3241394527802294e-06,
+      "loss": 1.384,
+      "step": 1071
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3234774933804057e-06,
+      "loss": 0.9767,
+      "step": 1072
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.3228155339805824e-06,
+      "loss": 2.2387,
+      "step": 1073
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.322153574580759e-06,
+      "loss": 1.284,
+      "step": 1074
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.321491615180936e-06,
+      "loss": 1.4632,
+      "step": 1075
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 2.320829655781112e-06,
+      "loss": 0.5757,
+      "step": 1076
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.320167696381289e-06,
+      "loss": 0.4469,
+      "step": 1077
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.319505736981465e-06,
+      "loss": 1.481,
+      "step": 1078
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3188437775816415e-06,
+      "loss": 4.1246,
+      "step": 1079
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.318181818181818e-06,
+      "loss": 1.7984,
+      "step": 1080
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.317519858781995e-06,
+      "loss": 0.4903,
+      "step": 1081
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3168578993821716e-06,
+      "loss": 0.3749,
+      "step": 1082
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.316195939982348e-06,
+      "loss": 0.2083,
+      "step": 1083
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.315533980582524e-06,
+      "loss": 0.6691,
+      "step": 1084
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.314872021182701e-06,
+      "loss": 0.9,
+      "step": 1085
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3142100617828776e-06,
+      "loss": 1.7306,
+      "step": 1086
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.313548102383054e-06,
+      "loss": 1.2296,
+      "step": 1087
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3128861429832306e-06,
+      "loss": 0.4419,
+      "step": 1088
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.312224183583407e-06,
+      "loss": 1.4095,
+      "step": 1089
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.311562224183583e-06,
+      "loss": 1.2568,
+      "step": 1090
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.31090026478376e-06,
+      "loss": 1.3092,
+      "step": 1091
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3102383053839366e-06,
+      "loss": 1.0082,
+      "step": 1092
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3095763459841133e-06,
+      "loss": 2.9665,
+      "step": 1093
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3089143865842896e-06,
+      "loss": 1.0152,
+      "step": 1094
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.308252427184466e-06,
+      "loss": 0.7148,
+      "step": 1095
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3075904677846426e-06,
+      "loss": 1.207,
+      "step": 1096
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3069285083848193e-06,
+      "loss": 0.9576,
+      "step": 1097
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3062665489849956e-06,
+      "loss": 1.554,
+      "step": 1098
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3056045895851723e-06,
+      "loss": 1.0194,
+      "step": 1099
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3049426301853486e-06,
+      "loss": 2.0795,
+      "step": 1100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.304280670785525e-06,
+      "loss": 0.8772,
+      "step": 1101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3036187113857016e-06,
+      "loss": 1.1846,
+      "step": 1102
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3029567519858783e-06,
+      "loss": 0.5912,
+      "step": 1103
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.302294792586055e-06,
+      "loss": 0.6594,
+      "step": 1104
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3016328331862313e-06,
+      "loss": 0.9622,
+      "step": 1105
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3009708737864076e-06,
+      "loss": 1.6626,
+      "step": 1106
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.3003089143865843e-06,
+      "loss": 2.0447,
+      "step": 1107
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.299646954986761e-06,
+      "loss": 0.3928,
+      "step": 1108
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2989849955869373e-06,
+      "loss": 0.8738,
+      "step": 1109
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.298323036187114e-06,
+      "loss": 0.5051,
+      "step": 1110
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2976610767872907e-06,
+      "loss": 0.9329,
+      "step": 1111
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2969991173874666e-06,
+      "loss": 0.3765,
+      "step": 1112
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2963371579876433e-06,
+      "loss": 0.5556,
+      "step": 1113
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.29567519858782e-06,
+      "loss": 1.6421,
+      "step": 1114
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2950132391879967e-06,
+      "loss": 1.9824,
+      "step": 1115
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.294351279788173e-06,
+      "loss": 0.4522,
+      "step": 1116
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2936893203883497e-06,
+      "loss": 0.9268,
+      "step": 1117
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.293027360988526e-06,
+      "loss": 0.8001,
+      "step": 1118
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2923654015887027e-06,
+      "loss": 0.8559,
+      "step": 1119
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.291703442188879e-06,
+      "loss": 3.9215,
+      "step": 1120
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2910414827890557e-06,
+      "loss": 0.4136,
+      "step": 1121
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 2.2903795233892325e-06,
+      "loss": 1.1221,
+      "step": 1122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2897175639894087e-06,
+      "loss": 0.3301,
+      "step": 1123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.289055604589585e-06,
+      "loss": 0.5679,
+      "step": 1124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2883936451897618e-06,
+      "loss": 1.5671,
+      "step": 1125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2877316857899385e-06,
+      "loss": 1.0712,
+      "step": 1126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2870697263901148e-06,
+      "loss": 2.348,
+      "step": 1127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2864077669902915e-06,
+      "loss": 3.7125,
+      "step": 1128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2857458075904678e-06,
+      "loss": 2.0828,
+      "step": 1129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2850838481906445e-06,
+      "loss": 0.6149,
+      "step": 1130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2844218887908208e-06,
+      "loss": 0.1656,
+      "step": 1131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2837599293909975e-06,
+      "loss": 1.8125,
+      "step": 1132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.283097969991174e-06,
+      "loss": 1.6195,
+      "step": 1133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2824360105913505e-06,
+      "loss": 0.7252,
+      "step": 1134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2817740511915268e-06,
+      "loss": 1.1366,
+      "step": 1135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2811120917917035e-06,
+      "loss": 1.9438,
+      "step": 1136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.28045013239188e-06,
+      "loss": 0.7845,
+      "step": 1137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2797881729920565e-06,
+      "loss": 2.5931,
+      "step": 1138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.279126213592233e-06,
+      "loss": 1.0065,
+      "step": 1139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2784642541924095e-06,
+      "loss": 2.0633,
+      "step": 1140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2778022947925858e-06,
+      "loss": 1.4708,
+      "step": 1141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2771403353927625e-06,
+      "loss": 0.808,
+      "step": 1142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.276478375992939e-06,
+      "loss": 1.458,
+      "step": 1143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.275816416593116e-06,
+      "loss": 0.245,
+      "step": 1144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.275154457193292e-06,
+      "loss": 0.3924,
+      "step": 1145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2744924977934685e-06,
+      "loss": 2.0727,
+      "step": 1146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.273830538393645e-06,
+      "loss": 2.1561,
+      "step": 1147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.273168578993822e-06,
+      "loss": 1.4833,
+      "step": 1148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.272506619593998e-06,
+      "loss": 1.2041,
+      "step": 1149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.271844660194175e-06,
+      "loss": 1.4027,
+      "step": 1150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2711827007943516e-06,
+      "loss": 0.8555,
+      "step": 1151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.270520741394528e-06,
+      "loss": 0.6855,
+      "step": 1152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.269858781994704e-06,
+      "loss": 1.0,
+      "step": 1153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.269196822594881e-06,
+      "loss": 0.4604,
+      "step": 1154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2685348631950576e-06,
+      "loss": 2.0603,
+      "step": 1155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.267872903795234e-06,
+      "loss": 2.1879,
+      "step": 1156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2672109443954106e-06,
+      "loss": 1.4733,
+      "step": 1157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.266548984995587e-06,
+      "loss": 0.9307,
+      "step": 1158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2658870255957636e-06,
+      "loss": 0.3071,
+      "step": 1159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.26522506619594e-06,
+      "loss": 1.3583,
+      "step": 1160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2645631067961166e-06,
+      "loss": 0.9559,
+      "step": 1161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2639011473962933e-06,
+      "loss": 0.4301,
+      "step": 1162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2632391879964696e-06,
+      "loss": 3.5645,
+      "step": 1163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.262577228596646e-06,
+      "loss": 2.7002,
+      "step": 1164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2619152691968226e-06,
+      "loss": 0.889,
+      "step": 1165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2612533097969993e-06,
+      "loss": 1.9639,
+      "step": 1166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2605913503971756e-06,
+      "loss": 1.2524,
+      "step": 1167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 2.2599293909973523e-06,
+      "loss": 1.2669,
+      "step": 1168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2592674315975286e-06,
+      "loss": 1.2689,
+      "step": 1169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2586054721977053e-06,
+      "loss": 1.2639,
+      "step": 1170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2579435127978816e-06,
+      "loss": 0.8234,
+      "step": 1171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2572815533980583e-06,
+      "loss": 0.5877,
+      "step": 1172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.256619593998235e-06,
+      "loss": 1.0581,
+      "step": 1173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2559576345984113e-06,
+      "loss": 1.0647,
+      "step": 1174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2552956751985876e-06,
+      "loss": 0.9852,
+      "step": 1175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2546337157987643e-06,
+      "loss": 0.2842,
+      "step": 1176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.253971756398941e-06,
+      "loss": 0.9998,
+      "step": 1177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2533097969991173e-06,
+      "loss": 0.4159,
+      "step": 1178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.252647837599294e-06,
+      "loss": 2.607,
+      "step": 1179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2519858781994703e-06,
+      "loss": 0.891,
+      "step": 1180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.251323918799647e-06,
+      "loss": 1.8175,
+      "step": 1181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2506619593998233e-06,
+      "loss": 0.7553,
+      "step": 1182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.25e-06,
+      "loss": 0.1453,
+      "step": 1183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2493380406001768e-06,
+      "loss": 0.9512,
+      "step": 1184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.248676081200353e-06,
+      "loss": 0.0455,
+      "step": 1185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2480141218005298e-06,
+      "loss": 0.9978,
+      "step": 1186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.247352162400706e-06,
+      "loss": 1.2774,
+      "step": 1187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2466902030008828e-06,
+      "loss": 1.9616,
+      "step": 1188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.246028243601059e-06,
+      "loss": 2.0288,
+      "step": 1189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2453662842012358e-06,
+      "loss": 0.0814,
+      "step": 1190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2447043248014125e-06,
+      "loss": 0.6682,
+      "step": 1191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2440423654015888e-06,
+      "loss": 1.6143,
+      "step": 1192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.243380406001765e-06,
+      "loss": 2.1181,
+      "step": 1193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2427184466019418e-06,
+      "loss": 0.9933,
+      "step": 1194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2420564872021185e-06,
+      "loss": 2.3481,
+      "step": 1195
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2413945278022948e-06,
+      "loss": 0.4879,
+      "step": 1196
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2407325684024715e-06,
+      "loss": 1.1047,
+      "step": 1197
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2400706090026478e-06,
+      "loss": 0.6975,
+      "step": 1198
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2394086496028245e-06,
+      "loss": 2.0057,
+      "step": 1199
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2387466902030008e-06,
+      "loss": 0.847,
+      "step": 1200
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2380847308031775e-06,
+      "loss": 1.8615,
+      "step": 1201
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.237422771403354e-06,
+      "loss": 2.5039,
+      "step": 1202
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2367608120035305e-06,
+      "loss": 2.5126,
+      "step": 1203
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2360988526037068e-06,
+      "loss": 0.8119,
+      "step": 1204
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2354368932038835e-06,
+      "loss": 1.8294,
+      "step": 1205
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.23477493380406e-06,
+      "loss": 0.6441,
+      "step": 1206
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2341129744042365e-06,
+      "loss": 1.3458,
+      "step": 1207
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.233451015004413e-06,
+      "loss": 0.8927,
+      "step": 1208
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2327890556045895e-06,
+      "loss": 0.8583,
+      "step": 1209
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.232127096204766e-06,
+      "loss": 1.9946,
+      "step": 1210
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.2314651368049425e-06,
+      "loss": 0.91,
+      "step": 1211
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.230803177405119e-06,
+      "loss": 0.9823,
+      "step": 1212
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.230141218005296e-06,
+      "loss": 1.3625,
+      "step": 1213
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 2.229479258605472e-06,
+      "loss": 1.2054,
+      "step": 1214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2288172992056485e-06,
+      "loss": 0.4629,
+      "step": 1215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2281553398058252e-06,
+      "loss": 1.737,
+      "step": 1216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.227493380406002e-06,
+      "loss": 0.3054,
+      "step": 1217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2268314210061782e-06,
+      "loss": 1.7333,
+      "step": 1218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.226169461606355e-06,
+      "loss": 2.5336,
+      "step": 1219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2255075022065316e-06,
+      "loss": 0.5173,
+      "step": 1220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.224845542806708e-06,
+      "loss": 0.7216,
+      "step": 1221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2241835834068842e-06,
+      "loss": 0.0751,
+      "step": 1222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.223521624007061e-06,
+      "loss": 0.4067,
+      "step": 1223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2228596646072376e-06,
+      "loss": 0.6606,
+      "step": 1224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2221977052074144e-06,
+      "loss": 1.9476,
+      "step": 1225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2215357458075906e-06,
+      "loss": 1.0256,
+      "step": 1226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.220873786407767e-06,
+      "loss": 2.0719,
+      "step": 1227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2202118270079436e-06,
+      "loss": 1.2964,
+      "step": 1228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.21954986760812e-06,
+      "loss": 3.2589,
+      "step": 1229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2188879082082966e-06,
+      "loss": 1.2375,
+      "step": 1230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2182259488084734e-06,
+      "loss": 0.971,
+      "step": 1231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2175639894086496e-06,
+      "loss": 1.1762,
+      "step": 1232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.216902030008826e-06,
+      "loss": 0.6393,
+      "step": 1233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2162400706090026e-06,
+      "loss": 1.0843,
+      "step": 1234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2155781112091794e-06,
+      "loss": 3.3803,
+      "step": 1235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2149161518093556e-06,
+      "loss": 3.023,
+      "step": 1236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2142541924095324e-06,
+      "loss": 0.0556,
+      "step": 1237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2135922330097087e-06,
+      "loss": 2.2103,
+      "step": 1238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2129302736098854e-06,
+      "loss": 0.7137,
+      "step": 1239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2122683142100617e-06,
+      "loss": 1.6361,
+      "step": 1240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2116063548102384e-06,
+      "loss": 1.8428,
+      "step": 1241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.210944395410415e-06,
+      "loss": 1.268,
+      "step": 1242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2102824360105914e-06,
+      "loss": 0.4098,
+      "step": 1243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2096204766107677e-06,
+      "loss": 0.707,
+      "step": 1244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2089585172109444e-06,
+      "loss": 1.5649,
+      "step": 1245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.208296557811121e-06,
+      "loss": 0.0781,
+      "step": 1246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2076345984112974e-06,
+      "loss": 2.0631,
+      "step": 1247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.206972639011474e-06,
+      "loss": 2.4804,
+      "step": 1248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2063106796116504e-06,
+      "loss": 1.0056,
+      "step": 1249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.205648720211827e-06,
+      "loss": 2.0198,
+      "step": 1250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2049867608120034e-06,
+      "loss": 2.0591,
+      "step": 1251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.20432480141218e-06,
+      "loss": 1.1239,
+      "step": 1252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.203662842012357e-06,
+      "loss": 0.8921,
+      "step": 1253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2030008826125335e-06,
+      "loss": 0.7237,
+      "step": 1254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.2023389232127094e-06,
+      "loss": 3.2813,
+      "step": 1255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.201676963812886e-06,
+      "loss": 1.9333,
+      "step": 1256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.201015004413063e-06,
+      "loss": 2.041,
+      "step": 1257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.200353045013239e-06,
+      "loss": 0.7944,
+      "step": 1258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.199691085613416e-06,
+      "loss": 2.2895,
+      "step": 1259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 2.1990291262135925e-06,
+      "loss": 0.8122,
+      "step": 1260
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.198367166813769e-06,
+      "loss": 0.4615,
+      "step": 1261
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.197705207413945e-06,
+      "loss": 2.2139,
+      "step": 1262
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.197043248014122e-06,
+      "loss": 1.7832,
+      "step": 1263
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1963812886142985e-06,
+      "loss": 0.6683,
+      "step": 1264
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1957193292144752e-06,
+      "loss": 2.9592,
+      "step": 1265
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1950573698146515e-06,
+      "loss": 1.8292,
+      "step": 1266
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.194395410414828e-06,
+      "loss": 1.6336,
+      "step": 1267
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1937334510150045e-06,
+      "loss": 2.041,
+      "step": 1268
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.193071491615181e-06,
+      "loss": 2.1022,
+      "step": 1269
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1924095322153575e-06,
+      "loss": 2.177,
+      "step": 1270
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1917475728155342e-06,
+      "loss": 1.1629,
+      "step": 1271
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1910856134157105e-06,
+      "loss": 1.2768,
+      "step": 1272
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.190423654015887e-06,
+      "loss": 1.3417,
+      "step": 1273
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1897616946160635e-06,
+      "loss": 0.3042,
+      "step": 1274
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1890997352162402e-06,
+      "loss": 0.1366,
+      "step": 1275
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.188437775816417e-06,
+      "loss": 1.6015,
+      "step": 1276
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1877758164165932e-06,
+      "loss": 0.9972,
+      "step": 1277
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1871138570167695e-06,
+      "loss": 1.7153,
+      "step": 1278
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1864518976169462e-06,
+      "loss": 0.0821,
+      "step": 1279
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1857899382171225e-06,
+      "loss": 0.615,
+      "step": 1280
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1851279788172992e-06,
+      "loss": 0.0649,
+      "step": 1281
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.184466019417476e-06,
+      "loss": 0.961,
+      "step": 1282
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1838040600176522e-06,
+      "loss": 1.5126,
+      "step": 1283
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1831421006178285e-06,
+      "loss": 2.9477,
+      "step": 1284
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1824801412180052e-06,
+      "loss": 0.1162,
+      "step": 1285
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.181818181818182e-06,
+      "loss": 0.0901,
+      "step": 1286
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1811562224183587e-06,
+      "loss": 0.7359,
+      "step": 1287
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.180494263018535e-06,
+      "loss": 1.8192,
+      "step": 1288
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1798323036187112e-06,
+      "loss": 1.4098,
+      "step": 1289
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.179170344218888e-06,
+      "loss": 0.5978,
+      "step": 1290
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1785083848190642e-06,
+      "loss": 0.9818,
+      "step": 1291
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.177846425419241e-06,
+      "loss": 1.8569,
+      "step": 1292
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1771844660194177e-06,
+      "loss": 0.4182,
+      "step": 1293
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1765225066195944e-06,
+      "loss": 0.0835,
+      "step": 1294
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1758605472197707e-06,
+      "loss": 0.6457,
+      "step": 1295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.175198587819947e-06,
+      "loss": 1.8923,
+      "step": 1296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1745366284201237e-06,
+      "loss": 2.0705,
+      "step": 1297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1738746690203004e-06,
+      "loss": 1.5797,
+      "step": 1298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1732127096204767e-06,
+      "loss": 0.9186,
+      "step": 1299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1725507502206534e-06,
+      "loss": 1.0648,
+      "step": 1300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1718887908208297e-06,
+      "loss": 0.6718,
+      "step": 1301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.171226831421006e-06,
+      "loss": 1.4666,
+      "step": 1302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1705648720211827e-06,
+      "loss": 0.9891,
+      "step": 1303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.1699029126213594e-06,
+      "loss": 1.5614,
+      "step": 1304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 2.169240953221536e-06,
+      "loss": 1.9225,
+      "step": 1305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1685789938217124e-06,
+      "loss": 0.9292,
+      "step": 1306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1679170344218887e-06,
+      "loss": 0.8083,
+      "step": 1307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1672550750220654e-06,
+      "loss": 0.1601,
+      "step": 1308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.166593115622242e-06,
+      "loss": 0.1111,
+      "step": 1309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1659311562224184e-06,
+      "loss": 0.7362,
+      "step": 1310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.165269196822595e-06,
+      "loss": 0.6211,
+      "step": 1311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1646072374227714e-06,
+      "loss": 0.8957,
+      "step": 1312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1639452780229477e-06,
+      "loss": 2.4377,
+      "step": 1313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1632833186231244e-06,
+      "loss": 0.3723,
+      "step": 1314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.162621359223301e-06,
+      "loss": 1.4013,
+      "step": 1315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.161959399823478e-06,
+      "loss": 0.4284,
+      "step": 1316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.161297440423654e-06,
+      "loss": 0.7937,
+      "step": 1317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1606354810238304e-06,
+      "loss": 1.2527,
+      "step": 1318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.159973521624007e-06,
+      "loss": 1.7999,
+      "step": 1319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.159311562224184e-06,
+      "loss": 0.8205,
+      "step": 1320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.15864960282436e-06,
+      "loss": 0.5712,
+      "step": 1321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.157987643424537e-06,
+      "loss": 0.4506,
+      "step": 1322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.157325684024713e-06,
+      "loss": 1.1465,
+      "step": 1323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1566637246248894e-06,
+      "loss": 0.9498,
+      "step": 1324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.156001765225066e-06,
+      "loss": 1.1567,
+      "step": 1325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.155339805825243e-06,
+      "loss": 0.8963,
+      "step": 1326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1546778464254195e-06,
+      "loss": 1.3172,
+      "step": 1327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.154015887025596e-06,
+      "loss": 3.3222,
+      "step": 1328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.153353927625772e-06,
+      "loss": 2.0724,
+      "step": 1329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.152691968225949e-06,
+      "loss": 0.7978,
+      "step": 1330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1520300088261255e-06,
+      "loss": 0.6675,
+      "step": 1331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.151368049426302e-06,
+      "loss": 1.5003,
+      "step": 1332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1507060900264785e-06,
+      "loss": 1.7664,
+      "step": 1333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1500441306266553e-06,
+      "loss": 1.2657,
+      "step": 1334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1493821712268315e-06,
+      "loss": 0.9805,
+      "step": 1335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.148720211827008e-06,
+      "loss": 1.0468,
+      "step": 1336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1480582524271845e-06,
+      "loss": 1.2757,
+      "step": 1337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1473962930273613e-06,
+      "loss": 1.1758,
+      "step": 1338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1467343336275375e-06,
+      "loss": 0.989,
+      "step": 1339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1460723742277143e-06,
+      "loss": 1.1892,
+      "step": 1340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1454104148278905e-06,
+      "loss": 1.7633,
+      "step": 1341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.144748455428067e-06,
+      "loss": 1.3622,
+      "step": 1342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1440864960282435e-06,
+      "loss": 0.8412,
+      "step": 1343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1434245366284203e-06,
+      "loss": 1.1003,
+      "step": 1344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.142762577228597e-06,
+      "loss": 2.0143,
+      "step": 1345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1421006178287733e-06,
+      "loss": 2.6456,
+      "step": 1346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1414386584289495e-06,
+      "loss": 0.2204,
+      "step": 1347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1407766990291263e-06,
+      "loss": 0.1102,
+      "step": 1348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.140114739629303e-06,
+      "loss": 1.6706,
+      "step": 1349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.1394527802294793e-06,
+      "loss": 0.5032,
+      "step": 1350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 2.138790820829656e-06,
+      "loss": 0.7067,
+      "step": 1351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1381288614298323e-06,
+      "loss": 0.9436,
+      "step": 1352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1374669020300086e-06,
+      "loss": 0.4213,
+      "step": 1353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1368049426301853e-06,
+      "loss": 2.4098,
+      "step": 1354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.136142983230362e-06,
+      "loss": 2.707,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1354810238305387e-06,
+      "loss": 2.7341,
+      "step": 1356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.134819064430715e-06,
+      "loss": 1.7864,
+      "step": 1357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1341571050308913e-06,
+      "loss": 0.6478,
+      "step": 1358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.133495145631068e-06,
+      "loss": 0.348,
+      "step": 1359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1328331862312447e-06,
+      "loss": 0.7301,
+      "step": 1360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.132171226831421e-06,
+      "loss": 1.0723,
+      "step": 1361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1315092674315977e-06,
+      "loss": 1.0829,
+      "step": 1362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.130847308031774e-06,
+      "loss": 0.5748,
+      "step": 1363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1301853486319503e-06,
+      "loss": 1.757,
+      "step": 1364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.129523389232127e-06,
+      "loss": 0.5573,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1288614298323037e-06,
+      "loss": 1.8474,
+      "step": 1366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1281994704324804e-06,
+      "loss": 1.1862,
+      "step": 1367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1275375110326567e-06,
+      "loss": 3.2239,
+      "step": 1368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1268755516328334e-06,
+      "loss": 0.1604,
+      "step": 1369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1262135922330097e-06,
+      "loss": 0.9915,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1255516328331864e-06,
+      "loss": 1.4562,
+      "step": 1371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1248896734333627e-06,
+      "loss": 3.4323,
+      "step": 1372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1242277140335394e-06,
+      "loss": 2.6676,
+      "step": 1373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.123565754633716e-06,
+      "loss": 1.893,
+      "step": 1374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1229037952338924e-06,
+      "loss": 1.1668,
+      "step": 1375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1222418358340687e-06,
+      "loss": 2.2798,
+      "step": 1376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1215798764342454e-06,
+      "loss": 0.3074,
+      "step": 1377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.120917917034422e-06,
+      "loss": 0.9027,
+      "step": 1378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1202559576345984e-06,
+      "loss": 0.0907,
+      "step": 1379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.119593998234775e-06,
+      "loss": 3.4626,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1189320388349514e-06,
+      "loss": 1.249,
+      "step": 1381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.118270079435128e-06,
+      "loss": 0.6761,
+      "step": 1382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1176081200353044e-06,
+      "loss": 2.1771,
+      "step": 1383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.116946160635481e-06,
+      "loss": 0.683,
+      "step": 1384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.116284201235658e-06,
+      "loss": 2.0662,
+      "step": 1385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.115622241835834e-06,
+      "loss": 0.4104,
+      "step": 1386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1149602824360104e-06,
+      "loss": 1.5106,
+      "step": 1387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.114298323036187e-06,
+      "loss": 0.6791,
+      "step": 1388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.113636363636364e-06,
+      "loss": 1.8655,
+      "step": 1389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.11297440423654e-06,
+      "loss": 1.6319,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.112312444836717e-06,
+      "loss": 2.3962,
+      "step": 1391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.111650485436893e-06,
+      "loss": 2.0546,
+      "step": 1392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.11098852603707e-06,
+      "loss": 1.2994,
+      "step": 1393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.110326566637246e-06,
+      "loss": 1.0568,
+      "step": 1394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.109664607237423e-06,
+      "loss": 0.196,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.1090026478375996e-06,
+      "loss": 1.7118,
+      "step": 1396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2.108340688437776e-06,
+      "loss": 0.8532,
+      "step": 1397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.107678729037952e-06,
+      "loss": 1.4444,
+      "step": 1398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.107016769638129e-06,
+      "loss": 2.1393,
+      "step": 1399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1063548102383056e-06,
+      "loss": 0.8262,
+      "step": 1400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.105692850838482e-06,
+      "loss": 1.3399,
+      "step": 1401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1050308914386586e-06,
+      "loss": 2.678,
+      "step": 1402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1043689320388353e-06,
+      "loss": 1.5234,
+      "step": 1403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1037069726390116e-06,
+      "loss": 1.9401,
+      "step": 1404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.103045013239188e-06,
+      "loss": 0.3855,
+      "step": 1405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1023830538393646e-06,
+      "loss": 1.0105,
+      "step": 1406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1017210944395413e-06,
+      "loss": 0.9651,
+      "step": 1407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1010591350397176e-06,
+      "loss": 0.7337,
+      "step": 1408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.1003971756398943e-06,
+      "loss": 1.5012,
+      "step": 1409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0997352162400706e-06,
+      "loss": 1.7181,
+      "step": 1410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0990732568402473e-06,
+      "loss": 1.1654,
+      "step": 1411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0984112974404236e-06,
+      "loss": 1.3474,
+      "step": 1412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0977493380406003e-06,
+      "loss": 0.2233,
+      "step": 1413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.097087378640777e-06,
+      "loss": 0.5226,
+      "step": 1414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0964254192409533e-06,
+      "loss": 1.5682,
+      "step": 1415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0957634598411296e-06,
+      "loss": 0.7676,
+      "step": 1416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0951015004413063e-06,
+      "loss": 0.1267,
+      "step": 1417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.094439541041483e-06,
+      "loss": 1.5695,
+      "step": 1418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0937775816416593e-06,
+      "loss": 0.7057,
+      "step": 1419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.093115622241836e-06,
+      "loss": 1.1681,
+      "step": 1420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0924536628420123e-06,
+      "loss": 0.3141,
+      "step": 1421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.091791703442189e-06,
+      "loss": 1.292,
+      "step": 1422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0911297440423653e-06,
+      "loss": 0.7713,
+      "step": 1423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.090467784642542e-06,
+      "loss": 0.7125,
+      "step": 1424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0898058252427187e-06,
+      "loss": 0.2458,
+      "step": 1425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.089143865842895e-06,
+      "loss": 0.9201,
+      "step": 1426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0884819064430713e-06,
+      "loss": 0.8883,
+      "step": 1427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.087819947043248e-06,
+      "loss": 1.4413,
+      "step": 1428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0871579876434247e-06,
+      "loss": 0.7685,
+      "step": 1429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.086496028243601e-06,
+      "loss": 0.1893,
+      "step": 1430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0858340688437777e-06,
+      "loss": 0.9086,
+      "step": 1431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.085172109443954e-06,
+      "loss": 0.8493,
+      "step": 1432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0845101500441307e-06,
+      "loss": 0.8539,
+      "step": 1433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.083848190644307e-06,
+      "loss": 1.553,
+      "step": 1434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0831862312444837e-06,
+      "loss": 1.1848,
+      "step": 1435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0825242718446604e-06,
+      "loss": 0.8524,
+      "step": 1436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.081862312444837e-06,
+      "loss": 0.2919,
+      "step": 1437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.081200353045013e-06,
+      "loss": 1.4471,
+      "step": 1438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0805383936451897e-06,
+      "loss": 1.891,
+      "step": 1439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0798764342453664e-06,
+      "loss": 2.6845,
+      "step": 1440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0792144748455427e-06,
+      "loss": 2.3642,
+      "step": 1441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.0785525154457194e-06,
+      "loss": 0.158,
+      "step": 1442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 2.077890556045896e-06,
+      "loss": 1.1433,
+      "step": 1443
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0772285966460724e-06,
+      "loss": 0.107,
+      "step": 1444
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0765666372462487e-06,
+      "loss": 0.5639,
+      "step": 1445
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0759046778464254e-06,
+      "loss": 2.487,
+      "step": 1446
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.075242718446602e-06,
+      "loss": 1.1528,
+      "step": 1447
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0745807590467784e-06,
+      "loss": 2.0152,
+      "step": 1448
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.073918799646955e-06,
+      "loss": 1.3653,
+      "step": 1449
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0732568402471314e-06,
+      "loss": 2.4241,
+      "step": 1450
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.072594880847308e-06,
+      "loss": 0.8249,
+      "step": 1451
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0719329214474844e-06,
+      "loss": 1.6299,
+      "step": 1452
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.071270962047661e-06,
+      "loss": 1.961,
+      "step": 1453
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.070609002647838e-06,
+      "loss": 0.7922,
+      "step": 1454
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.069947043248014e-06,
+      "loss": 0.9914,
+      "step": 1455
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0692850838481904e-06,
+      "loss": 0.6289,
+      "step": 1456
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.068623124448367e-06,
+      "loss": 1.4992,
+      "step": 1457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.067961165048544e-06,
+      "loss": 1.1053,
+      "step": 1458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.06729920564872e-06,
+      "loss": 0.1275,
+      "step": 1459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.066637246248897e-06,
+      "loss": 1.2947,
+      "step": 1460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.065975286849073e-06,
+      "loss": 1.0179,
+      "step": 1461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.06531332744925e-06,
+      "loss": 3.2484,
+      "step": 1462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.064651368049426e-06,
+      "loss": 0.4137,
+      "step": 1463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.063989408649603e-06,
+      "loss": 2.1523,
+      "step": 1464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0633274492497796e-06,
+      "loss": 1.087,
+      "step": 1465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.062665489849956e-06,
+      "loss": 1.4492,
+      "step": 1466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.062003530450132e-06,
+      "loss": 0.6603,
+      "step": 1467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.061341571050309e-06,
+      "loss": 1.9843,
+      "step": 1468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0606796116504856e-06,
+      "loss": 0.0896,
+      "step": 1469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.060017652250662e-06,
+      "loss": 2.6252,
+      "step": 1470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0593556928508386e-06,
+      "loss": 0.6728,
+      "step": 1471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.058693733451015e-06,
+      "loss": 1.4885,
+      "step": 1472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0580317740511916e-06,
+      "loss": 0.2776,
+      "step": 1473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.057369814651368e-06,
+      "loss": 3.3625,
+      "step": 1474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0567078552515446e-06,
+      "loss": 1.568,
+      "step": 1475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0560458958517213e-06,
+      "loss": 2.2214,
+      "step": 1476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.055383936451898e-06,
+      "loss": 1.6246,
+      "step": 1477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0547219770520743e-06,
+      "loss": 1.6823,
+      "step": 1478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0540600176522506e-06,
+      "loss": 0.4871,
+      "step": 1479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0533980582524273e-06,
+      "loss": 0.905,
+      "step": 1480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0527360988526036e-06,
+      "loss": 1.0667,
+      "step": 1481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0520741394527803e-06,
+      "loss": 0.8641,
+      "step": 1482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.051412180052957e-06,
+      "loss": 0.9494,
+      "step": 1483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0507502206531333e-06,
+      "loss": 2.4684,
+      "step": 1484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0500882612533096e-06,
+      "loss": 1.2901,
+      "step": 1485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0494263018534863e-06,
+      "loss": 1.3838,
+      "step": 1486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.048764342453663e-06,
+      "loss": 1.3699,
+      "step": 1487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.0481023830538397e-06,
+      "loss": 1.9247,
+      "step": 1488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 2.047440423654016e-06,
+      "loss": 0.5264,
+      "step": 1489
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0467784642541923e-06,
+      "loss": 2.6399,
+      "step": 1490
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.046116504854369e-06,
+      "loss": 3.0466,
+      "step": 1491
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0454545454545453e-06,
+      "loss": 1.2794,
+      "step": 1492
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.044792586054722e-06,
+      "loss": 1.3654,
+      "step": 1493
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0441306266548987e-06,
+      "loss": 1.1315,
+      "step": 1494
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.043468667255075e-06,
+      "loss": 1.6307,
+      "step": 1495
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0428067078552513e-06,
+      "loss": 0.391,
+      "step": 1496
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.042144748455428e-06,
+      "loss": 1.8507,
+      "step": 1497
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0414827890556047e-06,
+      "loss": 1.1643,
+      "step": 1498
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0408208296557815e-06,
+      "loss": 0.9397,
+      "step": 1499
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0401588702559577e-06,
+      "loss": 1.7716,
+      "step": 1500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.039496910856134e-06,
+      "loss": 1.0888,
+      "step": 1501
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0388349514563107e-06,
+      "loss": 1.4861,
+      "step": 1502
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.038172992056487e-06,
+      "loss": 1.4859,
+      "step": 1503
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0375110326566637e-06,
+      "loss": 1.3005,
+      "step": 1504
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0368490732568405e-06,
+      "loss": 1.3809,
+      "step": 1505
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0361871138570167e-06,
+      "loss": 0.8385,
+      "step": 1506
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.035525154457193e-06,
+      "loss": 1.7139,
+      "step": 1507
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0348631950573697e-06,
+      "loss": 1.8533,
+      "step": 1508
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0342012356575465e-06,
+      "loss": 2.2419,
+      "step": 1509
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.033539276257723e-06,
+      "loss": 0.8817,
+      "step": 1510
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0328773168578995e-06,
+      "loss": 0.5025,
+      "step": 1511
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.032215357458076e-06,
+      "loss": 0.1592,
+      "step": 1512
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0315533980582525e-06,
+      "loss": 2.0758,
+      "step": 1513
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0308914386584288e-06,
+      "loss": 2.8407,
+      "step": 1514
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0302294792586055e-06,
+      "loss": 2.2008,
+      "step": 1515
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.029567519858782e-06,
+      "loss": 2.0269,
+      "step": 1516
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.028905560458959e-06,
+      "loss": 0.5022,
+      "step": 1517
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.028243601059135e-06,
+      "loss": 1.1576,
+      "step": 1518
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0275816416593115e-06,
+      "loss": 2.1806,
+      "step": 1519
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.026919682259488e-06,
+      "loss": 1.0167,
+      "step": 1520
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.026257722859665e-06,
+      "loss": 2.0559,
+      "step": 1521
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.025595763459841e-06,
+      "loss": 1.0335,
+      "step": 1522
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.024933804060018e-06,
+      "loss": 1.8881,
+      "step": 1523
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.024271844660194e-06,
+      "loss": 1.0298,
+      "step": 1524
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0236098852603705e-06,
+      "loss": 0.4487,
+      "step": 1525
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.022947925860547e-06,
+      "loss": 1.8407,
+      "step": 1526
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.022285966460724e-06,
+      "loss": 0.4137,
+      "step": 1527
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0216240070609006e-06,
+      "loss": 0.6531,
+      "step": 1528
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.020962047661077e-06,
+      "loss": 1.5136,
+      "step": 1529
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.020300088261253e-06,
+      "loss": 0.2823,
+      "step": 1530
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.01963812886143e-06,
+      "loss": 1.0487,
+      "step": 1531
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0189761694616066e-06,
+      "loss": 1.0295,
+      "step": 1532
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.018314210061783e-06,
+      "loss": 0.235,
+      "step": 1533
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 2.0176522506619596e-06,
+      "loss": 1.1896,
+      "step": 1534
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.016990291262136e-06,
+      "loss": 1.0655,
+      "step": 1535
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.016328331862312e-06,
+      "loss": 1.9494,
+      "step": 1536
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.015666372462489e-06,
+      "loss": 0.4925,
+      "step": 1537
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0150044130626656e-06,
+      "loss": 1.2802,
+      "step": 1538
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0143424536628423e-06,
+      "loss": 0.2274,
+      "step": 1539
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0136804942630186e-06,
+      "loss": 1.7434,
+      "step": 1540
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.013018534863195e-06,
+      "loss": 1.0468,
+      "step": 1541
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0123565754633716e-06,
+      "loss": 0.2826,
+      "step": 1542
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0116946160635483e-06,
+      "loss": 1.2691,
+      "step": 1543
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0110326566637246e-06,
+      "loss": 1.2584,
+      "step": 1544
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0103706972639013e-06,
+      "loss": 0.5287,
+      "step": 1545
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0097087378640776e-06,
+      "loss": 2.2869,
+      "step": 1546
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.009046778464254e-06,
+      "loss": 1.764,
+      "step": 1547
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0083848190644306e-06,
+      "loss": 1.2927,
+      "step": 1548
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0077228596646073e-06,
+      "loss": 0.8419,
+      "step": 1549
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.007060900264784e-06,
+      "loss": 0.9089,
+      "step": 1550
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0063989408649603e-06,
+      "loss": 2.7541,
+      "step": 1551
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.005736981465137e-06,
+      "loss": 0.6436,
+      "step": 1552
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0050750220653133e-06,
+      "loss": 0.6765,
+      "step": 1553
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0044130626654896e-06,
+      "loss": 0.7398,
+      "step": 1554
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0037511032656663e-06,
+      "loss": 0.1606,
+      "step": 1555
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.003089143865843e-06,
+      "loss": 0.7639,
+      "step": 1556
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0024271844660198e-06,
+      "loss": 1.8988,
+      "step": 1557
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.001765225066196e-06,
+      "loss": 0.0942,
+      "step": 1558
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.0011032656663723e-06,
+      "loss": 1.6459,
+      "step": 1559
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 2.000441306266549e-06,
+      "loss": 0.25,
+      "step": 1560
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9997793468667258e-06,
+      "loss": 0.0817,
+      "step": 1561
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.999117387466902e-06,
+      "loss": 2.7139,
+      "step": 1562
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9984554280670788e-06,
+      "loss": 1.7421,
+      "step": 1563
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.997793468667255e-06,
+      "loss": 0.9668,
+      "step": 1564
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9971315092674313e-06,
+      "loss": 2.0339,
+      "step": 1565
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.996469549867608e-06,
+      "loss": 1.3632,
+      "step": 1566
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9958075904677848e-06,
+      "loss": 0.585,
+      "step": 1567
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9951456310679615e-06,
+      "loss": 0.8579,
+      "step": 1568
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9944836716681378e-06,
+      "loss": 0.0539,
+      "step": 1569
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.993821712268314e-06,
+      "loss": 0.9879,
+      "step": 1570
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9931597528684908e-06,
+      "loss": 0.5146,
+      "step": 1571
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9924977934686675e-06,
+      "loss": 0.7738,
+      "step": 1572
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9918358340688438e-06,
+      "loss": 0.1996,
+      "step": 1573
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9911738746690205e-06,
+      "loss": 0.578,
+      "step": 1574
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9905119152691968e-06,
+      "loss": 1.169,
+      "step": 1575
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.989849955869373e-06,
+      "loss": 0.5696,
+      "step": 1576
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9891879964695498e-06,
+      "loss": 0.6832,
+      "step": 1577
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9885260370697265e-06,
+      "loss": 1.357,
+      "step": 1578
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.987864077669903e-06,
+      "loss": 0.5826,
+      "step": 1579
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9872021182700795e-06,
+      "loss": 0.2678,
+      "step": 1580
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9865401588702558e-06,
+      "loss": 0.2474,
+      "step": 1581
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9858781994704325e-06,
+      "loss": 0.7908,
+      "step": 1582
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.985216240070609e-06,
+      "loss": 1.0877,
+      "step": 1583
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9845542806707855e-06,
+      "loss": 1.9279,
+      "step": 1584
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.983892321270962e-06,
+      "loss": 0.5115,
+      "step": 1585
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.983230361871139e-06,
+      "loss": 1.1373,
+      "step": 1586
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9825684024713148e-06,
+      "loss": 2.0473,
+      "step": 1587
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9819064430714915e-06,
+      "loss": 0.3114,
+      "step": 1588
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.981244483671668e-06,
+      "loss": 2.4672,
+      "step": 1589
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.980582524271845e-06,
+      "loss": 1.8366,
+      "step": 1590
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.979920564872021e-06,
+      "loss": 3.5298,
+      "step": 1591
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.979258605472198e-06,
+      "loss": 1.2721,
+      "step": 1592
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.978596646072374e-06,
+      "loss": 0.8405,
+      "step": 1593
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.977934686672551e-06,
+      "loss": 0.1258,
+      "step": 1594
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.977272727272727e-06,
+      "loss": 0.9784,
+      "step": 1595
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.976610767872904e-06,
+      "loss": 1.8268,
+      "step": 1596
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9759488084730806e-06,
+      "loss": 0.1068,
+      "step": 1597
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.975286849073257e-06,
+      "loss": 0.3239,
+      "step": 1598
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.974624889673433e-06,
+      "loss": 0.7618,
+      "step": 1599
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.97396293027361e-06,
+      "loss": 0.0584,
+      "step": 1600
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9733009708737866e-06,
+      "loss": 0.362,
+      "step": 1601
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.972639011473963e-06,
+      "loss": 0.5423,
+      "step": 1602
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9719770520741396e-06,
+      "loss": 0.3126,
+      "step": 1603
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.971315092674316e-06,
+      "loss": 1.1569,
+      "step": 1604
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9706531332744926e-06,
+      "loss": 2.0746,
+      "step": 1605
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.969991173874669e-06,
+      "loss": 0.4425,
+      "step": 1606
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9693292144748456e-06,
+      "loss": 1.6803,
+      "step": 1607
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9686672550750224e-06,
+      "loss": 1.2065,
+      "step": 1608
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9680052956751986e-06,
+      "loss": 3.5752,
+      "step": 1609
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.967343336275375e-06,
+      "loss": 2.5476,
+      "step": 1610
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9666813768755516e-06,
+      "loss": 0.2863,
+      "step": 1611
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9660194174757284e-06,
+      "loss": 1.0306,
+      "step": 1612
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9653574580759046e-06,
+      "loss": 0.2829,
+      "step": 1613
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9646954986760814e-06,
+      "loss": 0.4781,
+      "step": 1614
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9640335392762576e-06,
+      "loss": 1.8106,
+      "step": 1615
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9633715798764344e-06,
+      "loss": 1.0906,
+      "step": 1616
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9627096204766106e-06,
+      "loss": 0.1271,
+      "step": 1617
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620476610767874e-06,
+      "loss": 0.6014,
+      "step": 1618
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.961385701676964e-06,
+      "loss": 0.619,
+      "step": 1619
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607237422771404e-06,
+      "loss": 1.2874,
+      "step": 1620
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600617828773166e-06,
+      "loss": 0.1498,
+      "step": 1621
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9593998234774934e-06,
+      "loss": 0.5299,
+      "step": 1622
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.95873786407767e-06,
+      "loss": 0.1692,
+      "step": 1623
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9580759046778464e-06,
+      "loss": 0.5189,
+      "step": 1624
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.957413945278023e-06,
+      "loss": 1.7987,
+      "step": 1625
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9567519858781998e-06,
+      "loss": 2.8382,
+      "step": 1626
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956090026478376e-06,
+      "loss": 0.6434,
+      "step": 1627
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9554280670785524e-06,
+      "loss": 1.4717,
+      "step": 1628
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.954766107678729e-06,
+      "loss": 3.64,
+      "step": 1629
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9541041482789058e-06,
+      "loss": 1.1335,
+      "step": 1630
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.953442188879082e-06,
+      "loss": 1.5807,
+      "step": 1631
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.952780229479259e-06,
+      "loss": 1.276,
+      "step": 1632
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.952118270079435e-06,
+      "loss": 0.4203,
+      "step": 1633
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.951456310679612e-06,
+      "loss": 2.2169,
+      "step": 1634
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.950794351279788e-06,
+      "loss": 0.6377,
+      "step": 1635
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.950132391879965e-06,
+      "loss": 1.2964,
+      "step": 1636
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9494704324801415e-06,
+      "loss": 1.1075,
+      "step": 1637
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.948808473080318e-06,
+      "loss": 0.5345,
+      "step": 1638
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.948146513680494e-06,
+      "loss": 0.6081,
+      "step": 1639
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.947484554280671e-06,
+      "loss": 1.5022,
+      "step": 1640
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9468225948808475e-06,
+      "loss": 1.0868,
+      "step": 1641
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.946160635481024e-06,
+      "loss": 0.9315,
+      "step": 1642
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9454986760812005e-06,
+      "loss": 0.3167,
+      "step": 1643
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.944836716681377e-06,
+      "loss": 1.3777,
+      "step": 1644
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9441747572815535e-06,
+      "loss": 2.1167,
+      "step": 1645
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.94351279788173e-06,
+      "loss": 1.2515,
+      "step": 1646
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9428508384819065e-06,
+      "loss": 3.6053,
+      "step": 1647
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9421888790820832e-06,
+      "loss": 0.5834,
+      "step": 1648
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9415269196822595e-06,
+      "loss": 0.8591,
+      "step": 1649
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.940864960282436e-06,
+      "loss": 1.1738,
+      "step": 1650
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9402030008826125e-06,
+      "loss": 0.0548,
+      "step": 1651
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9395410414827892e-06,
+      "loss": 2.1708,
+      "step": 1652
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9388790820829655e-06,
+      "loss": 1.4601,
+      "step": 1653
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9382171226831422e-06,
+      "loss": 0.4494,
+      "step": 1654
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9375551632833185e-06,
+      "loss": 1.2123,
+      "step": 1655
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9368932038834952e-06,
+      "loss": 1.7652,
+      "step": 1656
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9362312444836715e-06,
+      "loss": 1.5395,
+      "step": 1657
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9355692850838482e-06,
+      "loss": 2.7167,
+      "step": 1658
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.934907325684025e-06,
+      "loss": 0.5823,
+      "step": 1659
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9342453662842012e-06,
+      "loss": 0.0494,
+      "step": 1660
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.933583406884378e-06,
+      "loss": 0.6776,
+      "step": 1661
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9329214474845542e-06,
+      "loss": 0.4778,
+      "step": 1662
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.932259488084731e-06,
+      "loss": 0.1772,
+      "step": 1663
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9315975286849072e-06,
+      "loss": 0.5413,
+      "step": 1664
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.930935569285084e-06,
+      "loss": 0.3498,
+      "step": 1665
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9302736098852607e-06,
+      "loss": 0.2661,
+      "step": 1666
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.929611650485437e-06,
+      "loss": 1.2547,
+      "step": 1667
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9289496910856132e-06,
+      "loss": 0.1107,
+      "step": 1668
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.92828773168579e-06,
+      "loss": 1.1202,
+      "step": 1669
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9276257722859667e-06,
+      "loss": 0.1067,
+      "step": 1670
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.926963812886143e-06,
+      "loss": 1.5006,
+      "step": 1671
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9263018534863197e-06,
+      "loss": 0.4114,
+      "step": 1672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.925639894086496e-06,
+      "loss": 0.3751,
+      "step": 1673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9249779346866727e-06,
+      "loss": 0.4578,
+      "step": 1674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.924315975286849e-06,
+      "loss": 1.1294,
+      "step": 1675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9236540158870257e-06,
+      "loss": 1.9949,
+      "step": 1676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9229920564872024e-06,
+      "loss": 0.1636,
+      "step": 1677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9223300970873787e-06,
+      "loss": 0.7121,
+      "step": 1678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.921668137687555e-06,
+      "loss": 0.559,
+      "step": 1679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9210061782877317e-06,
+      "loss": 0.6816,
+      "step": 1680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9203442188879084e-06,
+      "loss": 1.2507,
+      "step": 1681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9196822594880847e-06,
+      "loss": 0.9879,
+      "step": 1682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9190203000882614e-06,
+      "loss": 0.9837,
+      "step": 1683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9183583406884377e-06,
+      "loss": 0.8686,
+      "step": 1684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9176963812886144e-06,
+      "loss": 0.2196,
+      "step": 1685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9170344218887907e-06,
+      "loss": 0.9045,
+      "step": 1686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9163724624889674e-06,
+      "loss": 1.1152,
+      "step": 1687
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.915710503089144e-06,
+      "loss": 1.9388,
+      "step": 1688
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9150485436893204e-06,
+      "loss": 2.0552,
+      "step": 1689
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9143865842894967e-06,
+      "loss": 1.2667,
+      "step": 1690
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9137246248896734e-06,
+      "loss": 5.6166,
+      "step": 1691
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.91306266548985e-06,
+      "loss": 0.6645,
+      "step": 1692
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9124007060900264e-06,
+      "loss": 0.4933,
+      "step": 1693
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.911738746690203e-06,
+      "loss": 0.2732,
+      "step": 1694
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.91107678729038e-06,
+      "loss": 1.55,
+      "step": 1695
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.910414827890556e-06,
+      "loss": 0.3817,
+      "step": 1696
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9097528684907324e-06,
+      "loss": 0.2025,
+      "step": 1697
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.909090909090909e-06,
+      "loss": 1.198,
+      "step": 1698
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.908428949691086e-06,
+      "loss": 3.1669,
+      "step": 1699
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9077669902912625e-06,
+      "loss": 2.7306,
+      "step": 1700
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9071050308914386e-06,
+      "loss": 2.0556,
+      "step": 1701
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9064430714916153e-06,
+      "loss": 0.6592,
+      "step": 1702
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9057811120917918e-06,
+      "loss": 1.7921,
+      "step": 1703
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.905119152691968e-06,
+      "loss": 0.5029,
+      "step": 1704
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9044571932921448e-06,
+      "loss": 1.3624,
+      "step": 1705
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9037952338923213e-06,
+      "loss": 5.0414,
+      "step": 1706
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.903133274492498e-06,
+      "loss": 0.6695,
+      "step": 1707
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9024713150926743e-06,
+      "loss": 1.6344,
+      "step": 1708
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9018093556928508e-06,
+      "loss": 0.543,
+      "step": 1709
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9011473962930275e-06,
+      "loss": 0.9939,
+      "step": 1710
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.900485436893204e-06,
+      "loss": 1.6121,
+      "step": 1711
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8998234774933803e-06,
+      "loss": 1.9887,
+      "step": 1712
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.899161518093557e-06,
+      "loss": 1.4865,
+      "step": 1713
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8984995586937335e-06,
+      "loss": 1.1982,
+      "step": 1714
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8978375992939098e-06,
+      "loss": 2.1832,
+      "step": 1715
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8971756398940865e-06,
+      "loss": 0.6717,
+      "step": 1716
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.896513680494263e-06,
+      "loss": 1.2363,
+      "step": 1717
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8958517210944397e-06,
+      "loss": 2.0327,
+      "step": 1718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.895189761694616e-06,
+      "loss": 1.0095,
+      "step": 1719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8945278022947925e-06,
+      "loss": 1.424,
+      "step": 1720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8938658428949693e-06,
+      "loss": 0.5078,
+      "step": 1721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8932038834951458e-06,
+      "loss": 1.1769,
+      "step": 1722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.892541924095322e-06,
+      "loss": 0.6633,
+      "step": 1723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8918799646954988e-06,
+      "loss": 0.9703,
+      "step": 1724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8912180052956753e-06,
+      "loss": 0.1348,
+      "step": 1725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8905560458958515e-06,
+      "loss": 2.3768,
+      "step": 1726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8898940864960283e-06,
+      "loss": 0.8802,
+      "step": 1727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8892321270962048e-06,
+      "loss": 0.2049,
+      "step": 1728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8885701676963815e-06,
+      "loss": 0.3937,
+      "step": 1729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8879082082965578e-06,
+      "loss": 0.1341,
+      "step": 1730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8872462488967343e-06,
+      "loss": 1.6549,
+      "step": 1731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.886584289496911e-06,
+      "loss": 0.9812,
+      "step": 1732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8859223300970877e-06,
+      "loss": 0.5661,
+      "step": 1733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.885260370697264e-06,
+      "loss": 0.945,
+      "step": 1734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8845984112974405e-06,
+      "loss": 1.6126,
+      "step": 1735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8839364518976172e-06,
+      "loss": 0.9289,
+      "step": 1736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8832744924977935e-06,
+      "loss": 1.5085,
+      "step": 1737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.88261253309797e-06,
+      "loss": 1.1994,
+      "step": 1738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8819505736981467e-06,
+      "loss": 1.0566,
+      "step": 1739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8812886142983232e-06,
+      "loss": 0.9587,
+      "step": 1740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8806266548984995e-06,
+      "loss": 0.823,
+      "step": 1741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8799646954986762e-06,
+      "loss": 2.3499,
+      "step": 1742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8793027360988527e-06,
+      "loss": 1.6002,
+      "step": 1743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8786407766990294e-06,
+      "loss": 0.786,
+      "step": 1744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8779788172992057e-06,
+      "loss": 0.3947,
+      "step": 1745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8773168578993822e-06,
+      "loss": 0.6193,
+      "step": 1746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.876654898499559e-06,
+      "loss": 2.3605,
+      "step": 1747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8759929390997352e-06,
+      "loss": 0.151,
+      "step": 1748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8753309796999117e-06,
+      "loss": 0.9195,
+      "step": 1749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8746690203000884e-06,
+      "loss": 0.7345,
+      "step": 1750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.874007060900265e-06,
+      "loss": 0.8927,
+      "step": 1751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8733451015004412e-06,
+      "loss": 0.6023,
+      "step": 1752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.872683142100618e-06,
+      "loss": 1.2697,
+      "step": 1753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8720211827007944e-06,
+      "loss": 0.0895,
+      "step": 1754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8713592233009707e-06,
+      "loss": 2.484,
+      "step": 1755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8706972639011474e-06,
+      "loss": 1.4102,
+      "step": 1756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.870035304501324e-06,
+      "loss": 0.7424,
+      "step": 1757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8693733451015006e-06,
+      "loss": 1.6838,
+      "step": 1758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.868711385701677e-06,
+      "loss": 1.4741,
+      "step": 1759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8680494263018534e-06,
+      "loss": 3.8656,
+      "step": 1760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8673874669020301e-06,
+      "loss": 2.6887,
+      "step": 1761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8667255075022066e-06,
+      "loss": 1.6553,
+      "step": 1762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.866063548102383e-06,
+      "loss": 0.5171,
+      "step": 1763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8654015887025596e-06,
+      "loss": 1.3369,
+      "step": 1764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8647396293027361e-06,
+      "loss": 1.1581,
+      "step": 1765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8640776699029124e-06,
+      "loss": 0.5652,
+      "step": 1766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8634157105030891e-06,
+      "loss": 0.6752,
+      "step": 1767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8627537511032658e-06,
+      "loss": 0.5779,
+      "step": 1768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8620917917034423e-06,
+      "loss": 1.3455,
+      "step": 1769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8614298323036186e-06,
+      "loss": 1.1944,
+      "step": 1770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8607678729037953e-06,
+      "loss": 1.0069,
+      "step": 1771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8601059135039718e-06,
+      "loss": 2.5492,
+      "step": 1772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8594439541041486e-06,
+      "loss": 1.2157,
+      "step": 1773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8587819947043248e-06,
+      "loss": 0.3,
+      "step": 1774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8581200353045013e-06,
+      "loss": 0.2546,
+      "step": 1775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.857458075904678e-06,
+      "loss": 1.5297,
+      "step": 1776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8567961165048543e-06,
+      "loss": 0.5232,
+      "step": 1777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8561341571050308e-06,
+      "loss": 1.9843,
+      "step": 1778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8554721977052076e-06,
+      "loss": 0.3959,
+      "step": 1779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.854810238305384e-06,
+      "loss": 1.8716,
+      "step": 1780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8541482789055603e-06,
+      "loss": 1.0618,
+      "step": 1781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.853486319505737e-06,
+      "loss": 1.9217,
+      "step": 1782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8528243601059136e-06,
+      "loss": 1.1236,
+      "step": 1783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8521624007060903e-06,
+      "loss": 1.6569,
+      "step": 1784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8515004413062666e-06,
+      "loss": 0.8059,
+      "step": 1785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.850838481906443e-06,
+      "loss": 0.1277,
+      "step": 1786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8501765225066198e-06,
+      "loss": 0.8942,
+      "step": 1787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.849514563106796e-06,
+      "loss": 0.9274,
+      "step": 1788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8488526037069726e-06,
+      "loss": 1.5231,
+      "step": 1789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8481906443071493e-06,
+      "loss": 0.258,
+      "step": 1790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8475286849073258e-06,
+      "loss": 0.4729,
+      "step": 1791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.846866725507502e-06,
+      "loss": 1.4811,
+      "step": 1792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8462047661076788e-06,
+      "loss": 0.8935,
+      "step": 1793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8455428067078553e-06,
+      "loss": 1.7969,
+      "step": 1794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.844880847308032e-06,
+      "loss": 1.3316,
+      "step": 1795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8442188879082083e-06,
+      "loss": 1.7299,
+      "step": 1796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8435569285083848e-06,
+      "loss": 1.5738,
+      "step": 1797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8428949691085615e-06,
+      "loss": 1.2671,
+      "step": 1798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8422330097087378e-06,
+      "loss": 2.5787,
+      "step": 1799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8415710503089143e-06,
+      "loss": 0.472,
+      "step": 1800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.840909090909091e-06,
+      "loss": 1.7374,
+      "step": 1801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8402471315092675e-06,
+      "loss": 1.9176,
+      "step": 1802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8395851721094438e-06,
+      "loss": 1.18,
+      "step": 1803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8389232127096205e-06,
+      "loss": 1.0282,
+      "step": 1804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8382612533097972e-06,
+      "loss": 0.4748,
+      "step": 1805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8375992939099737e-06,
+      "loss": 1.5105,
+      "step": 1806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.83693733451015e-06,
+      "loss": 0.1376,
+      "step": 1807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8362753751103267e-06,
+      "loss": 1.6928,
+      "step": 1808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8356134157105032e-06,
+      "loss": 0.4812,
+      "step": 1809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8349514563106795e-06,
+      "loss": 1.0858,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8342894969108562e-06,
+      "loss": 1.433,
+      "step": 1811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8336275375110327e-06,
+      "loss": 0.8005,
+      "step": 1812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8329655781112094e-06,
+      "loss": 0.491,
+      "step": 1813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8323036187113857e-06,
+      "loss": 1.6689,
+      "step": 1814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8316416593115622e-06,
+      "loss": 0.7211,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.830979699911739e-06,
+      "loss": 1.0004,
+      "step": 1816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8303177405119154e-06,
+      "loss": 1.5806,
+      "step": 1817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8296557811120917e-06,
+      "loss": 0.9306,
+      "step": 1818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8289938217122684e-06,
+      "loss": 2.2408,
+      "step": 1819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.828331862312445e-06,
+      "loss": 0.9132,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8276699029126212e-06,
+      "loss": 1.2587,
+      "step": 1821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.827007943512798e-06,
+      "loss": 1.4221,
+      "step": 1822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8263459841129744e-06,
+      "loss": 0.9944,
+      "step": 1823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8256840247131511e-06,
+      "loss": 1.3006,
+      "step": 1824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8250220653133274e-06,
+      "loss": 0.1276,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.824360105913504e-06,
+      "loss": 2.1775,
+      "step": 1826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8236981465136806e-06,
+      "loss": 0.5248,
+      "step": 1827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8230361871138571e-06,
+      "loss": 1.8159,
+      "step": 1828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8223742277140334e-06,
+      "loss": 1.3559,
+      "step": 1829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8217122683142101e-06,
+      "loss": 0.2612,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8210503089143866e-06,
+      "loss": 0.6594,
+      "step": 1831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.820388349514563e-06,
+      "loss": 1.8094,
+      "step": 1832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8197263901147396e-06,
+      "loss": 0.8892,
+      "step": 1833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8190644307149162e-06,
+      "loss": 0.059,
+      "step": 1834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8184024713150929e-06,
+      "loss": 3.6407,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8177405119152692e-06,
+      "loss": 1.0768,
+      "step": 1836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8170785525154457e-06,
+      "loss": 2.1138,
+      "step": 1837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8164165931156224e-06,
+      "loss": 0.2047,
+      "step": 1838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.815754633715799e-06,
+      "loss": 1.9206,
+      "step": 1839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8150926743159752e-06,
+      "loss": 0.2845,
+      "step": 1840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8144307149161519e-06,
+      "loss": 1.907,
+      "step": 1841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8137687555163286e-06,
+      "loss": 2.3919,
+      "step": 1842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8131067961165047e-06,
+      "loss": 0.6209,
+      "step": 1843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8124448367166814e-06,
+      "loss": 0.9145,
+      "step": 1844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.811782877316858e-06,
+      "loss": 0.1768,
+      "step": 1845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8111209179170346e-06,
+      "loss": 2.4067,
+      "step": 1846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8104589585172109e-06,
+      "loss": 1.3652,
+      "step": 1847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8097969991173876e-06,
+      "loss": 2.4704,
+      "step": 1848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.809135039717564e-06,
+      "loss": 0.8898,
+      "step": 1849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8084730803177408e-06,
+      "loss": 1.8342,
+      "step": 1850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.807811120917917e-06,
+      "loss": 0.9522,
+      "step": 1851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8071491615180936e-06,
+      "loss": 0.6497,
+      "step": 1852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8064872021182703e-06,
+      "loss": 1.6562,
+      "step": 1853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8058252427184466e-06,
+      "loss": 1.341,
+      "step": 1854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.805163283318623e-06,
+      "loss": 1.0784,
+      "step": 1855
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8045013239187998e-06,
+      "loss": 1.352,
+      "step": 1856
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8038393645189763e-06,
+      "loss": 0.1277,
+      "step": 1857
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8031774051191526e-06,
+      "loss": 0.2103,
+      "step": 1858
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8025154457193293e-06,
+      "loss": 0.5741,
+      "step": 1859
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8018534863195058e-06,
+      "loss": 1.208,
+      "step": 1860
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.801191526919682e-06,
+      "loss": 0.6136,
+      "step": 1861
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8005295675198588e-06,
+      "loss": 0.8178,
+      "step": 1862
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7998676081200353e-06,
+      "loss": 1.6698,
+      "step": 1863
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.799205648720212e-06,
+      "loss": 2.2645,
+      "step": 1864
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7985436893203883e-06,
+      "loss": 1.0145,
+      "step": 1865
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7978817299205648e-06,
+      "loss": 0.6235,
+      "step": 1866
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7972197705207415e-06,
+      "loss": 1.4861,
+      "step": 1867
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.796557811120918e-06,
+      "loss": 0.0847,
+      "step": 1868
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7958958517210943e-06,
+      "loss": 0.3052,
+      "step": 1869
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.795233892321271e-06,
+      "loss": 0.0556,
+      "step": 1870
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7945719329214475e-06,
+      "loss": 0.5411,
+      "step": 1871
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7939099735216238e-06,
+      "loss": 2.1553,
+      "step": 1872
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7932480141218005e-06,
+      "loss": 0.086,
+      "step": 1873
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.792586054721977e-06,
+      "loss": 1.9746,
+      "step": 1874
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7919240953221537e-06,
+      "loss": 0.7666,
+      "step": 1875
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.79126213592233e-06,
+      "loss": 0.3415,
+      "step": 1876
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7906001765225065e-06,
+      "loss": 1.6629,
+      "step": 1877
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7899382171226832e-06,
+      "loss": 0.3562,
+      "step": 1878
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.78927625772286e-06,
+      "loss": 0.787,
+      "step": 1879
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7886142983230362e-06,
+      "loss": 1.1807,
+      "step": 1880
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7879523389232127e-06,
+      "loss": 1.2611,
+      "step": 1881
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7872903795233895e-06,
+      "loss": 0.5096,
+      "step": 1882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7866284201235657e-06,
+      "loss": 0.5422,
+      "step": 1883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7859664607237422e-06,
+      "loss": 0.7778,
+      "step": 1884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.785304501323919e-06,
+      "loss": 2.9028,
+      "step": 1885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7846425419240955e-06,
+      "loss": 0.6137,
+      "step": 1886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7839805825242717e-06,
+      "loss": 0.2414,
+      "step": 1887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7833186231244485e-06,
+      "loss": 0.6782,
+      "step": 1888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.782656663724625e-06,
+      "loss": 0.1831,
+      "step": 1889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7819947043248017e-06,
+      "loss": 0.2734,
+      "step": 1890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.781332744924978e-06,
+      "loss": 0.1695,
+      "step": 1891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7806707855251545e-06,
+      "loss": 1.5511,
+      "step": 1892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7800088261253312e-06,
+      "loss": 0.9448,
+      "step": 1893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7793468667255075e-06,
+      "loss": 2.1124,
+      "step": 1894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.778684907325684e-06,
+      "loss": 0.5322,
+      "step": 1895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7780229479258607e-06,
+      "loss": 1.3421,
+      "step": 1896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7773609885260372e-06,
+      "loss": 0.8151,
+      "step": 1897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7766990291262135e-06,
+      "loss": 1.3607,
+      "step": 1898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7760370697263902e-06,
+      "loss": 0.654,
+      "step": 1899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7753751103265667e-06,
+      "loss": 0.6436,
+      "step": 1900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.7747131509267434e-06,
+      "loss": 0.5657,
+      "step": 1901
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7740511915269197e-06,
+      "loss": 1.0367,
+      "step": 1902
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7733892321270962e-06,
+      "loss": 1.6427,
+      "step": 1903
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7727272727272729e-06,
+      "loss": 1.4411,
+      "step": 1904
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7720653133274492e-06,
+      "loss": 0.7721,
+      "step": 1905
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7714033539276257e-06,
+      "loss": 0.2371,
+      "step": 1906
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7707413945278024e-06,
+      "loss": 1.5582,
+      "step": 1907
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7700794351279789e-06,
+      "loss": 0.589,
+      "step": 1908
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7694174757281552e-06,
+      "loss": 1.8177,
+      "step": 1909
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7687555163283319e-06,
+      "loss": 2.4551,
+      "step": 1910
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7680935569285084e-06,
+      "loss": 0.3924,
+      "step": 1911
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.767431597528685e-06,
+      "loss": 0.8252,
+      "step": 1912
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7667696381288614e-06,
+      "loss": 0.4055,
+      "step": 1913
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.766107678729038e-06,
+      "loss": 0.233,
+      "step": 1914
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7654457193292146e-06,
+      "loss": 0.5621,
+      "step": 1915
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.764783759929391e-06,
+      "loss": 0.3034,
+      "step": 1916
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7641218005295676e-06,
+      "loss": 0.3873,
+      "step": 1917
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7634598411297441e-06,
+      "loss": 0.5525,
+      "step": 1918
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7627978817299208e-06,
+      "loss": 1.7346,
+      "step": 1919
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7621359223300971e-06,
+      "loss": 1.731,
+      "step": 1920
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7614739629302736e-06,
+      "loss": 0.7604,
+      "step": 1921
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7608120035304503e-06,
+      "loss": 0.2297,
+      "step": 1922
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7601500441306268e-06,
+      "loss": 0.4841,
+      "step": 1923
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7594880847308031e-06,
+      "loss": 1.8074,
+      "step": 1924
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7588261253309798e-06,
+      "loss": 1.5278,
+      "step": 1925
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7581641659311563e-06,
+      "loss": 0.7109,
+      "step": 1926
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7575022065313326e-06,
+      "loss": 0.907,
+      "step": 1927
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7568402471315093e-06,
+      "loss": 1.1328,
+      "step": 1928
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7561782877316858e-06,
+      "loss": 0.6453,
+      "step": 1929
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7555163283318625e-06,
+      "loss": 1.5173,
+      "step": 1930
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7548543689320388e-06,
+      "loss": 1.0506,
+      "step": 1931
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7541924095322153e-06,
+      "loss": 1.4224,
+      "step": 1932
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.753530450132392e-06,
+      "loss": 0.6415,
+      "step": 1933
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7528684907325685e-06,
+      "loss": 0.1932,
+      "step": 1934
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7522065313327448e-06,
+      "loss": 3.1459,
+      "step": 1935
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7515445719329215e-06,
+      "loss": 0.9673,
+      "step": 1936
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.750882612533098e-06,
+      "loss": 1.9356,
+      "step": 1937
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7502206531332743e-06,
+      "loss": 0.0751,
+      "step": 1938
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.749558693733451e-06,
+      "loss": 1.3668,
+      "step": 1939
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7488967343336275e-06,
+      "loss": 1.0542,
+      "step": 1940
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7482347749338043e-06,
+      "loss": 0.4398,
+      "step": 1941
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7475728155339805e-06,
+      "loss": 2.2161,
+      "step": 1942
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.746910856134157e-06,
+      "loss": 0.1646,
+      "step": 1943
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7462488967343338e-06,
+      "loss": 1.1391,
+      "step": 1944
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7455869373345103e-06,
+      "loss": 1.0208,
+      "step": 1945
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7449249779346865e-06,
+      "loss": 0.0709,
+      "step": 1946
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.7442630185348633e-06,
+      "loss": 1.8067,
+      "step": 1947
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7436010591350398e-06,
+      "loss": 2.2223,
+      "step": 1948
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.742939099735216e-06,
+      "loss": 1.2489,
+      "step": 1949
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7422771403353928e-06,
+      "loss": 0.1727,
+      "step": 1950
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7416151809355695e-06,
+      "loss": 1.3567,
+      "step": 1951
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.740953221535746e-06,
+      "loss": 0.5286,
+      "step": 1952
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7402912621359223e-06,
+      "loss": 2.0313,
+      "step": 1953
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.739629302736099e-06,
+      "loss": 2.524,
+      "step": 1954
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7389673433362755e-06,
+      "loss": 0.7508,
+      "step": 1955
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7383053839364518e-06,
+      "loss": 0.7559,
+      "step": 1956
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7376434245366285e-06,
+      "loss": 0.4702,
+      "step": 1957
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.736981465136805e-06,
+      "loss": 0.2943,
+      "step": 1958
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7363195057369817e-06,
+      "loss": 0.3048,
+      "step": 1959
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.735657546337158e-06,
+      "loss": 1.7535,
+      "step": 1960
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7349955869373345e-06,
+      "loss": 1.1308,
+      "step": 1961
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7343336275375112e-06,
+      "loss": 1.2815,
+      "step": 1962
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7336716681376877e-06,
+      "loss": 1.8607,
+      "step": 1963
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.733009708737864e-06,
+      "loss": 0.1415,
+      "step": 1964
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7323477493380407e-06,
+      "loss": 0.119,
+      "step": 1965
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7316857899382172e-06,
+      "loss": 0.6686,
+      "step": 1966
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7310238305383935e-06,
+      "loss": 3.0599,
+      "step": 1967
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7303618711385702e-06,
+      "loss": 1.6641,
+      "step": 1968
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7296999117387467e-06,
+      "loss": 1.9683,
+      "step": 1969
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7290379523389234e-06,
+      "loss": 2.2973,
+      "step": 1970
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7283759929390997e-06,
+      "loss": 0.7843,
+      "step": 1971
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7277140335392762e-06,
+      "loss": 0.9502,
+      "step": 1972
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.727052074139453e-06,
+      "loss": 1.5938,
+      "step": 1973
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7263901147396294e-06,
+      "loss": 1.9037,
+      "step": 1974
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7257281553398057e-06,
+      "loss": 1.8108,
+      "step": 1975
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7250661959399824e-06,
+      "loss": 1.5522,
+      "step": 1976
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.724404236540159e-06,
+      "loss": 0.3814,
+      "step": 1977
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7237422771403352e-06,
+      "loss": 0.5185,
+      "step": 1978
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.723080317740512e-06,
+      "loss": 0.0572,
+      "step": 1979
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7224183583406884e-06,
+      "loss": 1.7938,
+      "step": 1980
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7217563989408651e-06,
+      "loss": 0.9903,
+      "step": 1981
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7210944395410414e-06,
+      "loss": 0.9034,
+      "step": 1982
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.720432480141218e-06,
+      "loss": 0.8781,
+      "step": 1983
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7197705207413946e-06,
+      "loss": 2.4135,
+      "step": 1984
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7191085613415713e-06,
+      "loss": 0.3867,
+      "step": 1985
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7184466019417474e-06,
+      "loss": 1.1058,
+      "step": 1986
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7177846425419241e-06,
+      "loss": 1.4832,
+      "step": 1987
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7171226831421008e-06,
+      "loss": 0.3473,
+      "step": 1988
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.716460723742277e-06,
+      "loss": 1.4286,
+      "step": 1989
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7157987643424536e-06,
+      "loss": 0.7385,
+      "step": 1990
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7151368049426303e-06,
+      "loss": 4.713,
+      "step": 1991
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7144748455428068e-06,
+      "loss": 1.3669,
+      "step": 1992
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.7138128861429831e-06,
+      "loss": 1.5221,
+      "step": 1993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7131509267431598e-06,
+      "loss": 1.0008,
+      "step": 1994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7124889673433364e-06,
+      "loss": 0.8041,
+      "step": 1995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.711827007943513e-06,
+      "loss": 1.4771,
+      "step": 1996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7111650485436894e-06,
+      "loss": 1.0963,
+      "step": 1997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7105030891438659e-06,
+      "loss": 1.0733,
+      "step": 1998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7098411297440426e-06,
+      "loss": 1.3726,
+      "step": 1999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7091791703442189e-06,
+      "loss": 0.3041,
+      "step": 2000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7085172109443954e-06,
+      "loss": 0.9802,
+      "step": 2001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.707855251544572e-06,
+      "loss": 0.2351,
+      "step": 2002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7071932921447486e-06,
+      "loss": 1.1996,
+      "step": 2003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7065313327449249e-06,
+      "loss": 2.0743,
+      "step": 2004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7058693733451016e-06,
+      "loss": 1.6267,
+      "step": 2005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.705207413945278e-06,
+      "loss": 0.6134,
+      "step": 2006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7045454545454548e-06,
+      "loss": 0.909,
+      "step": 2007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.703883495145631e-06,
+      "loss": 2.5859,
+      "step": 2008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7032215357458076e-06,
+      "loss": 0.1211,
+      "step": 2009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7025595763459843e-06,
+      "loss": 1.3047,
+      "step": 2010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7018976169461606e-06,
+      "loss": 1.9629,
+      "step": 2011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.701235657546337e-06,
+      "loss": 0.6681,
+      "step": 2012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.7005736981465138e-06,
+      "loss": 2.0541,
+      "step": 2013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6999117387466903e-06,
+      "loss": 1.5023,
+      "step": 2014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6992497793468666e-06,
+      "loss": 1.0336,
+      "step": 2015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6985878199470433e-06,
+      "loss": 0.2717,
+      "step": 2016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6979258605472198e-06,
+      "loss": 1.328,
+      "step": 2017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6972639011473965e-06,
+      "loss": 1.804,
+      "step": 2018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6966019417475728e-06,
+      "loss": 3.8386,
+      "step": 2019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6959399823477493e-06,
+      "loss": 1.1629,
+      "step": 2020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.695278022947926e-06,
+      "loss": 1.7503,
+      "step": 2021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6946160635481023e-06,
+      "loss": 0.51,
+      "step": 2022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6939541041482788e-06,
+      "loss": 0.2166,
+      "step": 2023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6932921447484555e-06,
+      "loss": 1.4553,
+      "step": 2024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6926301853486322e-06,
+      "loss": 0.8748,
+      "step": 2025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6919682259488085e-06,
+      "loss": 0.9364,
+      "step": 2026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.691306266548985e-06,
+      "loss": 2.8698,
+      "step": 2027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6906443071491617e-06,
+      "loss": 0.6781,
+      "step": 2028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6899823477493382e-06,
+      "loss": 1.041,
+      "step": 2029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6893203883495145e-06,
+      "loss": 1.0757,
+      "step": 2030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6886584289496912e-06,
+      "loss": 1.0789,
+      "step": 2031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6879964695498677e-06,
+      "loss": 0.5857,
+      "step": 2032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.687334510150044e-06,
+      "loss": 0.5951,
+      "step": 2033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6866725507502207e-06,
+      "loss": 0.1286,
+      "step": 2034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6860105913503972e-06,
+      "loss": 1.2443,
+      "step": 2035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.685348631950574e-06,
+      "loss": 1.2797,
+      "step": 2036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6846866725507502e-06,
+      "loss": 1.9463,
+      "step": 2037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.6840247131509267e-06,
+      "loss": 2.1438,
+      "step": 2038
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6833627537511034e-06,
+      "loss": 0.5506,
+      "step": 2039
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.68270079435128e-06,
+      "loss": 0.8783,
+      "step": 2040
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6820388349514562e-06,
+      "loss": 1.5594,
+      "step": 2041
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.681376875551633e-06,
+      "loss": 0.3533,
+      "step": 2042
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6807149161518094e-06,
+      "loss": 0.5129,
+      "step": 2043
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6800529567519857e-06,
+      "loss": 1.6014,
+      "step": 2044
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6793909973521624e-06,
+      "loss": 2.6986,
+      "step": 2045
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.678729037952339e-06,
+      "loss": 0.8883,
+      "step": 2046
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6780670785525157e-06,
+      "loss": 0.9906,
+      "step": 2047
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.677405119152692e-06,
+      "loss": 0.2038,
+      "step": 2048
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6767431597528684e-06,
+      "loss": 0.1148,
+      "step": 2049
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6760812003530452e-06,
+      "loss": 0.7765,
+      "step": 2050
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6754192409532217e-06,
+      "loss": 0.7536,
+      "step": 2051
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.674757281553398e-06,
+      "loss": 0.2724,
+      "step": 2052
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6740953221535747e-06,
+      "loss": 1.5357,
+      "step": 2053
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6734333627537512e-06,
+      "loss": 1.6165,
+      "step": 2054
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6727714033539274e-06,
+      "loss": 1.4772,
+      "step": 2055
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6721094439541042e-06,
+      "loss": 0.6278,
+      "step": 2056
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6714474845542807e-06,
+      "loss": 0.0489,
+      "step": 2057
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6707855251544574e-06,
+      "loss": 4.5698,
+      "step": 2058
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6701235657546337e-06,
+      "loss": 0.2331,
+      "step": 2059
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6694616063548102e-06,
+      "loss": 0.8666,
+      "step": 2060
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6687996469549869e-06,
+      "loss": 0.5485,
+      "step": 2061
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6681376875551632e-06,
+      "loss": 0.5842,
+      "step": 2062
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6674757281553399e-06,
+      "loss": 0.3699,
+      "step": 2063
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6668137687555164e-06,
+      "loss": 0.2818,
+      "step": 2064
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.666151809355693e-06,
+      "loss": 1.2669,
+      "step": 2065
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6654898499558694e-06,
+      "loss": 0.8181,
+      "step": 2066
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6648278905560459e-06,
+      "loss": 0.9963,
+      "step": 2067
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6641659311562226e-06,
+      "loss": 1.0145,
+      "step": 2068
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.663503971756399e-06,
+      "loss": 1.5539,
+      "step": 2069
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6628420123565754e-06,
+      "loss": 2.3942,
+      "step": 2070
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.662180052956752e-06,
+      "loss": 2.0028,
+      "step": 2071
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6615180935569286e-06,
+      "loss": 1.5743,
+      "step": 2072
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6608561341571049e-06,
+      "loss": 0.8074,
+      "step": 2073
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6601941747572816e-06,
+      "loss": 1.5209,
+      "step": 2074
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.659532215357458e-06,
+      "loss": 0.243,
+      "step": 2075
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6588702559576348e-06,
+      "loss": 0.4929,
+      "step": 2076
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.658208296557811e-06,
+      "loss": 1.8297,
+      "step": 2077
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6575463371579876e-06,
+      "loss": 0.7977,
+      "step": 2078
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6568843777581643e-06,
+      "loss": 1.2966,
+      "step": 2079
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6562224183583408e-06,
+      "loss": 2.139,
+      "step": 2080
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.655560458958517e-06,
+      "loss": 0.5554,
+      "step": 2081
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6548984995586938e-06,
+      "loss": 0.3714,
+      "step": 2082
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6542365401588703e-06,
+      "loss": 1.8137,
+      "step": 2083
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.6535745807590466e-06,
+      "loss": 1.246,
+      "step": 2084
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6529126213592233e-06,
+      "loss": 0.5257,
+      "step": 2085
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6522506619593998e-06,
+      "loss": 2.1995,
+      "step": 2086
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6515887025595765e-06,
+      "loss": 0.0839,
+      "step": 2087
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6509267431597528e-06,
+      "loss": 0.5769,
+      "step": 2088
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6502647837599293e-06,
+      "loss": 1.4155,
+      "step": 2089
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.649602824360106e-06,
+      "loss": 1.3551,
+      "step": 2090
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6489408649602825e-06,
+      "loss": 0.4147,
+      "step": 2091
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6482789055604588e-06,
+      "loss": 1.0577,
+      "step": 2092
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6476169461606355e-06,
+      "loss": 0.3236,
+      "step": 2093
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.646954986760812e-06,
+      "loss": 1.6333,
+      "step": 2094
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6462930273609883e-06,
+      "loss": 2.2959,
+      "step": 2095
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.645631067961165e-06,
+      "loss": 0.875,
+      "step": 2096
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6449691085613417e-06,
+      "loss": 0.9391,
+      "step": 2097
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6443071491615182e-06,
+      "loss": 1.9846,
+      "step": 2098
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6436451897616945e-06,
+      "loss": 1.5092,
+      "step": 2099
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6429832303618712e-06,
+      "loss": 0.8601,
+      "step": 2100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6423212709620477e-06,
+      "loss": 0.679,
+      "step": 2101
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6416593115622245e-06,
+      "loss": 0.8846,
+      "step": 2102
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6409973521624007e-06,
+      "loss": 0.3366,
+      "step": 2103
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6403353927625772e-06,
+      "loss": 1.659,
+      "step": 2104
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.639673433362754e-06,
+      "loss": 1.154,
+      "step": 2105
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6390114739629302e-06,
+      "loss": 2.7739,
+      "step": 2106
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6383495145631067e-06,
+      "loss": 0.7855,
+      "step": 2107
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6376875551632835e-06,
+      "loss": 2.274,
+      "step": 2108
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.63702559576346e-06,
+      "loss": 1.4983,
+      "step": 2109
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6363636363636363e-06,
+      "loss": 1.6673,
+      "step": 2110
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.635701676963813e-06,
+      "loss": 1.6842,
+      "step": 2111
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6350397175639895e-06,
+      "loss": 1.1946,
+      "step": 2112
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6343777581641662e-06,
+      "loss": 1.6557,
+      "step": 2113
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6337157987643425e-06,
+      "loss": 0.833,
+      "step": 2114
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.633053839364519e-06,
+      "loss": 0.9624,
+      "step": 2115
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6323918799646957e-06,
+      "loss": 0.5681,
+      "step": 2116
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.631729920564872e-06,
+      "loss": 0.0744,
+      "step": 2117
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6310679611650485e-06,
+      "loss": 1.2677,
+      "step": 2118
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6304060017652252e-06,
+      "loss": 2.4139,
+      "step": 2119
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6297440423654017e-06,
+      "loss": 0.4168,
+      "step": 2120
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.629082082965578e-06,
+      "loss": 0.7381,
+      "step": 2121
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6284201235657547e-06,
+      "loss": 1.6871,
+      "step": 2122
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6277581641659312e-06,
+      "loss": 1.2091,
+      "step": 2123
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.627096204766108e-06,
+      "loss": 0.8506,
+      "step": 2124
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6264342453662842e-06,
+      "loss": 1.2705,
+      "step": 2125
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6257722859664607e-06,
+      "loss": 0.8651,
+      "step": 2126
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6251103265666374e-06,
+      "loss": 0.465,
+      "step": 2127
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6244483671668137e-06,
+      "loss": 0.9001,
+      "step": 2128
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.6237864077669902e-06,
+      "loss": 1.5885,
+      "step": 2129
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.623124448367167e-06,
+      "loss": 0.4841,
+      "step": 2130
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6224624889673434e-06,
+      "loss": 1.9873,
+      "step": 2131
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6218005295675197e-06,
+      "loss": 0.4875,
+      "step": 2132
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6211385701676964e-06,
+      "loss": 1.9549,
+      "step": 2133
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6204766107678731e-06,
+      "loss": 1.2431,
+      "step": 2134
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6198146513680496e-06,
+      "loss": 1.2847,
+      "step": 2135
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.619152691968226e-06,
+      "loss": 0.7107,
+      "step": 2136
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6184907325684026e-06,
+      "loss": 1.2926,
+      "step": 2137
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6178287731685791e-06,
+      "loss": 0.7779,
+      "step": 2138
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6171668137687554e-06,
+      "loss": 0.6517,
+      "step": 2139
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6165048543689321e-06,
+      "loss": 1.1144,
+      "step": 2140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6158428949691086e-06,
+      "loss": 0.6124,
+      "step": 2141
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6151809355692853e-06,
+      "loss": 1.0781,
+      "step": 2142
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6145189761694616e-06,
+      "loss": 0.9622,
+      "step": 2143
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6138570167696381e-06,
+      "loss": 1.0795,
+      "step": 2144
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6131950573698148e-06,
+      "loss": 1.9302,
+      "step": 2145
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6125330979699913e-06,
+      "loss": 0.822,
+      "step": 2146
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6118711385701676e-06,
+      "loss": 0.5938,
+      "step": 2147
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6112091791703443e-06,
+      "loss": 0.4298,
+      "step": 2148
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6105472197705208e-06,
+      "loss": 0.2625,
+      "step": 2149
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6098852603706971e-06,
+      "loss": 0.4413,
+      "step": 2150
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6092233009708738e-06,
+      "loss": 1.7344,
+      "step": 2151
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6085613415710503e-06,
+      "loss": 2.5282,
+      "step": 2152
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.607899382171227e-06,
+      "loss": 1.1365,
+      "step": 2153
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6072374227714033e-06,
+      "loss": 3.2368,
+      "step": 2154
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6065754633715798e-06,
+      "loss": 0.8449,
+      "step": 2155
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6059135039717566e-06,
+      "loss": 0.876,
+      "step": 2156
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.605251544571933e-06,
+      "loss": 1.207,
+      "step": 2157
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6045895851721093e-06,
+      "loss": 0.5957,
+      "step": 2158
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.603927625772286e-06,
+      "loss": 1.0618,
+      "step": 2159
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6032656663724626e-06,
+      "loss": 0.3805,
+      "step": 2160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6026037069726388e-06,
+      "loss": 1.0333,
+      "step": 2161
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6019417475728156e-06,
+      "loss": 1.2136,
+      "step": 2162
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.601279788172992e-06,
+      "loss": 1.355,
+      "step": 2163
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.6006178287731688e-06,
+      "loss": 1.4987,
+      "step": 2164
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.599955869373345e-06,
+      "loss": 1.5354,
+      "step": 2165
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5992939099735216e-06,
+      "loss": 0.5459,
+      "step": 2166
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5986319505736983e-06,
+      "loss": 1.5024,
+      "step": 2167
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5979699911738746e-06,
+      "loss": 0.7831,
+      "step": 2168
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.597308031774051e-06,
+      "loss": 0.9432,
+      "step": 2169
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5966460723742278e-06,
+      "loss": 2.6329,
+      "step": 2170
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5959841129744045e-06,
+      "loss": 0.0921,
+      "step": 2171
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5953221535745808e-06,
+      "loss": 2.0696,
+      "step": 2172
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5946601941747573e-06,
+      "loss": 0.7829,
+      "step": 2173
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.593998234774934e-06,
+      "loss": 0.3198,
+      "step": 2174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5933362753751105e-06,
+      "loss": 1.5142,
+      "step": 2175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.5926743159752868e-06,
+      "loss": 1.4195,
+      "step": 2176
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5920123565754635e-06,
+      "loss": 0.2111,
+      "step": 2177
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.59135039717564e-06,
+      "loss": 0.0544,
+      "step": 2178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5906884377758163e-06,
+      "loss": 1.4272,
+      "step": 2179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.590026478375993e-06,
+      "loss": 1.3259,
+      "step": 2180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5893645189761695e-06,
+      "loss": 0.3289,
+      "step": 2181
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5887025595763462e-06,
+      "loss": 0.9591,
+      "step": 2182
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5880406001765225e-06,
+      "loss": 4.3381,
+      "step": 2183
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.587378640776699e-06,
+      "loss": 2.7429,
+      "step": 2184
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5867166813768757e-06,
+      "loss": 2.022,
+      "step": 2185
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5860547219770522e-06,
+      "loss": 0.4223,
+      "step": 2186
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5853927625772285e-06,
+      "loss": 1.3567,
+      "step": 2187
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5847308031774052e-06,
+      "loss": 0.8507,
+      "step": 2188
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5840688437775817e-06,
+      "loss": 1.2225,
+      "step": 2189
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.583406884377758e-06,
+      "loss": 0.4731,
+      "step": 2190
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5827449249779347e-06,
+      "loss": 2.2011,
+      "step": 2191
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5820829655781112e-06,
+      "loss": 0.0994,
+      "step": 2192
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.581421006178288e-06,
+      "loss": 1.0086,
+      "step": 2193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5807590467784642e-06,
+      "loss": 1.1221,
+      "step": 2194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5800970873786407e-06,
+      "loss": 1.0236,
+      "step": 2195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5794351279788174e-06,
+      "loss": 0.339,
+      "step": 2196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.578773168578994e-06,
+      "loss": 1.0803,
+      "step": 2197
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5781112091791702e-06,
+      "loss": 0.6449,
+      "step": 2198
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.577449249779347e-06,
+      "loss": 3.0228,
+      "step": 2199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5767872903795234e-06,
+      "loss": 0.8181,
+      "step": 2200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5761253309796997e-06,
+      "loss": 1.7251,
+      "step": 2201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5754633715798764e-06,
+      "loss": 2.0637,
+      "step": 2202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.574801412180053e-06,
+      "loss": 1.9249,
+      "step": 2203
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5741394527802296e-06,
+      "loss": 1.0595,
+      "step": 2204
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.573477493380406e-06,
+      "loss": 0.8976,
+      "step": 2205
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5728155339805824e-06,
+      "loss": 0.7703,
+      "step": 2206
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5721535745807591e-06,
+      "loss": 1.1347,
+      "step": 2207
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5714916151809359e-06,
+      "loss": 1.2203,
+      "step": 2208
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5708296557811121e-06,
+      "loss": 1.4388,
+      "step": 2209
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5701676963812886e-06,
+      "loss": 0.5058,
+      "step": 2210
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5695057369814654e-06,
+      "loss": 0.3485,
+      "step": 2211
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5688437775816416e-06,
+      "loss": 2.4127,
+      "step": 2212
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5681818181818181e-06,
+      "loss": 1.0316,
+      "step": 2213
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5675198587819949e-06,
+      "loss": 2.3199,
+      "step": 2214
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5668578993821714e-06,
+      "loss": 0.2015,
+      "step": 2215
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5661959399823476e-06,
+      "loss": 0.5843,
+      "step": 2216
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5655339805825244e-06,
+      "loss": 0.2493,
+      "step": 2217
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5648720211827009e-06,
+      "loss": 0.5659,
+      "step": 2218
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5642100617828776e-06,
+      "loss": 2.6566,
+      "step": 2219
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5635481023830539e-06,
+      "loss": 1.4337,
+      "step": 2220
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.5628861429832304e-06,
+      "loss": 3.5158,
+      "step": 2221
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.562224183583407e-06,
+      "loss": 4.1356,
+      "step": 2222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5615622241835834e-06,
+      "loss": 1.3203,
+      "step": 2223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5609002647837599e-06,
+      "loss": 0.1016,
+      "step": 2224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5602383053839366e-06,
+      "loss": 1.2635,
+      "step": 2225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.559576345984113e-06,
+      "loss": 1.3449,
+      "step": 2226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5589143865842894e-06,
+      "loss": 0.3344,
+      "step": 2227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.558252427184466e-06,
+      "loss": 1.6867,
+      "step": 2228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5575904677846426e-06,
+      "loss": 0.6759,
+      "step": 2229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5569285083848193e-06,
+      "loss": 0.823,
+      "step": 2230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5562665489849956e-06,
+      "loss": 1.2083,
+      "step": 2231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.555604589585172e-06,
+      "loss": 0.6695,
+      "step": 2232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5549426301853488e-06,
+      "loss": 0.5731,
+      "step": 2233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.554280670785525e-06,
+      "loss": 0.3751,
+      "step": 2234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5536187113857016e-06,
+      "loss": 0.4023,
+      "step": 2235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5529567519858783e-06,
+      "loss": 1.2034,
+      "step": 2236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5522947925860548e-06,
+      "loss": 0.9821,
+      "step": 2237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.551632833186231e-06,
+      "loss": 0.0728,
+      "step": 2238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5509708737864078e-06,
+      "loss": 0.0979,
+      "step": 2239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5503089143865843e-06,
+      "loss": 2.1052,
+      "step": 2240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.549646954986761e-06,
+      "loss": 0.685,
+      "step": 2241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5489849955869373e-06,
+      "loss": 0.282,
+      "step": 2242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.548323036187114e-06,
+      "loss": 1.8607,
+      "step": 2243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5476610767872905e-06,
+      "loss": 1.5221,
+      "step": 2244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5469991173874668e-06,
+      "loss": 0.3702,
+      "step": 2245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5463371579876435e-06,
+      "loss": 0.8006,
+      "step": 2246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.54567519858782e-06,
+      "loss": 1.5549,
+      "step": 2247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5450132391879967e-06,
+      "loss": 1.9436,
+      "step": 2248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.544351279788173e-06,
+      "loss": 0.959,
+      "step": 2249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5436893203883495e-06,
+      "loss": 0.7561,
+      "step": 2250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5430273609885262e-06,
+      "loss": 1.1125,
+      "step": 2251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5423654015887027e-06,
+      "loss": 0.2056,
+      "step": 2252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.541703442188879e-06,
+      "loss": 0.8232,
+      "step": 2253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5410414827890557e-06,
+      "loss": 0.9762,
+      "step": 2254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5403795233892322e-06,
+      "loss": 3.3306,
+      "step": 2255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5397175639894085e-06,
+      "loss": 0.9101,
+      "step": 2256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5390556045895852e-06,
+      "loss": 2.9342,
+      "step": 2257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5383936451897617e-06,
+      "loss": 0.1275,
+      "step": 2258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5377316857899384e-06,
+      "loss": 0.3725,
+      "step": 2259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5370697263901147e-06,
+      "loss": 1.6182,
+      "step": 2260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5364077669902912e-06,
+      "loss": 1.4866,
+      "step": 2261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.535745807590468e-06,
+      "loss": 1.3296,
+      "step": 2262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5350838481906444e-06,
+      "loss": 1.0647,
+      "step": 2263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5344218887908207e-06,
+      "loss": 2.6411,
+      "step": 2264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5337599293909974e-06,
+      "loss": 1.0138,
+      "step": 2265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.533097969991174e-06,
+      "loss": 3.0696,
+      "step": 2266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.5324360105913502e-06,
+      "loss": 0.9242,
+      "step": 2267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.531774051191527e-06,
+      "loss": 1.3484,
+      "step": 2268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5311120917917034e-06,
+      "loss": 2.3076,
+      "step": 2269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5304501323918802e-06,
+      "loss": 0.5519,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5297881729920565e-06,
+      "loss": 0.7366,
+      "step": 2271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.529126213592233e-06,
+      "loss": 2.7651,
+      "step": 2272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5284642541924097e-06,
+      "loss": 0.4293,
+      "step": 2273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.527802294792586e-06,
+      "loss": 0.0371,
+      "step": 2274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5271403353927625e-06,
+      "loss": 1.9877,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5264783759929392e-06,
+      "loss": 1.1782,
+      "step": 2276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5258164165931157e-06,
+      "loss": 0.7743,
+      "step": 2277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.525154457193292e-06,
+      "loss": 0.9661,
+      "step": 2278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5244924977934687e-06,
+      "loss": 1.6221,
+      "step": 2279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5238305383936454e-06,
+      "loss": 1.221,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5231685789938219e-06,
+      "loss": 0.3016,
+      "step": 2281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5225066195939982e-06,
+      "loss": 0.5942,
+      "step": 2282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5218446601941749e-06,
+      "loss": 1.2691,
+      "step": 2283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5211827007943514e-06,
+      "loss": 1.4721,
+      "step": 2284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5205207413945277e-06,
+      "loss": 0.1968,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5198587819947044e-06,
+      "loss": 0.5282,
+      "step": 2286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5191968225948809e-06,
+      "loss": 0.183,
+      "step": 2287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5185348631950576e-06,
+      "loss": 1.5963,
+      "step": 2288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5178729037952339e-06,
+      "loss": 0.2455,
+      "step": 2289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5172109443954104e-06,
+      "loss": 0.8718,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.516548984995587e-06,
+      "loss": 0.6574,
+      "step": 2291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5158870255957636e-06,
+      "loss": 1.9657,
+      "step": 2292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5152250661959399e-06,
+      "loss": 1.5187,
+      "step": 2293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5145631067961166e-06,
+      "loss": 0.6718,
+      "step": 2294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.513901147396293e-06,
+      "loss": 2.5879,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5132391879964694e-06,
+      "loss": 0.7032,
+      "step": 2296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.512577228596646e-06,
+      "loss": 0.3299,
+      "step": 2297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5119152691968226e-06,
+      "loss": 1.2031,
+      "step": 2298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5112533097969993e-06,
+      "loss": 1.1789,
+      "step": 2299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5105913503971756e-06,
+      "loss": 0.2257,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.509929390997352e-06,
+      "loss": 0.6586,
+      "step": 2301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5092674315975288e-06,
+      "loss": 2.1567,
+      "step": 2302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5086054721977053e-06,
+      "loss": 2.6329,
+      "step": 2303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5079435127978816e-06,
+      "loss": 1.6384,
+      "step": 2304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5072815533980583e-06,
+      "loss": 0.5184,
+      "step": 2305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5066195939982348e-06,
+      "loss": 0.7622,
+      "step": 2306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5059576345984111e-06,
+      "loss": 1.4664,
+      "step": 2307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5052956751985878e-06,
+      "loss": 1.9732,
+      "step": 2308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5046337157987643e-06,
+      "loss": 0.9208,
+      "step": 2309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.503971756398941e-06,
+      "loss": 1.7535,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5033097969991173e-06,
+      "loss": 2.4425,
+      "step": 2311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5026478375992938e-06,
+      "loss": 0.5617,
+      "step": 2312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.5019858781994705e-06,
+      "loss": 0.1059,
+      "step": 2313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5013239187996472e-06,
+      "loss": 0.3134,
+      "step": 2314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5006619593998233e-06,
+      "loss": 1.5896,
+      "step": 2315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.5e-06,
+      "loss": 0.7194,
+      "step": 2316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4993380406001765e-06,
+      "loss": 0.3134,
+      "step": 2317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.498676081200353e-06,
+      "loss": 1.5672,
+      "step": 2318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4980141218005295e-06,
+      "loss": 0.9398,
+      "step": 2319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4973521624007063e-06,
+      "loss": 0.6678,
+      "step": 2320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4966902030008825e-06,
+      "loss": 1.9461,
+      "step": 2321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4960282436010593e-06,
+      "loss": 1.3127,
+      "step": 2322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4953662842012358e-06,
+      "loss": 0.9401,
+      "step": 2323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4947043248014123e-06,
+      "loss": 0.3529,
+      "step": 2324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4940423654015888e-06,
+      "loss": 2.0538,
+      "step": 2325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4933804060017653e-06,
+      "loss": 2.2762,
+      "step": 2326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4927184466019418e-06,
+      "loss": 1.0541,
+      "step": 2327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4920564872021183e-06,
+      "loss": 1.9109,
+      "step": 2328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4913945278022948e-06,
+      "loss": 2.0226,
+      "step": 2329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4907325684024713e-06,
+      "loss": 3.4395,
+      "step": 2330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.490070609002648e-06,
+      "loss": 0.3873,
+      "step": 2331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4894086496028243e-06,
+      "loss": 1.6199,
+      "step": 2332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.488746690203001e-06,
+      "loss": 1.7775,
+      "step": 2333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4880847308031775e-06,
+      "loss": 0.9889,
+      "step": 2334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.487422771403354e-06,
+      "loss": 0.2409,
+      "step": 2335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4867608120035305e-06,
+      "loss": 0.077,
+      "step": 2336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4860988526037072e-06,
+      "loss": 1.6504,
+      "step": 2337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4854368932038835e-06,
+      "loss": 1.1567,
+      "step": 2338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.48477493380406e-06,
+      "loss": 2.6966,
+      "step": 2339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4841129744042367e-06,
+      "loss": 0.2485,
+      "step": 2340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.483451015004413e-06,
+      "loss": 0.8671,
+      "step": 2341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4827890556045897e-06,
+      "loss": 1.6778,
+      "step": 2342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4821270962047662e-06,
+      "loss": 0.7073,
+      "step": 2343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4814651368049427e-06,
+      "loss": 1.3924,
+      "step": 2344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4808031774051192e-06,
+      "loss": 0.1408,
+      "step": 2345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4801412180052957e-06,
+      "loss": 0.6079,
+      "step": 2346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4794792586054722e-06,
+      "loss": 1.8016,
+      "step": 2347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4788172992056487e-06,
+      "loss": 1.6995,
+      "step": 2348
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4781553398058252e-06,
+      "loss": 0.8309,
+      "step": 2349
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4774933804060017e-06,
+      "loss": 1.1294,
+      "step": 2350
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4768314210061784e-06,
+      "loss": 0.5003,
+      "step": 2351
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4761694616063547e-06,
+      "loss": 0.1955,
+      "step": 2352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4755075022065314e-06,
+      "loss": 0.1598,
+      "step": 2353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.474845542806708e-06,
+      "loss": 2.155,
+      "step": 2354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4741835834068844e-06,
+      "loss": 1.0628,
+      "step": 2355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.473521624007061e-06,
+      "loss": 0.2785,
+      "step": 2356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4728596646072376e-06,
+      "loss": 0.9939,
+      "step": 2357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.472197705207414e-06,
+      "loss": 0.5191,
+      "step": 2358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.4715357458075904e-06,
+      "loss": 0.9805,
+      "step": 2359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4708737864077671e-06,
+      "loss": 0.0483,
+      "step": 2360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4702118270079434e-06,
+      "loss": 0.449,
+      "step": 2361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4695498676081201e-06,
+      "loss": 1.9365,
+      "step": 2362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4688879082082966e-06,
+      "loss": 0.1206,
+      "step": 2363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4682259488084731e-06,
+      "loss": 0.6281,
+      "step": 2364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4675639894086496e-06,
+      "loss": 1.3186,
+      "step": 2365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4669020300088261e-06,
+      "loss": 0.0844,
+      "step": 2366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4662400706090026e-06,
+      "loss": 2.9354,
+      "step": 2367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4655781112091793e-06,
+      "loss": 1.7099,
+      "step": 2368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4649161518093556e-06,
+      "loss": 1.1024,
+      "step": 2369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4642541924095321e-06,
+      "loss": 1.331,
+      "step": 2370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4635922330097088e-06,
+      "loss": 0.2381,
+      "step": 2371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4629302736098853e-06,
+      "loss": 1.2181,
+      "step": 2372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4622683142100618e-06,
+      "loss": 0.6457,
+      "step": 2373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4616063548102383e-06,
+      "loss": 0.1393,
+      "step": 2374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4609443954104148e-06,
+      "loss": 0.5647,
+      "step": 2375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4602824360105913e-06,
+      "loss": 0.3458,
+      "step": 2376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.459620476610768e-06,
+      "loss": 1.1028,
+      "step": 2377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4589585172109443e-06,
+      "loss": 2.2304,
+      "step": 2378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.458296557811121e-06,
+      "loss": 1.7984,
+      "step": 2379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4576345984112976e-06,
+      "loss": 0.3818,
+      "step": 2380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4569726390114738e-06,
+      "loss": 0.0421,
+      "step": 2381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4563106796116506e-06,
+      "loss": 2.1,
+      "step": 2382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.455648720211827e-06,
+      "loss": 0.886,
+      "step": 2383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4549867608120036e-06,
+      "loss": 0.931,
+      "step": 2384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.45432480141218e-06,
+      "loss": 0.6149,
+      "step": 2385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4536628420123566e-06,
+      "loss": 0.4154,
+      "step": 2386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.453000882612533e-06,
+      "loss": 1.3699,
+      "step": 2387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4523389232127098e-06,
+      "loss": 2.1501,
+      "step": 2388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.451676963812886e-06,
+      "loss": 1.3033,
+      "step": 2389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4510150044130628e-06,
+      "loss": 2.9512,
+      "step": 2390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4503530450132393e-06,
+      "loss": 0.3813,
+      "step": 2391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4496910856134158e-06,
+      "loss": 2.3304,
+      "step": 2392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4490291262135923e-06,
+      "loss": 1.9611,
+      "step": 2393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4483671668137688e-06,
+      "loss": 0.4983,
+      "step": 2394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4477052074139453e-06,
+      "loss": 0.3366,
+      "step": 2395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4470432480141218e-06,
+      "loss": 0.3977,
+      "step": 2396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4463812886142985e-06,
+      "loss": 1.2348,
+      "step": 2397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4457193292144748e-06,
+      "loss": 0.1616,
+      "step": 2398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4450573698146515e-06,
+      "loss": 0.8582,
+      "step": 2399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.444395410414828e-06,
+      "loss": 1.1539,
+      "step": 2400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4437334510150043e-06,
+      "loss": 0.1219,
+      "step": 2401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.443071491615181e-06,
+      "loss": 1.7691,
+      "step": 2402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4424095322153575e-06,
+      "loss": 0.4189,
+      "step": 2403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.441747572815534e-06,
+      "loss": 1.2555,
+      "step": 2404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.4410856134157105e-06,
+      "loss": 0.3024,
+      "step": 2405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.440423654015887e-06,
+      "loss": 0.5491,
+      "step": 2406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4397616946160635e-06,
+      "loss": 1.2116,
+      "step": 2407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4390997352162402e-06,
+      "loss": 1.5516,
+      "step": 2408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4384377758164167e-06,
+      "loss": 0.0613,
+      "step": 2409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4377758164165932e-06,
+      "loss": 0.6675,
+      "step": 2410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4371138570167697e-06,
+      "loss": 0.3965,
+      "step": 2411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4364518976169462e-06,
+      "loss": 1.1343,
+      "step": 2412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4357899382171227e-06,
+      "loss": 0.2887,
+      "step": 2413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4351279788172992e-06,
+      "loss": 1.7875,
+      "step": 2414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4344660194174757e-06,
+      "loss": 0.9417,
+      "step": 2415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4338040600176522e-06,
+      "loss": 1.0175,
+      "step": 2416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.433142100617829e-06,
+      "loss": 0.4649,
+      "step": 2417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4324801412180052e-06,
+      "loss": 0.2326,
+      "step": 2418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.431818181818182e-06,
+      "loss": 1.3123,
+      "step": 2419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4311562224183584e-06,
+      "loss": 0.9312,
+      "step": 2420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.430494263018535e-06,
+      "loss": 1.5257,
+      "step": 2421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4298323036187114e-06,
+      "loss": 1.7505,
+      "step": 2422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.429170344218888e-06,
+      "loss": 0.1726,
+      "step": 2423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4285083848190644e-06,
+      "loss": 1.1435,
+      "step": 2424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.427846425419241e-06,
+      "loss": 0.9882,
+      "step": 2425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4271844660194176e-06,
+      "loss": 0.8836,
+      "step": 2426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.426522506619594e-06,
+      "loss": 2.8149,
+      "step": 2427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4258605472197706e-06,
+      "loss": 0.7033,
+      "step": 2428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4251985878199471e-06,
+      "loss": 0.3811,
+      "step": 2429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4245366284201237e-06,
+      "loss": 1.0278,
+      "step": 2430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4238746690203002e-06,
+      "loss": 0.2537,
+      "step": 2431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4232127096204767e-06,
+      "loss": 2.0956,
+      "step": 2432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4225507502206532e-06,
+      "loss": 1.6407,
+      "step": 2433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4218887908208297e-06,
+      "loss": 1.4841,
+      "step": 2434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4212268314210062e-06,
+      "loss": 0.6686,
+      "step": 2435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4205648720211827e-06,
+      "loss": 1.3661,
+      "step": 2436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4199029126213594e-06,
+      "loss": 1.4358,
+      "step": 2437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4192409532215357e-06,
+      "loss": 0.5202,
+      "step": 2438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4185789938217124e-06,
+      "loss": 0.3543,
+      "step": 2439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4179170344218889e-06,
+      "loss": 1.2724,
+      "step": 2440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4172550750220654e-06,
+      "loss": 2.894,
+      "step": 2441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4165931156222419e-06,
+      "loss": 0.7934,
+      "step": 2442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4159311562224186e-06,
+      "loss": 0.6595,
+      "step": 2443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4152691968225949e-06,
+      "loss": 0.9666,
+      "step": 2444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4146072374227714e-06,
+      "loss": 0.5984,
+      "step": 2445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.413945278022948e-06,
+      "loss": 0.3847,
+      "step": 2446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4132833186231244e-06,
+      "loss": 1.4458,
+      "step": 2447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.412621359223301e-06,
+      "loss": 0.5254,
+      "step": 2448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4119593998234776e-06,
+      "loss": 1.0904,
+      "step": 2449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.411297440423654e-06,
+      "loss": 2.2039,
+      "step": 2450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.4106354810238306e-06,
+      "loss": 1.7948,
+      "step": 2451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.409973521624007e-06,
+      "loss": 2.3353,
+      "step": 2452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4093115622241836e-06,
+      "loss": 0.572,
+      "step": 2453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.40864960282436e-06,
+      "loss": 3.0678,
+      "step": 2454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4079876434245366e-06,
+      "loss": 0.7547,
+      "step": 2455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.407325684024713e-06,
+      "loss": 1.4336,
+      "step": 2456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4066637246248898e-06,
+      "loss": 1.0432,
+      "step": 2457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.406001765225066e-06,
+      "loss": 2.3977,
+      "step": 2458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4053398058252428e-06,
+      "loss": 1.0156,
+      "step": 2459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4046778464254193e-06,
+      "loss": 0.8261,
+      "step": 2460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4040158870255958e-06,
+      "loss": 0.7411,
+      "step": 2461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4033539276257723e-06,
+      "loss": 1.8697,
+      "step": 2462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.402691968225949e-06,
+      "loss": 0.7355,
+      "step": 2463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4020300088261253e-06,
+      "loss": 1.8944,
+      "step": 2464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4013680494263018e-06,
+      "loss": 0.1146,
+      "step": 2465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4007060900264785e-06,
+      "loss": 0.6041,
+      "step": 2466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.4000441306266548e-06,
+      "loss": 1.2624,
+      "step": 2467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3993821712268315e-06,
+      "loss": 0.4584,
+      "step": 2468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.398720211827008e-06,
+      "loss": 1.3783,
+      "step": 2469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3980582524271845e-06,
+      "loss": 0.325,
+      "step": 2470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.397396293027361e-06,
+      "loss": 1.6753,
+      "step": 2471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3967343336275375e-06,
+      "loss": 0.8218,
+      "step": 2472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.396072374227714e-06,
+      "loss": 0.0793,
+      "step": 2473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3954104148278907e-06,
+      "loss": 2.5946,
+      "step": 2474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.394748455428067e-06,
+      "loss": 0.9945,
+      "step": 2475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3940864960282435e-06,
+      "loss": 1.1682,
+      "step": 2476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3934245366284202e-06,
+      "loss": 2.6234,
+      "step": 2477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3927625772285965e-06,
+      "loss": 1.8794,
+      "step": 2478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3921006178287732e-06,
+      "loss": 2.7923,
+      "step": 2479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3914386584289497e-06,
+      "loss": 2.3472,
+      "step": 2480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3907766990291262e-06,
+      "loss": 0.5249,
+      "step": 2481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3901147396293027e-06,
+      "loss": 0.4732,
+      "step": 2482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3894527802294795e-06,
+      "loss": 0.6603,
+      "step": 2483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3887908208296557e-06,
+      "loss": 1.1192,
+      "step": 2484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3881288614298325e-06,
+      "loss": 1.2703,
+      "step": 2485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.387466902030009e-06,
+      "loss": 1.6816,
+      "step": 2486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3868049426301852e-06,
+      "loss": 0.754,
+      "step": 2487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.386142983230362e-06,
+      "loss": 1.7847,
+      "step": 2488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3854810238305385e-06,
+      "loss": 0.4841,
+      "step": 2489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.384819064430715e-06,
+      "loss": 1.4961,
+      "step": 2490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3841571050308915e-06,
+      "loss": 1.0028,
+      "step": 2491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.383495145631068e-06,
+      "loss": 1.2598,
+      "step": 2492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3828331862312445e-06,
+      "loss": 0.7303,
+      "step": 2493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3821712268314212e-06,
+      "loss": 0.7583,
+      "step": 2494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3815092674315975e-06,
+      "loss": 0.5085,
+      "step": 2495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3808473080317742e-06,
+      "loss": 1.1501,
+      "step": 2496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.3801853486319507e-06,
+      "loss": 0.0512,
+      "step": 2497
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.379523389232127e-06,
+      "loss": 1.0839,
+      "step": 2498
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3788614298323037e-06,
+      "loss": 0.5707,
+      "step": 2499
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3781994704324802e-06,
+      "loss": 1.7744,
+      "step": 2500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3775375110326567e-06,
+      "loss": 2.709,
+      "step": 2501
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3768755516328332e-06,
+      "loss": 0.4942,
+      "step": 2502
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3762135922330099e-06,
+      "loss": 0.1716,
+      "step": 2503
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3755516328331862e-06,
+      "loss": 0.0721,
+      "step": 2504
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3748896734333629e-06,
+      "loss": 0.5978,
+      "step": 2505
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3742277140335394e-06,
+      "loss": 2.2435,
+      "step": 2506
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3735657546337157e-06,
+      "loss": 1.4703,
+      "step": 2507
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3729037952338924e-06,
+      "loss": 0.7667,
+      "step": 2508
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3722418358340689e-06,
+      "loss": 1.8934,
+      "step": 2509
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3715798764342454e-06,
+      "loss": 1.1045,
+      "step": 2510
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.370917917034422e-06,
+      "loss": 0.4048,
+      "step": 2511
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3702559576345984e-06,
+      "loss": 0.6188,
+      "step": 2512
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.369593998234775e-06,
+      "loss": 0.4526,
+      "step": 2513
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3689320388349516e-06,
+      "loss": 2.7534,
+      "step": 2514
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.368270079435128e-06,
+      "loss": 0.6709,
+      "step": 2515
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3676081200353046e-06,
+      "loss": 1.6379,
+      "step": 2516
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3669461606354811e-06,
+      "loss": 1.0304,
+      "step": 2517
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3662842012356574e-06,
+      "loss": 0.5644,
+      "step": 2518
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3656222418358341e-06,
+      "loss": 0.5412,
+      "step": 2519
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3649602824360106e-06,
+      "loss": 0.7479,
+      "step": 2520
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3642983230361871e-06,
+      "loss": 0.6782,
+      "step": 2521
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3636363636363636e-06,
+      "loss": 0.0338,
+      "step": 2522
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3629744042365403e-06,
+      "loss": 0.5502,
+      "step": 2523
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3623124448367166e-06,
+      "loss": 0.9221,
+      "step": 2524
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3616504854368933e-06,
+      "loss": 1.1312,
+      "step": 2525
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3609885260370698e-06,
+      "loss": 0.5047,
+      "step": 2526
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3603265666372463e-06,
+      "loss": 0.7314,
+      "step": 2527
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3596646072374228e-06,
+      "loss": 0.6265,
+      "step": 2528
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3590026478375993e-06,
+      "loss": 2.3014,
+      "step": 2529
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3583406884377758e-06,
+      "loss": 0.2163,
+      "step": 2530
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3576787290379523e-06,
+      "loss": 0.314,
+      "step": 2531
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3570167696381288e-06,
+      "loss": 0.98,
+      "step": 2532
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3563548102383053e-06,
+      "loss": 0.1216,
+      "step": 2533
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.355692850838482e-06,
+      "loss": 0.5582,
+      "step": 2534
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3550308914386583e-06,
+      "loss": 2.3471,
+      "step": 2535
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.354368932038835e-06,
+      "loss": 0.5945,
+      "step": 2536
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3537069726390115e-06,
+      "loss": 0.4679,
+      "step": 2537
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.353045013239188e-06,
+      "loss": 1.4139,
+      "step": 2538
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3523830538393645e-06,
+      "loss": 0.7471,
+      "step": 2539
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.351721094439541e-06,
+      "loss": 0.0657,
+      "step": 2540
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3510591350397175e-06,
+      "loss": 0.9656,
+      "step": 2541
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.350397175639894e-06,
+      "loss": 1.2669,
+      "step": 2542
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.3497352162400708e-06,
+      "loss": 1.4021,
+      "step": 2543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.349073256840247e-06,
+      "loss": 0.1684,
+      "step": 2544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3484112974404238e-06,
+      "loss": 0.4574,
+      "step": 2545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3477493380406003e-06,
+      "loss": 0.6849,
+      "step": 2546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3470873786407768e-06,
+      "loss": 0.8593,
+      "step": 2547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3464254192409533e-06,
+      "loss": 0.2891,
+      "step": 2548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3457634598411298e-06,
+      "loss": 1.8866,
+      "step": 2549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3451015004413063e-06,
+      "loss": 0.3301,
+      "step": 2550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3444395410414828e-06,
+      "loss": 2.6454,
+      "step": 2551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3437775816416593e-06,
+      "loss": 1.5468,
+      "step": 2552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3431156222418358e-06,
+      "loss": 0.024,
+      "step": 2553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3424536628420125e-06,
+      "loss": 2.4356,
+      "step": 2554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.341791703442189e-06,
+      "loss": 0.1167,
+      "step": 2555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3411297440423655e-06,
+      "loss": 2.4171,
+      "step": 2556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.340467784642542e-06,
+      "loss": 0.6979,
+      "step": 2557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3398058252427185e-06,
+      "loss": 1.0877,
+      "step": 2558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.339143865842895e-06,
+      "loss": 0.5235,
+      "step": 2559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3384819064430715e-06,
+      "loss": 0.8317,
+      "step": 2560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.337819947043248e-06,
+      "loss": 0.6859,
+      "step": 2561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3371579876434245e-06,
+      "loss": 1.535,
+      "step": 2562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3364960282436012e-06,
+      "loss": 1.7272,
+      "step": 2563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3358340688437775e-06,
+      "loss": 0.8676,
+      "step": 2564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3351721094439542e-06,
+      "loss": 1.6413,
+      "step": 2565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3345101500441307e-06,
+      "loss": 1.1518,
+      "step": 2566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3338481906443072e-06,
+      "loss": 0.2672,
+      "step": 2567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3331862312444837e-06,
+      "loss": 1.6086,
+      "step": 2568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3325242718446602e-06,
+      "loss": 2.2763,
+      "step": 2569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3318623124448367e-06,
+      "loss": 2.2217,
+      "step": 2570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3312003530450132e-06,
+      "loss": 2.0429,
+      "step": 2571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.33053839364519e-06,
+      "loss": 1.1756,
+      "step": 2572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3298764342453662e-06,
+      "loss": 0.5992,
+      "step": 2573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.329214474845543e-06,
+      "loss": 0.7262,
+      "step": 2574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3285525154457194e-06,
+      "loss": 2.1599,
+      "step": 2575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.327890556045896e-06,
+      "loss": 0.2261,
+      "step": 2576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3272285966460724e-06,
+      "loss": 3.1659,
+      "step": 2577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.326566637246249e-06,
+      "loss": 1.3604,
+      "step": 2578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3259046778464254e-06,
+      "loss": 0.0724,
+      "step": 2579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3252427184466021e-06,
+      "loss": 1.1294,
+      "step": 2580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3245807590467784e-06,
+      "loss": 1.0802,
+      "step": 2581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.323918799646955e-06,
+      "loss": 3.9138,
+      "step": 2582
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3232568402471316e-06,
+      "loss": 0.9305,
+      "step": 2583
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.322594880847308e-06,
+      "loss": 0.8022,
+      "step": 2584
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3219329214474846e-06,
+      "loss": 0.1759,
+      "step": 2585
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3212709620476611e-06,
+      "loss": 1.3898,
+      "step": 2586
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3206090026478376e-06,
+      "loss": 0.8344,
+      "step": 2587
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.3199470432480141e-06,
+      "loss": 1.902,
+      "step": 2588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3192850838481906e-06,
+      "loss": 0.3008,
+      "step": 2589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3186231244483671e-06,
+      "loss": 0.5563,
+      "step": 2590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3179611650485439e-06,
+      "loss": 0.3985,
+      "step": 2591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3172992056487204e-06,
+      "loss": 1.3544,
+      "step": 2592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3166372462488966e-06,
+      "loss": 1.0264,
+      "step": 2593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3159752868490734e-06,
+      "loss": 1.4166,
+      "step": 2594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3153133274492499e-06,
+      "loss": 1.6042,
+      "step": 2595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3146513680494264e-06,
+      "loss": 2.9583,
+      "step": 2596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3139894086496029e-06,
+      "loss": 0.5811,
+      "step": 2597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3133274492497794e-06,
+      "loss": 0.4267,
+      "step": 2598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3126654898499559e-06,
+      "loss": 0.6321,
+      "step": 2599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3120035304501326e-06,
+      "loss": 1.4177,
+      "step": 2600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3113415710503089e-06,
+      "loss": 3.0235,
+      "step": 2601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3106796116504854e-06,
+      "loss": 1.6129,
+      "step": 2602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.310017652250662e-06,
+      "loss": 0.1088,
+      "step": 2603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3093556928508384e-06,
+      "loss": 1.3457,
+      "step": 2604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.308693733451015e-06,
+      "loss": 0.0868,
+      "step": 2605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3080317740511916e-06,
+      "loss": 1.8316,
+      "step": 2606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.307369814651368e-06,
+      "loss": 0.748,
+      "step": 2607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3067078552515446e-06,
+      "loss": 1.1454,
+      "step": 2608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3060458958517213e-06,
+      "loss": 1.2295,
+      "step": 2609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3053839364518976e-06,
+      "loss": 2.276,
+      "step": 2610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3047219770520743e-06,
+      "loss": 0.4329,
+      "step": 2611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3040600176522508e-06,
+      "loss": 0.2707,
+      "step": 2612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.303398058252427e-06,
+      "loss": 1.7076,
+      "step": 2613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3027360988526038e-06,
+      "loss": 0.8498,
+      "step": 2614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3020741394527803e-06,
+      "loss": 0.2797,
+      "step": 2615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3014121800529568e-06,
+      "loss": 3.4325,
+      "step": 2616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3007502206531333e-06,
+      "loss": 1.4865,
+      "step": 2617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.3000882612533098e-06,
+      "loss": 1.6052,
+      "step": 2618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2994263018534863e-06,
+      "loss": 1.1086,
+      "step": 2619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.298764342453663e-06,
+      "loss": 1.2892,
+      "step": 2620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2981023830538393e-06,
+      "loss": 1.8037,
+      "step": 2621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.297440423654016e-06,
+      "loss": 0.4149,
+      "step": 2622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2967784642541925e-06,
+      "loss": 0.7041,
+      "step": 2623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2961165048543688e-06,
+      "loss": 1.6528,
+      "step": 2624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2954545454545455e-06,
+      "loss": 0.3605,
+      "step": 2625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.294792586054722e-06,
+      "loss": 0.0907,
+      "step": 2626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2941306266548985e-06,
+      "loss": 1.2659,
+      "step": 2627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.293468667255075e-06,
+      "loss": 1.4552,
+      "step": 2628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2928067078552517e-06,
+      "loss": 0.4167,
+      "step": 2629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.292144748455428e-06,
+      "loss": 0.7469,
+      "step": 2630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2914827890556047e-06,
+      "loss": 1.6082,
+      "step": 2631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2908208296557812e-06,
+      "loss": 3.8337,
+      "step": 2632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2901588702559577e-06,
+      "loss": 0.5448,
+      "step": 2633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.2894969108561342e-06,
+      "loss": 1.3916,
+      "step": 2634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2888349514563107e-06,
+      "loss": 0.6745,
+      "step": 2635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2881729920564872e-06,
+      "loss": 2.1176,
+      "step": 2636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2875110326566637e-06,
+      "loss": 0.3865,
+      "step": 2637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2868490732568402e-06,
+      "loss": 1.2613,
+      "step": 2638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2861871138570167e-06,
+      "loss": 0.3062,
+      "step": 2639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2855251544571934e-06,
+      "loss": 0.9087,
+      "step": 2640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2848631950573697e-06,
+      "loss": 0.864,
+      "step": 2641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2842012356575464e-06,
+      "loss": 0.7786,
+      "step": 2642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.283539276257723e-06,
+      "loss": 1.9636,
+      "step": 2643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2828773168578994e-06,
+      "loss": 1.4992,
+      "step": 2644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.282215357458076e-06,
+      "loss": 1.3259,
+      "step": 2645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2815533980582524e-06,
+      "loss": 0.1375,
+      "step": 2646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.280891438658429e-06,
+      "loss": 2.9085,
+      "step": 2647
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2802294792586054e-06,
+      "loss": 2.7038,
+      "step": 2648
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2795675198587822e-06,
+      "loss": 1.4731,
+      "step": 2649
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2789055604589584e-06,
+      "loss": 1.061,
+      "step": 2650
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2782436010591352e-06,
+      "loss": 1.3977,
+      "step": 2651
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2775816416593117e-06,
+      "loss": 1.1386,
+      "step": 2652
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2769196822594882e-06,
+      "loss": 0.6628,
+      "step": 2653
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2762577228596647e-06,
+      "loss": 0.2974,
+      "step": 2654
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2755957634598412e-06,
+      "loss": 0.8792,
+      "step": 2655
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2749338040600177e-06,
+      "loss": 0.4347,
+      "step": 2656
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2742718446601942e-06,
+      "loss": 0.5018,
+      "step": 2657
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2736098852603707e-06,
+      "loss": 1.6179,
+      "step": 2658
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2729479258605472e-06,
+      "loss": 1.1807,
+      "step": 2659
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2722859664607239e-06,
+      "loss": 0.5841,
+      "step": 2660
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2716240070609002e-06,
+      "loss": 1.0301,
+      "step": 2661
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2709620476610769e-06,
+      "loss": 0.737,
+      "step": 2662
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2703000882612534e-06,
+      "loss": 1.3128,
+      "step": 2663
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2696381288614299e-06,
+      "loss": 0.8089,
+      "step": 2664
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2689761694616064e-06,
+      "loss": 2.117,
+      "step": 2665
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2683142100617829e-06,
+      "loss": 1.0667,
+      "step": 2666
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2676522506619594e-06,
+      "loss": 0.5023,
+      "step": 2667
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2669902912621359e-06,
+      "loss": 0.9911,
+      "step": 2668
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2663283318623126e-06,
+      "loss": 1.1612,
+      "step": 2669
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2656663724624889e-06,
+      "loss": 1.9523,
+      "step": 2670
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2650044130626656e-06,
+      "loss": 1.7694,
+      "step": 2671
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.264342453662842e-06,
+      "loss": 0.9692,
+      "step": 2672
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2636804942630186e-06,
+      "loss": 0.6885,
+      "step": 2673
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.263018534863195e-06,
+      "loss": 0.8205,
+      "step": 2674
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2623565754633716e-06,
+      "loss": 1.8808,
+      "step": 2675
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.261694616063548e-06,
+      "loss": 1.3992,
+      "step": 2676
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2610326566637246e-06,
+      "loss": 1.1102,
+      "step": 2677
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.260370697263901e-06,
+      "loss": 1.7506,
+      "step": 2678
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2597087378640776e-06,
+      "loss": 0.1288,
+      "step": 2679
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.2590467784642543e-06,
+      "loss": 1.6459,
+      "step": 2680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2583848190644306e-06,
+      "loss": 1.4507,
+      "step": 2681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2577228596646073e-06,
+      "loss": 0.5347,
+      "step": 2682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2570609002647838e-06,
+      "loss": 1.1892,
+      "step": 2683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2563989408649603e-06,
+      "loss": 0.9296,
+      "step": 2684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2557369814651368e-06,
+      "loss": 0.0936,
+      "step": 2685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2550750220653135e-06,
+      "loss": 3.3554,
+      "step": 2686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2544130626654898e-06,
+      "loss": 0.243,
+      "step": 2687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2537511032656663e-06,
+      "loss": 2.3801,
+      "step": 2688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.253089143865843e-06,
+      "loss": 2.4848,
+      "step": 2689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2524271844660193e-06,
+      "loss": 1.145,
+      "step": 2690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.251765225066196e-06,
+      "loss": 0.8211,
+      "step": 2691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2511032656663725e-06,
+      "loss": 0.3201,
+      "step": 2692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.250441306266549e-06,
+      "loss": 1.1964,
+      "step": 2693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2497793468667255e-06,
+      "loss": 0.4226,
+      "step": 2694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.249117387466902e-06,
+      "loss": 1.4421,
+      "step": 2695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2484554280670785e-06,
+      "loss": 1.1376,
+      "step": 2696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2477934686672552e-06,
+      "loss": 0.8248,
+      "step": 2697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2471315092674315e-06,
+      "loss": 0.6289,
+      "step": 2698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.246469549867608e-06,
+      "loss": 2.7298,
+      "step": 2699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2458075904677847e-06,
+      "loss": 1.005,
+      "step": 2700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2451456310679612e-06,
+      "loss": 0.6132,
+      "step": 2701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2444836716681377e-06,
+      "loss": 1.0542,
+      "step": 2702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2438217122683142e-06,
+      "loss": 0.7392,
+      "step": 2703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2431597528684907e-06,
+      "loss": 0.4323,
+      "step": 2704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2424977934686673e-06,
+      "loss": 1.908,
+      "step": 2705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.241835834068844e-06,
+      "loss": 0.3861,
+      "step": 2706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2411738746690203e-06,
+      "loss": 0.1055,
+      "step": 2707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2405119152691968e-06,
+      "loss": 2.9402,
+      "step": 2708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2398499558693735e-06,
+      "loss": 0.6689,
+      "step": 2709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2391879964695498e-06,
+      "loss": 0.7665,
+      "step": 2710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2385260370697265e-06,
+      "loss": 0.4858,
+      "step": 2711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.237864077669903e-06,
+      "loss": 1.514,
+      "step": 2712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2372021182700795e-06,
+      "loss": 0.9768,
+      "step": 2713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.236540158870256e-06,
+      "loss": 0.4531,
+      "step": 2714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2358781994704325e-06,
+      "loss": 0.6549,
+      "step": 2715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.235216240070609e-06,
+      "loss": 2.6675,
+      "step": 2716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2345542806707857e-06,
+      "loss": 0.0911,
+      "step": 2717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.233892321270962e-06,
+      "loss": 1.6741,
+      "step": 2718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2332303618711385e-06,
+      "loss": 0.5823,
+      "step": 2719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2325684024713152e-06,
+      "loss": 0.4306,
+      "step": 2720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2319064430714917e-06,
+      "loss": 0.6155,
+      "step": 2721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2312444836716682e-06,
+      "loss": 0.7787,
+      "step": 2722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2305825242718447e-06,
+      "loss": 0.2809,
+      "step": 2723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2299205648720212e-06,
+      "loss": 0.0966,
+      "step": 2724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2292586054721977e-06,
+      "loss": 0.6065,
+      "step": 2725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.2285966460723744e-06,
+      "loss": 0.3957,
+      "step": 2726
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2279346866725507e-06,
+      "loss": 0.8455,
+      "step": 2727
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2272727272727274e-06,
+      "loss": 1.7116,
+      "step": 2728
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.226610767872904e-06,
+      "loss": 0.7325,
+      "step": 2729
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2259488084730802e-06,
+      "loss": 0.7858,
+      "step": 2730
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.225286849073257e-06,
+      "loss": 0.6307,
+      "step": 2731
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2246248896734334e-06,
+      "loss": 1.0081,
+      "step": 2732
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.22396293027361e-06,
+      "loss": 1.4984,
+      "step": 2733
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2233009708737864e-06,
+      "loss": 0.6847,
+      "step": 2734
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.222639011473963e-06,
+      "loss": 0.5799,
+      "step": 2735
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2219770520741394e-06,
+      "loss": 0.4689,
+      "step": 2736
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2213150926743161e-06,
+      "loss": 0.5231,
+      "step": 2737
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2206531332744926e-06,
+      "loss": 0.0672,
+      "step": 2738
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2199911738746691e-06,
+      "loss": 0.9273,
+      "step": 2739
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2193292144748456e-06,
+      "loss": 1.4274,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2186672550750221e-06,
+      "loss": 0.2688,
+      "step": 2741
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2180052956751986e-06,
+      "loss": 0.7535,
+      "step": 2742
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2173433362753751e-06,
+      "loss": 0.3619,
+      "step": 2743
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2166813768755516e-06,
+      "loss": 0.9937,
+      "step": 2744
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2160194174757281e-06,
+      "loss": 0.6268,
+      "step": 2745
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2153574580759048e-06,
+      "loss": 0.8107,
+      "step": 2746
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2146954986760811e-06,
+      "loss": 1.1374,
+      "step": 2747
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2140335392762578e-06,
+      "loss": 0.8304,
+      "step": 2748
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2133715798764343e-06,
+      "loss": 1.0893,
+      "step": 2749
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2127096204766108e-06,
+      "loss": 0.6483,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2120476610767873e-06,
+      "loss": 0.9463,
+      "step": 2751
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2113857016769638e-06,
+      "loss": 1.1622,
+      "step": 2752
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2107237422771403e-06,
+      "loss": 1.6314,
+      "step": 2753
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2100617828773168e-06,
+      "loss": 1.3916,
+      "step": 2754
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2093998234774936e-06,
+      "loss": 0.8199,
+      "step": 2755
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2087378640776698e-06,
+      "loss": 0.0638,
+      "step": 2756
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2080759046778466e-06,
+      "loss": 1.4808,
+      "step": 2757
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.207413945278023e-06,
+      "loss": 2.4838,
+      "step": 2758
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2067519858781996e-06,
+      "loss": 2.5688,
+      "step": 2759
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.206090026478376e-06,
+      "loss": 1.1179,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2054280670785526e-06,
+      "loss": 0.8425,
+      "step": 2761
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.204766107678729e-06,
+      "loss": 0.5035,
+      "step": 2762
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2041041482789056e-06,
+      "loss": 0.8617,
+      "step": 2763
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.203442188879082e-06,
+      "loss": 1.0706,
+      "step": 2764
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2027802294792586e-06,
+      "loss": 0.6489,
+      "step": 2765
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2021182700794353e-06,
+      "loss": 0.1257,
+      "step": 2766
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2014563106796116e-06,
+      "loss": 0.6085,
+      "step": 2767
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2007943512797883e-06,
+      "loss": 0.5092,
+      "step": 2768
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.2001323918799648e-06,
+      "loss": 1.4962,
+      "step": 2769
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1994704324801413e-06,
+      "loss": 1.5125,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1988084730803178e-06,
+      "loss": 1.7712,
+      "step": 2771
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.1981465136804943e-06,
+      "loss": 2.6818,
+      "step": 2772
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1974845542806708e-06,
+      "loss": 1.0654,
+      "step": 2773
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1968225948808473e-06,
+      "loss": 2.7952,
+      "step": 2774
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.196160635481024e-06,
+      "loss": 0.9049,
+      "step": 2775
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1954986760812003e-06,
+      "loss": 0.2321,
+      "step": 2776
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.194836716681377e-06,
+      "loss": 1.5255,
+      "step": 2777
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1941747572815535e-06,
+      "loss": 0.1128,
+      "step": 2778
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.19351279788173e-06,
+      "loss": 1.1152,
+      "step": 2779
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1928508384819065e-06,
+      "loss": 0.885,
+      "step": 2780
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.192188879082083e-06,
+      "loss": 1.0291,
+      "step": 2781
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1915269196822595e-06,
+      "loss": 1.9325,
+      "step": 2782
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.190864960282436e-06,
+      "loss": 0.6827,
+      "step": 2783
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1902030008826125e-06,
+      "loss": 1.1849,
+      "step": 2784
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.189541041482789e-06,
+      "loss": 0.8816,
+      "step": 2785
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1888790820829657e-06,
+      "loss": 0.2959,
+      "step": 2786
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.188217122683142e-06,
+      "loss": 1.4543,
+      "step": 2787
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1875551632833187e-06,
+      "loss": 1.7087,
+      "step": 2788
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1868932038834952e-06,
+      "loss": 1.3229,
+      "step": 2789
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1862312444836717e-06,
+      "loss": 3.9809,
+      "step": 2790
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1855692850838482e-06,
+      "loss": 0.5706,
+      "step": 2791
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.184907325684025e-06,
+      "loss": 2.0482,
+      "step": 2792
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1842453662842012e-06,
+      "loss": 2.2642,
+      "step": 2793
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1835834068843777e-06,
+      "loss": 1.5126,
+      "step": 2794
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1829214474845544e-06,
+      "loss": 0.0394,
+      "step": 2795
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1822594880847307e-06,
+      "loss": 0.7172,
+      "step": 2796
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1815975286849074e-06,
+      "loss": 1.6218,
+      "step": 2797
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.180935569285084e-06,
+      "loss": 1.2088,
+      "step": 2798
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1802736098852604e-06,
+      "loss": 1.5627,
+      "step": 2799
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.179611650485437e-06,
+      "loss": 0.5293,
+      "step": 2800
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1789496910856134e-06,
+      "loss": 2.1892,
+      "step": 2801
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.17828773168579e-06,
+      "loss": 2.1993,
+      "step": 2802
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1776257722859666e-06,
+      "loss": 0.0733,
+      "step": 2803
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.176963812886143e-06,
+      "loss": 0.8377,
+      "step": 2804
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1763018534863194e-06,
+      "loss": 3.3965,
+      "step": 2805
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1756398940864961e-06,
+      "loss": 0.5695,
+      "step": 2806
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1749779346866724e-06,
+      "loss": 2.7834,
+      "step": 2807
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1743159752868491e-06,
+      "loss": 0.4034,
+      "step": 2808
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1736540158870256e-06,
+      "loss": 0.7185,
+      "step": 2809
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1729920564872021e-06,
+      "loss": 1.6749,
+      "step": 2810
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1723300970873786e-06,
+      "loss": 0.0742,
+      "step": 2811
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1716681376875554e-06,
+      "loss": 0.5162,
+      "step": 2812
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1710061782877316e-06,
+      "loss": 0.9948,
+      "step": 2813
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1703442188879081e-06,
+      "loss": 0.5964,
+      "step": 2814
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1696822594880849e-06,
+      "loss": 0.2877,
+      "step": 2815
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1690203000882611e-06,
+      "loss": 0.382,
+      "step": 2816
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.1683583406884379e-06,
+      "loss": 1.5812,
+      "step": 2817
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1676963812886144e-06,
+      "loss": 1.3563,
+      "step": 2818
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1670344218887909e-06,
+      "loss": 1.4136,
+      "step": 2819
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1663724624889674e-06,
+      "loss": 0.6785,
+      "step": 2820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1657105030891439e-06,
+      "loss": 0.111,
+      "step": 2821
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1650485436893204e-06,
+      "loss": 1.1861,
+      "step": 2822
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.164386584289497e-06,
+      "loss": 0.3681,
+      "step": 2823
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1637246248896734e-06,
+      "loss": 1.0429,
+      "step": 2824
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1630626654898499e-06,
+      "loss": 0.499,
+      "step": 2825
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1624007060900266e-06,
+      "loss": 1.023,
+      "step": 2826
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1617387466902029e-06,
+      "loss": 0.9098,
+      "step": 2827
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1610767872903796e-06,
+      "loss": 1.4818,
+      "step": 2828
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.160414827890556e-06,
+      "loss": 1.285,
+      "step": 2829
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1597528684907326e-06,
+      "loss": 1.5058,
+      "step": 2830
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.159090909090909e-06,
+      "loss": 0.8501,
+      "step": 2831
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1584289496910858e-06,
+      "loss": 1.7981,
+      "step": 2832
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.157766990291262e-06,
+      "loss": 0.9995,
+      "step": 2833
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1571050308914388e-06,
+      "loss": 1.6077,
+      "step": 2834
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1564430714916153e-06,
+      "loss": 0.4637,
+      "step": 2835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1557811120917916e-06,
+      "loss": 0.5845,
+      "step": 2836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1551191526919683e-06,
+      "loss": 0.5633,
+      "step": 2837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1544571932921448e-06,
+      "loss": 1.5673,
+      "step": 2838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1537952338923213e-06,
+      "loss": 1.5335,
+      "step": 2839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1531332744924978e-06,
+      "loss": 0.9553,
+      "step": 2840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1524713150926743e-06,
+      "loss": 1.2184,
+      "step": 2841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1518093556928508e-06,
+      "loss": 1.2109,
+      "step": 2842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1511473962930275e-06,
+      "loss": 0.6722,
+      "step": 2843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1504854368932038e-06,
+      "loss": 0.6552,
+      "step": 2844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1498234774933805e-06,
+      "loss": 0.9383,
+      "step": 2845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.149161518093557e-06,
+      "loss": 0.1308,
+      "step": 2846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1484995586937333e-06,
+      "loss": 1.7325,
+      "step": 2847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.14783759929391e-06,
+      "loss": 0.3883,
+      "step": 2848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1471756398940865e-06,
+      "loss": 1.7186,
+      "step": 2849
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.146513680494263e-06,
+      "loss": 0.2422,
+      "step": 2850
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1458517210944395e-06,
+      "loss": 0.7981,
+      "step": 2851
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1451897616946162e-06,
+      "loss": 1.1791,
+      "step": 2852
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1445278022947925e-06,
+      "loss": 3.5645,
+      "step": 2853
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1438658428949692e-06,
+      "loss": 0.4967,
+      "step": 2854
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1432038834951457e-06,
+      "loss": 0.5555,
+      "step": 2855
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1425419240953222e-06,
+      "loss": 0.1791,
+      "step": 2856
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1418799646954987e-06,
+      "loss": 0.2457,
+      "step": 2857
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1412180052956752e-06,
+      "loss": 2.1792,
+      "step": 2858
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1405560458958517e-06,
+      "loss": 1.7119,
+      "step": 2859
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1398940864960282e-06,
+      "loss": 0.474,
+      "step": 2860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1392321270962047e-06,
+      "loss": 0.9572,
+      "step": 2861
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.1385701676963812e-06,
+      "loss": 0.6006,
+      "step": 2862
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.137908208296558e-06,
+      "loss": 0.5752,
+      "step": 2863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1372462488967342e-06,
+      "loss": 1.0213,
+      "step": 2864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.136584289496911e-06,
+      "loss": 1.076,
+      "step": 2865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1359223300970875e-06,
+      "loss": 1.2673,
+      "step": 2866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.135260370697264e-06,
+      "loss": 1.0625,
+      "step": 2867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1345984112974405e-06,
+      "loss": 0.7914,
+      "step": 2868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.133936451897617e-06,
+      "loss": 0.9434,
+      "step": 2869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1332744924977935e-06,
+      "loss": 1.1796,
+      "step": 2870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.13261253309797e-06,
+      "loss": 0.4676,
+      "step": 2871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1319505736981467e-06,
+      "loss": 0.3383,
+      "step": 2872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.131288614298323e-06,
+      "loss": 0.5241,
+      "step": 2873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1306266548984997e-06,
+      "loss": 1.2013,
+      "step": 2874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1299646954986762e-06,
+      "loss": 1.1268,
+      "step": 2875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1293027360988527e-06,
+      "loss": 0.2607,
+      "step": 2876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1286407766990292e-06,
+      "loss": 1.2586,
+      "step": 2877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1279788172992057e-06,
+      "loss": 2.5521,
+      "step": 2878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1273168578993822e-06,
+      "loss": 0.5473,
+      "step": 2879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1266548984995587e-06,
+      "loss": 1.4046,
+      "step": 2880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1259929390997352e-06,
+      "loss": 2.8492,
+      "step": 2881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1253309796999117e-06,
+      "loss": 1.9119,
+      "step": 2882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1246690203000884e-06,
+      "loss": 0.7623,
+      "step": 2883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1240070609002649e-06,
+      "loss": 3.0763,
+      "step": 2884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1233451015004414e-06,
+      "loss": 0.2254,
+      "step": 2885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1226831421006179e-06,
+      "loss": 1.6613,
+      "step": 2886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1220211827007944e-06,
+      "loss": 1.2115,
+      "step": 2887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1213592233009709e-06,
+      "loss": 0.5981,
+      "step": 2888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1206972639011474e-06,
+      "loss": 0.4388,
+      "step": 2889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1200353045013239e-06,
+      "loss": 1.4494,
+      "step": 2890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1193733451015004e-06,
+      "loss": 1.9328,
+      "step": 2891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.118711385701677e-06,
+      "loss": 0.6925,
+      "step": 2892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1180494263018534e-06,
+      "loss": 1.2596,
+      "step": 2893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.11738746690203e-06,
+      "loss": 2.4574,
+      "step": 2894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1167255075022066e-06,
+      "loss": 0.2515,
+      "step": 2895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.116063548102383e-06,
+      "loss": 1.242,
+      "step": 2896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1154015887025596e-06,
+      "loss": 1.3031,
+      "step": 2897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.114739629302736e-06,
+      "loss": 1.5822,
+      "step": 2898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1140776699029126e-06,
+      "loss": 0.9982,
+      "step": 2899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1134157105030891e-06,
+      "loss": 1.1151,
+      "step": 2900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1127537511032658e-06,
+      "loss": 1.2145,
+      "step": 2901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1120917917034421e-06,
+      "loss": 0.4757,
+      "step": 2902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1114298323036188e-06,
+      "loss": 0.3167,
+      "step": 2903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1107678729037953e-06,
+      "loss": 0.3523,
+      "step": 2904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1101059135039718e-06,
+      "loss": 0.9603,
+      "step": 2905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1094439541041483e-06,
+      "loss": 1.4097,
+      "step": 2906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1087819947043248e-06,
+      "loss": 0.3005,
+      "step": 2907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1081200353045013e-06,
+      "loss": 0.3332,
+      "step": 2908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.1074580759046778e-06,
+      "loss": 0.3539,
+      "step": 2909
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1067961165048543e-06,
+      "loss": 1.7311,
+      "step": 2910
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1061341571050308e-06,
+      "loss": 0.7149,
+      "step": 2911
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1054721977052075e-06,
+      "loss": 0.0442,
+      "step": 2912
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1048102383053838e-06,
+      "loss": 0.7266,
+      "step": 2913
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1041482789055605e-06,
+      "loss": 0.6735,
+      "step": 2914
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.103486319505737e-06,
+      "loss": 1.9802,
+      "step": 2915
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1028243601059135e-06,
+      "loss": 0.6086,
+      "step": 2916
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.10216240070609e-06,
+      "loss": 0.7916,
+      "step": 2917
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1015004413062668e-06,
+      "loss": 1.7714,
+      "step": 2918
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.100838481906443e-06,
+      "loss": 1.2435,
+      "step": 2919
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.1001765225066195e-06,
+      "loss": 1.0556,
+      "step": 2920
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0995145631067963e-06,
+      "loss": 1.1836,
+      "step": 2921
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0988526037069725e-06,
+      "loss": 0.4959,
+      "step": 2922
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0981906443071493e-06,
+      "loss": 0.4881,
+      "step": 2923
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0975286849073258e-06,
+      "loss": 1.621,
+      "step": 2924
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0968667255075023e-06,
+      "loss": 0.2008,
+      "step": 2925
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0962047661076788e-06,
+      "loss": 1.0747,
+      "step": 2926
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0955428067078553e-06,
+      "loss": 1.5859,
+      "step": 2927
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0948808473080318e-06,
+      "loss": 2.9099,
+      "step": 2928
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0942188879082085e-06,
+      "loss": 0.895,
+      "step": 2929
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0935569285083848e-06,
+      "loss": 0.8008,
+      "step": 2930
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0928949691085613e-06,
+      "loss": 0.5909,
+      "step": 2931
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.092233009708738e-06,
+      "loss": 0.5164,
+      "step": 2932
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0915710503089143e-06,
+      "loss": 0.4544,
+      "step": 2933
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.090909090909091e-06,
+      "loss": 0.0778,
+      "step": 2934
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0902471315092675e-06,
+      "loss": 0.9032,
+      "step": 2935
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.089585172109444e-06,
+      "loss": 1.9278,
+      "step": 2936
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0889232127096205e-06,
+      "loss": 0.0418,
+      "step": 2937
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0882612533097972e-06,
+      "loss": 0.6368,
+      "step": 2938
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0875992939099735e-06,
+      "loss": 1.2453,
+      "step": 2939
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0869373345101502e-06,
+      "loss": 2.6808,
+      "step": 2940
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0862753751103267e-06,
+      "loss": 0.7588,
+      "step": 2941
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.085613415710503e-06,
+      "loss": 0.6183,
+      "step": 2942
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0849514563106797e-06,
+      "loss": 0.6482,
+      "step": 2943
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0842894969108562e-06,
+      "loss": 0.9427,
+      "step": 2944
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0836275375110327e-06,
+      "loss": 1.3203,
+      "step": 2945
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0829655781112092e-06,
+      "loss": 1.7216,
+      "step": 2946
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0823036187113857e-06,
+      "loss": 0.0339,
+      "step": 2947
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0816416593115622e-06,
+      "loss": 1.1144,
+      "step": 2948
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.080979699911739e-06,
+      "loss": 0.8057,
+      "step": 2949
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0803177405119152e-06,
+      "loss": 0.248,
+      "step": 2950
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.079655781112092e-06,
+      "loss": 0.272,
+      "step": 2951
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0789938217122684e-06,
+      "loss": 0.4512,
+      "step": 2952
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0783318623124447e-06,
+      "loss": 1.3192,
+      "step": 2953
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.0776699029126214e-06,
+      "loss": 0.8342,
+      "step": 2954
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.077007943512798e-06,
+      "loss": 0.0637,
+      "step": 2955
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0763459841129744e-06,
+      "loss": 1.7672,
+      "step": 2956
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.075684024713151e-06,
+      "loss": 0.9881,
+      "step": 2957
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0750220653133276e-06,
+      "loss": 1.9522,
+      "step": 2958
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.074360105913504e-06,
+      "loss": 0.4196,
+      "step": 2959
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0736981465136806e-06,
+      "loss": 1.3025,
+      "step": 2960
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0730361871138571e-06,
+      "loss": 1.7515,
+      "step": 2961
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0723742277140334e-06,
+      "loss": 0.4282,
+      "step": 2962
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0717122683142101e-06,
+      "loss": 0.6807,
+      "step": 2963
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0710503089143866e-06,
+      "loss": 1.3137,
+      "step": 2964
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0703883495145631e-06,
+      "loss": 0.7445,
+      "step": 2965
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0697263901147396e-06,
+      "loss": 1.5282,
+      "step": 2966
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0690644307149161e-06,
+      "loss": 0.2884,
+      "step": 2967
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0684024713150926e-06,
+      "loss": 0.363,
+      "step": 2968
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0677405119152693e-06,
+      "loss": 1.0554,
+      "step": 2969
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0670785525154456e-06,
+      "loss": 1.2908,
+      "step": 2970
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0664165931156223e-06,
+      "loss": 1.8463,
+      "step": 2971
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0657546337157988e-06,
+      "loss": 1.8173,
+      "step": 2972
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0650926743159751e-06,
+      "loss": 0.659,
+      "step": 2973
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0644307149161518e-06,
+      "loss": 0.4862,
+      "step": 2974
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0637687555163283e-06,
+      "loss": 0.4771,
+      "step": 2975
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0631067961165048e-06,
+      "loss": 0.6967,
+      "step": 2976
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0624448367166813e-06,
+      "loss": 1.8448,
+      "step": 2977
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.061782877316858e-06,
+      "loss": 0.1301,
+      "step": 2978
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0611209179170344e-06,
+      "loss": 1.3185,
+      "step": 2979
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.060458958517211e-06,
+      "loss": 0.8072,
+      "step": 2980
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0597969991173876e-06,
+      "loss": 0.7017,
+      "step": 2981
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.059135039717564e-06,
+      "loss": 2.2304,
+      "step": 2982
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0584730803177406e-06,
+      "loss": 0.0908,
+      "step": 2983
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.057811120917917e-06,
+      "loss": 1.4417,
+      "step": 2984
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0571491615180936e-06,
+      "loss": 1.2665,
+      "step": 2985
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.05648720211827e-06,
+      "loss": 0.0861,
+      "step": 2986
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0558252427184466e-06,
+      "loss": 0.9827,
+      "step": 2987
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.055163283318623e-06,
+      "loss": 0.5273,
+      "step": 2988
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0545013239187998e-06,
+      "loss": 1.0641,
+      "step": 2989
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.053839364518976e-06,
+      "loss": 0.1396,
+      "step": 2990
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0531774051191528e-06,
+      "loss": 0.6529,
+      "step": 2991
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0525154457193293e-06,
+      "loss": 0.5999,
+      "step": 2992
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0518534863195058e-06,
+      "loss": 0.9785,
+      "step": 2993
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0511915269196823e-06,
+      "loss": 1.0088,
+      "step": 2994
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0505295675198588e-06,
+      "loss": 1.0204,
+      "step": 2995
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0498676081200353e-06,
+      "loss": 1.1889,
+      "step": 2996
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0492056487202118e-06,
+      "loss": 1.1033,
+      "step": 2997
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0485436893203885e-06,
+      "loss": 1.3201,
+      "step": 2998
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0478817299205648e-06,
+      "loss": 2.2408,
+      "step": 2999
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.0472197705207415e-06,
+      "loss": 1.3312,
+      "step": 3000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.046557811120918e-06,
+      "loss": 1.2402,
+      "step": 3001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0458958517210945e-06,
+      "loss": 1.1494,
+      "step": 3002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.045233892321271e-06,
+      "loss": 1.0762,
+      "step": 3003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0445719329214475e-06,
+      "loss": 1.5871,
+      "step": 3004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.043909973521624e-06,
+      "loss": 0.1639,
+      "step": 3005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0432480141218005e-06,
+      "loss": 1.5894,
+      "step": 3006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.042586054721977e-06,
+      "loss": 0.9321,
+      "step": 3007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0419240953221535e-06,
+      "loss": 0.5634,
+      "step": 3008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0412621359223302e-06,
+      "loss": 0.0734,
+      "step": 3009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0406001765225065e-06,
+      "loss": 0.8447,
+      "step": 3010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0399382171226832e-06,
+      "loss": 1.5755,
+      "step": 3011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0392762577228597e-06,
+      "loss": 0.8206,
+      "step": 3012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0386142983230362e-06,
+      "loss": 0.8341,
+      "step": 3013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0379523389232127e-06,
+      "loss": 0.5277,
+      "step": 3014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0372903795233892e-06,
+      "loss": 0.6258,
+      "step": 3015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0366284201235657e-06,
+      "loss": 0.7128,
+      "step": 3016
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0359664607237422e-06,
+      "loss": 1.1768,
+      "step": 3017
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.035304501323919e-06,
+      "loss": 0.8126,
+      "step": 3018
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0346425419240952e-06,
+      "loss": 1.8508,
+      "step": 3019
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.033980582524272e-06,
+      "loss": 0.9321,
+      "step": 3020
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0333186231244484e-06,
+      "loss": 2.0563,
+      "step": 3021
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.032656663724625e-06,
+      "loss": 2.2975,
+      "step": 3022
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0319947043248014e-06,
+      "loss": 0.4815,
+      "step": 3023
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.031332744924978e-06,
+      "loss": 0.5314,
+      "step": 3024
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0306707855251544e-06,
+      "loss": 0.0525,
+      "step": 3025
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.030008826125331e-06,
+      "loss": 1.1339,
+      "step": 3026
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0293468667255074e-06,
+      "loss": 0.6392,
+      "step": 3027
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.028684907325684e-06,
+      "loss": 0.7209,
+      "step": 3028
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0280229479258607e-06,
+      "loss": 1.5975,
+      "step": 3029
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0273609885260372e-06,
+      "loss": 1.6794,
+      "step": 3030
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0266990291262137e-06,
+      "loss": 1.434,
+      "step": 3031
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0260370697263902e-06,
+      "loss": 1.3334,
+      "step": 3032
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0253751103265667e-06,
+      "loss": 1.2222,
+      "step": 3033
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0247131509267432e-06,
+      "loss": 1.3936,
+      "step": 3034
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0240511915269199e-06,
+      "loss": 0.3684,
+      "step": 3035
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0233892321270962e-06,
+      "loss": 0.4687,
+      "step": 3036
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0227272727272727e-06,
+      "loss": 1.2179,
+      "step": 3037
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0220653133274494e-06,
+      "loss": 0.0739,
+      "step": 3038
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0214033539276257e-06,
+      "loss": 0.8333,
+      "step": 3039
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0207413945278024e-06,
+      "loss": 2.4258,
+      "step": 3040
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0200794351279789e-06,
+      "loss": 1.3485,
+      "step": 3041
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0194174757281554e-06,
+      "loss": 0.7166,
+      "step": 3042
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0187555163283319e-06,
+      "loss": 0.8506,
+      "step": 3043
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0180935569285084e-06,
+      "loss": 0.6303,
+      "step": 3044
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0174315975286849e-06,
+      "loss": 0.5692,
+      "step": 3045
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.0167696381288616e-06,
+      "loss": 2.396,
+      "step": 3046
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.016107678729038e-06,
+      "loss": 1.2275,
+      "step": 3047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0154457193292144e-06,
+      "loss": 1.0868,
+      "step": 3048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.014783759929391e-06,
+      "loss": 1.7453,
+      "step": 3049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0141218005295676e-06,
+      "loss": 1.6605,
+      "step": 3050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.013459841129744e-06,
+      "loss": 0.693,
+      "step": 3051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0127978817299206e-06,
+      "loss": 1.6304,
+      "step": 3052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.012135922330097e-06,
+      "loss": 1.0094,
+      "step": 3053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0114739629302736e-06,
+      "loss": 0.7682,
+      "step": 3054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0108120035304503e-06,
+      "loss": 1.1304,
+      "step": 3055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0101500441306266e-06,
+      "loss": 1.2902,
+      "step": 3056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0094880847308033e-06,
+      "loss": 1.6229,
+      "step": 3057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0088261253309798e-06,
+      "loss": 1.6661,
+      "step": 3058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.008164165931156e-06,
+      "loss": 2.9624,
+      "step": 3059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0075022065313328e-06,
+      "loss": 0.2274,
+      "step": 3060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0068402471315093e-06,
+      "loss": 0.6129,
+      "step": 3061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0061782877316858e-06,
+      "loss": 0.9871,
+      "step": 3062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0055163283318623e-06,
+      "loss": 0.7021,
+      "step": 3063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0048543689320388e-06,
+      "loss": 1.8318,
+      "step": 3064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0041924095322153e-06,
+      "loss": 0.9984,
+      "step": 3065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.003530450132392e-06,
+      "loss": 1.6574,
+      "step": 3066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0028684907325685e-06,
+      "loss": 0.951,
+      "step": 3067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0022065313327448e-06,
+      "loss": 1.0082,
+      "step": 3068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0015445719329215e-06,
+      "loss": 0.1888,
+      "step": 3069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.000882612533098e-06,
+      "loss": 1.9588,
+      "step": 3070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.0002206531332745e-06,
+      "loss": 0.3796,
+      "step": 3071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.99558693733451e-07,
+      "loss": 0.7528,
+      "step": 3072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.988967343336275e-07,
+      "loss": 0.5425,
+      "step": 3073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.98234774933804e-07,
+      "loss": 0.4884,
+      "step": 3074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.975728155339807e-07,
+      "loss": 1.0875,
+      "step": 3075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.96910856134157e-07,
+      "loss": 0.6469,
+      "step": 3076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.962488967343337e-07,
+      "loss": 2.1627,
+      "step": 3077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.955869373345102e-07,
+      "loss": 0.6663,
+      "step": 3078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.949249779346865e-07,
+      "loss": 0.944,
+      "step": 3079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.942630185348632e-07,
+      "loss": 2.3003,
+      "step": 3080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.936010591350397e-07,
+      "loss": 2.9852,
+      "step": 3081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.929390997352162e-07,
+      "loss": 0.0827,
+      "step": 3082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.922771403353927e-07,
+      "loss": 0.2601,
+      "step": 3083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.916151809355695e-07,
+      "loss": 3.0514,
+      "step": 3084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.909532215357457e-07,
+      "loss": 1.6765,
+      "step": 3085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.902912621359225e-07,
+      "loss": 0.0635,
+      "step": 3086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.89629302736099e-07,
+      "loss": 0.8115,
+      "step": 3087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.889673433362755e-07,
+      "loss": 0.3742,
+      "step": 3088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.88305383936452e-07,
+      "loss": 1.1082,
+      "step": 3089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.876434245366285e-07,
+      "loss": 0.676,
+      "step": 3090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.86981465136805e-07,
+      "loss": 0.1095,
+      "step": 3091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.863195057369815e-07,
+      "loss": 1.1516,
+      "step": 3092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.85657546337158e-07,
+      "loss": 1.25,
+      "step": 3093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.849955869373345e-07,
+      "loss": 2.6321,
+      "step": 3094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.843336275375112e-07,
+      "loss": 0.5547,
+      "step": 3095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.836716681376875e-07,
+      "loss": 1.061,
+      "step": 3096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.830097087378642e-07,
+      "loss": 0.121,
+      "step": 3097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.823477493380407e-07,
+      "loss": 1.6902,
+      "step": 3098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.816857899382172e-07,
+      "loss": 0.6063,
+      "step": 3099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.810238305383937e-07,
+      "loss": 0.9318,
+      "step": 3100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.803618711385702e-07,
+      "loss": 2.54,
+      "step": 3101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.796999117387467e-07,
+      "loss": 2.0318,
+      "step": 3102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.790379523389232e-07,
+      "loss": 2.5138,
+      "step": 3103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.783759929390999e-07,
+      "loss": 0.812,
+      "step": 3104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.777140335392762e-07,
+      "loss": 1.918,
+      "step": 3105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.770520741394529e-07,
+      "loss": 1.2303,
+      "step": 3106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.763901147396294e-07,
+      "loss": 0.2019,
+      "step": 3107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.75728155339806e-07,
+      "loss": 1.4502,
+      "step": 3108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.750661959399824e-07,
+      "loss": 1.0553,
+      "step": 3109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.74404236540159e-07,
+      "loss": 2.1926,
+      "step": 3110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.737422771403354e-07,
+      "loss": 0.6215,
+      "step": 3111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.73080317740512e-07,
+      "loss": 1.1597,
+      "step": 3112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.724183583406884e-07,
+      "loss": 1.9212,
+      "step": 3113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.71756398940865e-07,
+      "loss": 0.3816,
+      "step": 3114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.710944395410416e-07,
+      "loss": 0.1239,
+      "step": 3115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.70432480141218e-07,
+      "loss": 0.7181,
+      "step": 3116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.697705207413946e-07,
+      "loss": 0.8858,
+      "step": 3117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.691085613415711e-07,
+      "loss": 1.3531,
+      "step": 3118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.684466019417476e-07,
+      "loss": 0.4392,
+      "step": 3119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.677846425419241e-07,
+      "loss": 1.5549,
+      "step": 3120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.671226831421006e-07,
+      "loss": 0.4148,
+      "step": 3121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.664607237422771e-07,
+      "loss": 0.1487,
+      "step": 3122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.657987643424536e-07,
+      "loss": 0.6977,
+      "step": 3123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.651368049426303e-07,
+      "loss": 1.0744,
+      "step": 3124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.644748455428066e-07,
+      "loss": 0.6608,
+      "step": 3125
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.638128861429833e-07,
+      "loss": 0.5458,
+      "step": 3126
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.631509267431598e-07,
+      "loss": 1.7257,
+      "step": 3127
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.624889673433363e-07,
+      "loss": 0.5063,
+      "step": 3128
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.618270079435128e-07,
+      "loss": 1.2214,
+      "step": 3129
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.611650485436893e-07,
+      "loss": 1.3366,
+      "step": 3130
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.605030891438658e-07,
+      "loss": 0.7538,
+      "step": 3131
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.598411297440423e-07,
+      "loss": 0.1637,
+      "step": 3132
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.591791703442188e-07,
+      "loss": 0.4012,
+      "step": 3133
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.585172109443953e-07,
+      "loss": 1.0087,
+      "step": 3134
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.57855251544572e-07,
+      "loss": 0.6637,
+      "step": 3135
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.571932921447483e-07,
+      "loss": 0.4743,
+      "step": 3136
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.56531332744925e-07,
+      "loss": 2.151,
+      "step": 3137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.558693733451015e-07,
+      "loss": 0.6072,
+      "step": 3138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.55207413945278e-07,
+      "loss": 1.0542,
+      "step": 3139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.545454545454546e-07,
+      "loss": 1.7353,
+      "step": 3140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.538834951456313e-07,
+      "loss": 1.8788,
+      "step": 3141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.532215357458077e-07,
+      "loss": 1.38,
+      "step": 3142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.52559576345984e-07,
+      "loss": 1.0739,
+      "step": 3143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.518976169461607e-07,
+      "loss": 1.7513,
+      "step": 3144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.512356575463372e-07,
+      "loss": 1.2446,
+      "step": 3145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.505736981465138e-07,
+      "loss": 1.7963,
+      "step": 3146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.499117387466902e-07,
+      "loss": 0.4947,
+      "step": 3147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.492497793468668e-07,
+      "loss": 2.6051,
+      "step": 3148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.485878199470433e-07,
+      "loss": 1.5031,
+      "step": 3149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.479258605472199e-07,
+      "loss": 0.8069,
+      "step": 3150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.472639011473963e-07,
+      "loss": 0.9392,
+      "step": 3151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.466019417475729e-07,
+      "loss": 0.9475,
+      "step": 3152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.459399823477494e-07,
+      "loss": 0.6906,
+      "step": 3153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.452780229479258e-07,
+      "loss": 0.5997,
+      "step": 3154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.446160635481024e-07,
+      "loss": 0.6319,
+      "step": 3155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.439541041482789e-07,
+      "loss": 1.686,
+      "step": 3156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.432921447484555e-07,
+      "loss": 0.8619,
+      "step": 3157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.42630185348632e-07,
+      "loss": 0.603,
+      "step": 3158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.419682259488086e-07,
+      "loss": 0.4932,
+      "step": 3159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.41306266548985e-07,
+      "loss": 2.3929,
+      "step": 3160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.406443071491616e-07,
+      "loss": 1.7158,
+      "step": 3161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.399823477493381e-07,
+      "loss": 0.4657,
+      "step": 3162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.393203883495147e-07,
+      "loss": 1.5063,
+      "step": 3163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.386584289496911e-07,
+      "loss": 3.1709,
+      "step": 3164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.379964695498676e-07,
+      "loss": 1.3913,
+      "step": 3165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.373345101500442e-07,
+      "loss": 0.5029,
+      "step": 3166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.366725507502206e-07,
+      "loss": 0.8273,
+      "step": 3167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.360105913503972e-07,
+      "loss": 2.5389,
+      "step": 3168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.353486319505737e-07,
+      "loss": 2.5626,
+      "step": 3169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.346866725507503e-07,
+      "loss": 0.8794,
+      "step": 3170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.340247131509267e-07,
+      "loss": 0.6838,
+      "step": 3171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.333627537511033e-07,
+      "loss": 0.5338,
+      "step": 3172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.327007943512798e-07,
+      "loss": 0.4535,
+      "step": 3173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.320388349514562e-07,
+      "loss": 2.4795,
+      "step": 3174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.313768755516329e-07,
+      "loss": 0.685,
+      "step": 3175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.307149161518093e-07,
+      "loss": 0.4982,
+      "step": 3176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.300529567519859e-07,
+      "loss": 0.7941,
+      "step": 3177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.293909973521624e-07,
+      "loss": 1.1536,
+      "step": 3178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.28729037952339e-07,
+      "loss": 0.1885,
+      "step": 3179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.280670785525154e-07,
+      "loss": 0.2542,
+      "step": 3180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.27405119152692e-07,
+      "loss": 0.1855,
+      "step": 3181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.267431597528685e-07,
+      "loss": 0.469,
+      "step": 3182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.260812003530451e-07,
+      "loss": 1.1623,
+      "step": 3183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.254192409532215e-07,
+      "loss": 2.2996,
+      "step": 3184
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.24757281553398e-07,
+      "loss": 0.1092,
+      "step": 3185
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.240953221535746e-07,
+      "loss": 1.0444,
+      "step": 3186
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.23433362753751e-07,
+      "loss": 2.6421,
+      "step": 3187
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.227714033539276e-07,
+      "loss": 1.495,
+      "step": 3188
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.221094439541041e-07,
+      "loss": 1.4715,
+      "step": 3189
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.214474845542807e-07,
+      "loss": 0.6397,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.207855251544571e-07,
+      "loss": 1.3636,
+      "step": 3191
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.201235657546337e-07,
+      "loss": 1.8862,
+      "step": 3192
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.194616063548102e-07,
+      "loss": 1.7406,
+      "step": 3193
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.187996469549869e-07,
+      "loss": 3.4139,
+      "step": 3194
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.181376875551634e-07,
+      "loss": 0.9087,
+      "step": 3195
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.174757281553397e-07,
+      "loss": 2.0144,
+      "step": 3196
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.168137687555164e-07,
+      "loss": 0.1401,
+      "step": 3197
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.161518093556929e-07,
+      "loss": 0.0448,
+      "step": 3198
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.154898499558695e-07,
+      "loss": 0.1257,
+      "step": 3199
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.148278905560459e-07,
+      "loss": 0.2315,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.141659311562225e-07,
+      "loss": 1.9799,
+      "step": 3201
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.13503971756399e-07,
+      "loss": 0.7533,
+      "step": 3202
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.128420123565756e-07,
+      "loss": 0.2163,
+      "step": 3203
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.12180052956752e-07,
+      "loss": 0.3956,
+      "step": 3204
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.115180935569286e-07,
+      "loss": 0.9731,
+      "step": 3205
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.108561341571051e-07,
+      "loss": 1.0796,
+      "step": 3206
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.101941747572815e-07,
+      "loss": 0.5416,
+      "step": 3207
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.095322153574581e-07,
+      "loss": 1.843,
+      "step": 3208
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.088702559576346e-07,
+      "loss": 0.8141,
+      "step": 3209
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.082082965578112e-07,
+      "loss": 1.2609,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.075463371579876e-07,
+      "loss": 2.2731,
+      "step": 3211
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.068843777581643e-07,
+      "loss": 1.655,
+      "step": 3212
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.062224183583407e-07,
+      "loss": 1.0436,
+      "step": 3213
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.055604589585173e-07,
+      "loss": 2.5583,
+      "step": 3214
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.048984995586938e-07,
+      "loss": 0.452,
+      "step": 3215
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.042365401588704e-07,
+      "loss": 1.5631,
+      "step": 3216
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.035745807590468e-07,
+      "loss": 1.289,
+      "step": 3217
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.029126213592233e-07,
+      "loss": 2.4465,
+      "step": 3218
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.022506619593999e-07,
+      "loss": 0.5637,
+      "step": 3219
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.015887025595763e-07,
+      "loss": 1.9893,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.009267431597529e-07,
+      "loss": 2.285,
+      "step": 3221
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.002647837599294e-07,
+      "loss": 3.0509,
+      "step": 3222
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.99602824360106e-07,
+      "loss": 1.3071,
+      "step": 3223
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.989408649602824e-07,
+      "loss": 0.2333,
+      "step": 3224
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.98278905560459e-07,
+      "loss": 0.805,
+      "step": 3225
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.976169461606355e-07,
+      "loss": 0.5399,
+      "step": 3226
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.969549867608119e-07,
+      "loss": 0.5471,
+      "step": 3227
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.962930273609885e-07,
+      "loss": 1.0579,
+      "step": 3228
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.95631067961165e-07,
+      "loss": 1.1227,
+      "step": 3229
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.949691085613416e-07,
+      "loss": 0.981,
+      "step": 3230
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.943071491615181e-07,
+      "loss": 1.1602,
+      "step": 3231
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.936451897616947e-07,
+      "loss": 1.1763,
+      "step": 3232
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.929832303618711e-07,
+      "loss": 1.1517,
+      "step": 3233
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.923212709620477e-07,
+      "loss": 1.4741,
+      "step": 3234
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.916593115622242e-07,
+      "loss": 1.3059,
+      "step": 3235
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.909973521624008e-07,
+      "loss": 0.9102,
+      "step": 3236
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.903353927625772e-07,
+      "loss": 0.4148,
+      "step": 3237
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.896734333627537e-07,
+      "loss": 0.4178,
+      "step": 3238
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.890114739629303e-07,
+      "loss": 0.1441,
+      "step": 3239
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.883495145631067e-07,
+      "loss": 0.4019,
+      "step": 3240
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.876875551632833e-07,
+      "loss": 1.5273,
+      "step": 3241
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.870255957634598e-07,
+      "loss": 0.1107,
+      "step": 3242
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.863636363636364e-07,
+      "loss": 0.294,
+      "step": 3243
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.857016769638128e-07,
+      "loss": 0.5314,
+      "step": 3244
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.850397175639894e-07,
+      "loss": 1.3845,
+      "step": 3245
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.843777581641659e-07,
+      "loss": 1.0819,
+      "step": 3246
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.837157987643426e-07,
+      "loss": 1.0604,
+      "step": 3247
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.83053839364519e-07,
+      "loss": 0.7134,
+      "step": 3248
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.823918799646954e-07,
+      "loss": 1.6057,
+      "step": 3249
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.817299205648721e-07,
+      "loss": 0.0404,
+      "step": 3250
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.810679611650486e-07,
+      "loss": 0.161,
+      "step": 3251
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.804060017652252e-07,
+      "loss": 0.3977,
+      "step": 3252
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.797440423654016e-07,
+      "loss": 1.9399,
+      "step": 3253
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.790820829655782e-07,
+      "loss": 0.0825,
+      "step": 3254
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.784201235657547e-07,
+      "loss": 0.1376,
+      "step": 3255
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.777581641659313e-07,
+      "loss": 1.075,
+      "step": 3256
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.770962047661077e-07,
+      "loss": 0.6803,
+      "step": 3257
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.764342453662843e-07,
+      "loss": 1.1609,
+      "step": 3258
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.757722859664608e-07,
+      "loss": 0.9025,
+      "step": 3259
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.751103265666372e-07,
+      "loss": 0.6689,
+      "step": 3260
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.744483671668138e-07,
+      "loss": 1.7713,
+      "step": 3261
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.737864077669903e-07,
+      "loss": 1.7161,
+      "step": 3262
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.731244483671669e-07,
+      "loss": 1.1139,
+      "step": 3263
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.724624889673433e-07,
+      "loss": 2.167,
+      "step": 3264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.718005295675199e-07,
+      "loss": 1.7673,
+      "step": 3265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.711385701676964e-07,
+      "loss": 0.0493,
+      "step": 3266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.70476610767873e-07,
+      "loss": 0.982,
+      "step": 3267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.698146513680495e-07,
+      "loss": 0.7817,
+      "step": 3268
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.691526919682259e-07,
+      "loss": 0.3692,
+      "step": 3269
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.684907325684025e-07,
+      "loss": 0.6802,
+      "step": 3270
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.67828773168579e-07,
+      "loss": 0.0935,
+      "step": 3271
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.671668137687556e-07,
+      "loss": 1.1649,
+      "step": 3272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.66504854368932e-07,
+      "loss": 0.2433,
+      "step": 3273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.658428949691086e-07,
+      "loss": 0.6803,
+      "step": 3274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.651809355692851e-07,
+      "loss": 1.8136,
+      "step": 3275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.645189761694617e-07,
+      "loss": 1.4382,
+      "step": 3276
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.638570167696381e-07,
+      "loss": 0.0236,
+      "step": 3277
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.631950573698147e-07,
+      "loss": 0.6046,
+      "step": 3278
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.625330979699912e-07,
+      "loss": 0.6853,
+      "step": 3279
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.618711385701676e-07,
+      "loss": 1.3241,
+      "step": 3280
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.612091791703442e-07,
+      "loss": 1.3433,
+      "step": 3281
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.605472197705207e-07,
+      "loss": 1.3242,
+      "step": 3282
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.598852603706973e-07,
+      "loss": 0.1007,
+      "step": 3283
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.592233009708737e-07,
+      "loss": 0.4827,
+      "step": 3284
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.585613415710504e-07,
+      "loss": 1.0857,
+      "step": 3285
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.578993821712268e-07,
+      "loss": 0.6694,
+      "step": 3286
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.572374227714034e-07,
+      "loss": 1.1601,
+      "step": 3287
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.565754633715799e-07,
+      "loss": 0.4159,
+      "step": 3288
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.559135039717565e-07,
+      "loss": 2.8309,
+      "step": 3289
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.552515445719329e-07,
+      "loss": 1.1867,
+      "step": 3290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.545895851721094e-07,
+      "loss": 0.1291,
+      "step": 3291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.53927625772286e-07,
+      "loss": 0.7901,
+      "step": 3292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.532656663724624e-07,
+      "loss": 1.0871,
+      "step": 3293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.52603706972639e-07,
+      "loss": 1.7745,
+      "step": 3294
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.519417475728155e-07,
+      "loss": 0.0581,
+      "step": 3295
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.512797881729921e-07,
+      "loss": 0.4959,
+      "step": 3296
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.506178287731685e-07,
+      "loss": 0.5561,
+      "step": 3297
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.499558693733451e-07,
+      "loss": 1.0664,
+      "step": 3298
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.492939099735216e-07,
+      "loss": 1.678,
+      "step": 3299
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.486319505736982e-07,
+      "loss": 0.1089,
+      "step": 3300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.479699911738746e-07,
+      "loss": 0.4987,
+      "step": 3301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.473080317740511e-07,
+      "loss": 1.1897,
+      "step": 3302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.466460723742278e-07,
+      "loss": 0.1503,
+      "step": 3303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.459841129744043e-07,
+      "loss": 0.6468,
+      "step": 3304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.453221535745809e-07,
+      "loss": 0.6169,
+      "step": 3305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.446601941747573e-07,
+      "loss": 0.9039,
+      "step": 3306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.439982347749339e-07,
+      "loss": 3.0402,
+      "step": 3307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.433362753751104e-07,
+      "loss": 0.6123,
+      "step": 3308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.42674315975287e-07,
+      "loss": 3.2171,
+      "step": 3309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.420123565754634e-07,
+      "loss": 0.8973,
+      "step": 3310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.4135039717564e-07,
+      "loss": 0.7437,
+      "step": 3311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.406884377758165e-07,
+      "loss": 0.5253,
+      "step": 3312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.400264783759929e-07,
+      "loss": 0.9499,
+      "step": 3313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.393645189761695e-07,
+      "loss": 0.2829,
+      "step": 3314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.38702559576346e-07,
+      "loss": 2.5861,
+      "step": 3315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.380406001765226e-07,
+      "loss": 2.564,
+      "step": 3316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.37378640776699e-07,
+      "loss": 0.2352,
+      "step": 3317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.367166813768756e-07,
+      "loss": 0.7341,
+      "step": 3318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.360547219770521e-07,
+      "loss": 3.2955,
+      "step": 3319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.353927625772287e-07,
+      "loss": 1.3504,
+      "step": 3320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.347308031774051e-07,
+      "loss": 0.9679,
+      "step": 3321
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.340688437775816e-07,
+      "loss": 0.143,
+      "step": 3322
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.334068843777582e-07,
+      "loss": 1.2443,
+      "step": 3323
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.327449249779347e-07,
+      "loss": 1.4066,
+      "step": 3324
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.320829655781113e-07,
+      "loss": 0.3951,
+      "step": 3325
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.314210061782877e-07,
+      "loss": 0.2719,
+      "step": 3326
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.307590467784643e-07,
+      "loss": 0.5365,
+      "step": 3327
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.300970873786408e-07,
+      "loss": 0.0957,
+      "step": 3328
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.294351279788174e-07,
+      "loss": 0.5128,
+      "step": 3329
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.287731685789938e-07,
+      "loss": 0.5629,
+      "step": 3330
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.281112091791704e-07,
+      "loss": 1.9819,
+      "step": 3331
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.274492497793469e-07,
+      "loss": 1.5322,
+      "step": 3332
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.267872903795233e-07,
+      "loss": 0.3482,
+      "step": 3333
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.261253309796999e-07,
+      "loss": 0.9801,
+      "step": 3334
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.254633715798764e-07,
+      "loss": 0.9575,
+      "step": 3335
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.24801412180053e-07,
+      "loss": 0.2,
+      "step": 3336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.241394527802294e-07,
+      "loss": 0.4643,
+      "step": 3337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.23477493380406e-07,
+      "loss": 1.7277,
+      "step": 3338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.228155339805825e-07,
+      "loss": 0.9495,
+      "step": 3339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.221535745807591e-07,
+      "loss": 1.3442,
+      "step": 3340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.214916151809356e-07,
+      "loss": 0.7166,
+      "step": 3341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.208296557811122e-07,
+      "loss": 0.2674,
+      "step": 3342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.201676963812886e-07,
+      "loss": 0.2178,
+      "step": 3343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.195057369814651e-07,
+      "loss": 0.3799,
+      "step": 3344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.188437775816417e-07,
+      "loss": 0.0894,
+      "step": 3345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.181818181818181e-07,
+      "loss": 0.7408,
+      "step": 3346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.175198587819947e-07,
+      "loss": 2.0048,
+      "step": 3347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.168578993821712e-07,
+      "loss": 0.7029,
+      "step": 3348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.161959399823478e-07,
+      "loss": 0.9779,
+      "step": 3349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.155339805825242e-07,
+      "loss": 0.7646,
+      "step": 3350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.148720211827008e-07,
+      "loss": 1.7443,
+      "step": 3351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.142100617828773e-07,
+      "loss": 0.3614,
+      "step": 3352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.13548102383054e-07,
+      "loss": 0.7444,
+      "step": 3353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.128861429832303e-07,
+      "loss": 0.312,
+      "step": 3354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.122241835834068e-07,
+      "loss": 0.9663,
+      "step": 3355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.115622241835834e-07,
+      "loss": 0.6522,
+      "step": 3356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.109002647837598e-07,
+      "loss": 0.5662,
+      "step": 3357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.102383053839366e-07,
+      "loss": 0.6356,
+      "step": 3358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.09576345984113e-07,
+      "loss": 0.1186,
+      "step": 3359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.089143865842896e-07,
+      "loss": 0.6439,
+      "step": 3360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.082524271844661e-07,
+      "loss": 1.1796,
+      "step": 3361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.075904677846427e-07,
+      "loss": 1.784,
+      "step": 3362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.069285083848191e-07,
+      "loss": 1.8556,
+      "step": 3363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.062665489849957e-07,
+      "loss": 0.4966,
+      "step": 3364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.056045895851722e-07,
+      "loss": 0.1087,
+      "step": 3365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.049426301853486e-07,
+      "loss": 0.6409,
+      "step": 3366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.042806707855252e-07,
+      "loss": 0.6712,
+      "step": 3367
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.036187113857017e-07,
+      "loss": 1.2501,
+      "step": 3368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.029567519858783e-07,
+      "loss": 0.9423,
+      "step": 3369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.022947925860547e-07,
+      "loss": 1.1168,
+      "step": 3370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.016328331862313e-07,
+      "loss": 1.0008,
+      "step": 3371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.009708737864078e-07,
+      "loss": 3.0737,
+      "step": 3372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.003089143865844e-07,
+      "loss": 1.5612,
+      "step": 3373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.996469549867608e-07,
+      "loss": 2.21,
+      "step": 3374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.989849955869373e-07,
+      "loss": 1.3695,
+      "step": 3375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.983230361871139e-07,
+      "loss": 2.988,
+      "step": 3376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.976610767872904e-07,
+      "loss": 0.2841,
+      "step": 3377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.96999117387467e-07,
+      "loss": 0.9293,
+      "step": 3378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.963371579876434e-07,
+      "loss": 2.4423,
+      "step": 3379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.9567519858782e-07,
+      "loss": 0.885,
+      "step": 3380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.950132391879965e-07,
+      "loss": 0.1436,
+      "step": 3381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.943512797881731e-07,
+      "loss": 0.075,
+      "step": 3382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.936893203883495e-07,
+      "loss": 1.2595,
+      "step": 3383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.930273609885261e-07,
+      "loss": 2.2322,
+      "step": 3384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.923654015887026e-07,
+      "loss": 0.8104,
+      "step": 3385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.91703442188879e-07,
+      "loss": 0.1477,
+      "step": 3386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.910414827890556e-07,
+      "loss": 2.6833,
+      "step": 3387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.903795233892321e-07,
+      "loss": 1.2292,
+      "step": 3388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.897175639894087e-07,
+      "loss": 0.203,
+      "step": 3389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.890556045895851e-07,
+      "loss": 1.2387,
+      "step": 3390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.883936451897617e-07,
+      "loss": 0.5857,
+      "step": 3391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.877316857899382e-07,
+      "loss": 0.0657,
+      "step": 3392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.870697263901148e-07,
+      "loss": 0.9356,
+      "step": 3393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.864077669902912e-07,
+      "loss": 3.2322,
+      "step": 3394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.857458075904679e-07,
+      "loss": 1.0624,
+      "step": 3395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.850838481906443e-07,
+      "loss": 3.0599,
+      "step": 3396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.844218887908208e-07,
+      "loss": 0.9396,
+      "step": 3397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.837599293909974e-07,
+      "loss": 1.0458,
+      "step": 3398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.830979699911738e-07,
+      "loss": 1.3355,
+      "step": 3399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.824360105913504e-07,
+      "loss": 0.2811,
+      "step": 3400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.817740511915269e-07,
+      "loss": 1.0537,
+      "step": 3401
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.811120917917035e-07,
+      "loss": 1.704,
+      "step": 3402
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.804501323918799e-07,
+      "loss": 0.5448,
+      "step": 3403
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.797881729920565e-07,
+      "loss": 0.0258,
+      "step": 3404
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.79126213592233e-07,
+      "loss": 0.7478,
+      "step": 3405
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.784642541924096e-07,
+      "loss": 1.2064,
+      "step": 3406
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.77802294792586e-07,
+      "loss": 1.6699,
+      "step": 3407
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.771403353927625e-07,
+      "loss": 1.0762,
+      "step": 3408
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.764783759929391e-07,
+      "loss": 1.5856,
+      "step": 3409
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.758164165931155e-07,
+      "loss": 1.4385,
+      "step": 3410
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.751544571932921e-07,
+      "loss": 0.5601,
+      "step": 3411
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.744924977934686e-07,
+      "loss": 0.3089,
+      "step": 3412
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.738305383936453e-07,
+      "loss": 2.2261,
+      "step": 3413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.731685789938218e-07,
+      "loss": 1.2612,
+      "step": 3414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.725066195939984e-07,
+      "loss": 1.7206,
+      "step": 3415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.718446601941748e-07,
+      "loss": 0.808,
+      "step": 3416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.711827007943514e-07,
+      "loss": 2.2243,
+      "step": 3417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.705207413945279e-07,
+      "loss": 0.7551,
+      "step": 3418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.698587819947043e-07,
+      "loss": 0.8125,
+      "step": 3419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.691968225948809e-07,
+      "loss": 0.2954,
+      "step": 3420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.685348631950574e-07,
+      "loss": 0.3542,
+      "step": 3421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.67872903795234e-07,
+      "loss": 0.4867,
+      "step": 3422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.672109443954104e-07,
+      "loss": 0.1166,
+      "step": 3423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.66548984995587e-07,
+      "loss": 1.437,
+      "step": 3424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.658870255957635e-07,
+      "loss": 1.8795,
+      "step": 3425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.652250661959401e-07,
+      "loss": 0.9415,
+      "step": 3426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.645631067961165e-07,
+      "loss": 1.5526,
+      "step": 3427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.63901147396293e-07,
+      "loss": 1.1717,
+      "step": 3428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.632391879964696e-07,
+      "loss": 0.191,
+      "step": 3429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.62577228596646e-07,
+      "loss": 0.4674,
+      "step": 3430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.619152691968227e-07,
+      "loss": 0.5555,
+      "step": 3431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.612533097969991e-07,
+      "loss": 2.1182,
+      "step": 3432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.605913503971757e-07,
+      "loss": 0.48,
+      "step": 3433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.599293909973522e-07,
+      "loss": 1.7464,
+      "step": 3434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.592674315975288e-07,
+      "loss": 0.9806,
+      "step": 3435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.586054721977052e-07,
+      "loss": 1.8804,
+      "step": 3436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.579435127978818e-07,
+      "loss": 0.8354,
+      "step": 3437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.572815533980583e-07,
+      "loss": 1.1096,
+      "step": 3438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.566195939982347e-07,
+      "loss": 1.1127,
+      "step": 3439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.559576345984113e-07,
+      "loss": 1.1958,
+      "step": 3440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.552956751985878e-07,
+      "loss": 1.8085,
+      "step": 3441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.546337157987644e-07,
+      "loss": 1.1288,
+      "step": 3442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.539717563989408e-07,
+      "loss": 0.7993,
+      "step": 3443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.533097969991174e-07,
+      "loss": 0.9997,
+      "step": 3444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.526478375992939e-07,
+      "loss": 0.7948,
+      "step": 3445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.519858781994705e-07,
+      "loss": 0.64,
+      "step": 3446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.513239187996469e-07,
+      "loss": 0.9838,
+      "step": 3447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.506619593998236e-07,
+      "loss": 0.6074,
+      "step": 3448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.5e-07,
+      "loss": 0.2468,
+      "step": 3449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.493380406001765e-07,
+      "loss": 0.7893,
+      "step": 3450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.486760812003531e-07,
+      "loss": 0.4801,
+      "step": 3451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.480141218005296e-07,
+      "loss": 2.2013,
+      "step": 3452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.473521624007061e-07,
+      "loss": 0.5928,
+      "step": 3453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.466902030008826e-07,
+      "loss": 0.9378,
+      "step": 3454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.460282436010591e-07,
+      "loss": 1.706,
+      "step": 3455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.453662842012356e-07,
+      "loss": 2.0573,
+      "step": 3456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.447043248014121e-07,
+      "loss": 1.4544,
+      "step": 3457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.440423654015887e-07,
+      "loss": 1.0013,
+      "step": 3458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.433804060017652e-07,
+      "loss": 0.752,
+      "step": 3459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.427184466019417e-07,
+      "loss": 0.6819,
+      "step": 3460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.420564872021183e-07,
+      "loss": 0.1121,
+      "step": 3461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.413945278022948e-07,
+      "loss": 1.9678,
+      "step": 3462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.407325684024713e-07,
+      "loss": 1.4227,
+      "step": 3463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.400706090026478e-07,
+      "loss": 1.0303,
+      "step": 3464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.394086496028243e-07,
+      "loss": 1.9057,
+      "step": 3465
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.387466902030008e-07,
+      "loss": 0.7362,
+      "step": 3466
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.380847308031773e-07,
+      "loss": 2.3646,
+      "step": 3467
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.37422771403354e-07,
+      "loss": 0.7577,
+      "step": 3468
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.367608120035305e-07,
+      "loss": 1.2796,
+      "step": 3469
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.36098852603707e-07,
+      "loss": 2.5055,
+      "step": 3470
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.354368932038836e-07,
+      "loss": 2.7934,
+      "step": 3471
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.347749338040601e-07,
+      "loss": 1.3947,
+      "step": 3472
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.341129744042366e-07,
+      "loss": 1.1704,
+      "step": 3473
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.334510150044131e-07,
+      "loss": 2.1617,
+      "step": 3474
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.327890556045897e-07,
+      "loss": 1.429,
+      "step": 3475
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.321270962047661e-07,
+      "loss": 0.8878,
+      "step": 3476
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.314651368049427e-07,
+      "loss": 1.3508,
+      "step": 3477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.308031774051192e-07,
+      "loss": 1.74,
+      "step": 3478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.301412180052957e-07,
+      "loss": 0.5089,
+      "step": 3479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.294792586054722e-07,
+      "loss": 1.5184,
+      "step": 3480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.288172992056488e-07,
+      "loss": 0.2334,
+      "step": 3481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.281553398058253e-07,
+      "loss": 1.0127,
+      "step": 3482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.274933804060018e-07,
+      "loss": 0.8704,
+      "step": 3483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.268314210061783e-07,
+      "loss": 0.3168,
+      "step": 3484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.261694616063549e-07,
+      "loss": 0.5808,
+      "step": 3485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.255075022065314e-07,
+      "loss": 0.4711,
+      "step": 3486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.248455428067079e-07,
+      "loss": 4.5397,
+      "step": 3487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.241835834068844e-07,
+      "loss": 2.4275,
+      "step": 3488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.235216240070609e-07,
+      "loss": 0.2833,
+      "step": 3489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.228596646072374e-07,
+      "loss": 1.3352,
+      "step": 3490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.22197705207414e-07,
+      "loss": 2.2192,
+      "step": 3491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.215357458075905e-07,
+      "loss": 1.4418,
+      "step": 3492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.20873786407767e-07,
+      "loss": 1.3596,
+      "step": 3493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.202118270079435e-07,
+      "loss": 1.2565,
+      "step": 3494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.195498676081201e-07,
+      "loss": 1.0965,
+      "step": 3495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.188879082082966e-07,
+      "loss": 1.3528,
+      "step": 3496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.182259488084731e-07,
+      "loss": 1.2155,
+      "step": 3497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.175639894086496e-07,
+      "loss": 0.3767,
+      "step": 3498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.169020300088261e-07,
+      "loss": 0.5681,
+      "step": 3499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.162400706090026e-07,
+      "loss": 0.511,
+      "step": 3500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.155781112091792e-07,
+      "loss": 1.4131,
+      "step": 3501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.149161518093557e-07,
+      "loss": 2.2282,
+      "step": 3502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.142541924095322e-07,
+      "loss": 2.1686,
+      "step": 3503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.135922330097088e-07,
+      "loss": 0.4662,
+      "step": 3504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.129302736098853e-07,
+      "loss": 0.8544,
+      "step": 3505
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.122683142100618e-07,
+      "loss": 2.2051,
+      "step": 3506
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.116063548102383e-07,
+      "loss": 0.9675,
+      "step": 3507
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.109443954104148e-07,
+      "loss": 1.3712,
+      "step": 3508
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.102824360105913e-07,
+      "loss": 0.8923,
+      "step": 3509
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.096204766107678e-07,
+      "loss": 0.9165,
+      "step": 3510
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.089585172109444e-07,
+      "loss": 0.299,
+      "step": 3511
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.082965578111209e-07,
+      "loss": 1.8049,
+      "step": 3512
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.076345984112974e-07,
+      "loss": 0.096,
+      "step": 3513
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.06972639011474e-07,
+      "loss": 0.5328,
+      "step": 3514
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.063106796116505e-07,
+      "loss": 0.3439,
+      "step": 3515
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.05648720211827e-07,
+      "loss": 0.7213,
+      "step": 3516
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.049867608120035e-07,
+      "loss": 2.2335,
+      "step": 3517
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.0432480141218e-07,
+      "loss": 1.5411,
+      "step": 3518
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.036628420123565e-07,
+      "loss": 1.8862,
+      "step": 3519
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.03000882612533e-07,
+      "loss": 1.071,
+      "step": 3520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.023389232127097e-07,
+      "loss": 0.1423,
+      "step": 3521
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.016769638128862e-07,
+      "loss": 0.4976,
+      "step": 3522
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.010150044130627e-07,
+      "loss": 2.2426,
+      "step": 3523
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.003530450132393e-07,
+      "loss": 1.9912,
+      "step": 3524
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.996910856134158e-07,
+      "loss": 0.5245,
+      "step": 3525
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.990291262135923e-07,
+      "loss": 1.5419,
+      "step": 3526
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.983671668137688e-07,
+      "loss": 0.0873,
+      "step": 3527
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.977052074139454e-07,
+      "loss": 1.9136,
+      "step": 3528
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.970432480141218e-07,
+      "loss": 0.3704,
+      "step": 3529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.963812886142983e-07,
+      "loss": 0.211,
+      "step": 3530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.957193292144749e-07,
+      "loss": 1.1785,
+      "step": 3531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.950573698146514e-07,
+      "loss": 1.0475,
+      "step": 3532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.943954104148279e-07,
+      "loss": 0.1232,
+      "step": 3533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.937334510150045e-07,
+      "loss": 0.9899,
+      "step": 3534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.93071491615181e-07,
+      "loss": 0.5723,
+      "step": 3535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.924095322153575e-07,
+      "loss": 1.2576,
+      "step": 3536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.91747572815534e-07,
+      "loss": 0.6514,
+      "step": 3537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.910856134157106e-07,
+      "loss": 1.7699,
+      "step": 3538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.904236540158871e-07,
+      "loss": 1.8325,
+      "step": 3539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.897616946160635e-07,
+      "loss": 1.8668,
+      "step": 3540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.890997352162401e-07,
+      "loss": 1.2494,
+      "step": 3541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.884377758164166e-07,
+      "loss": 0.0666,
+      "step": 3542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.877758164165931e-07,
+      "loss": 0.3891,
+      "step": 3543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.871138570167697e-07,
+      "loss": 0.9718,
+      "step": 3544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.864518976169462e-07,
+      "loss": 0.2315,
+      "step": 3545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.857899382171227e-07,
+      "loss": 2.7917,
+      "step": 3546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.851279788172992e-07,
+      "loss": 1.1689,
+      "step": 3547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.844660194174758e-07,
+      "loss": 0.1999,
+      "step": 3548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.838040600176523e-07,
+      "loss": 0.4112,
+      "step": 3549
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.831421006178287e-07,
+      "loss": 1.5764,
+      "step": 3550
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 6.824801412180053e-07,
+      "loss": 0.5404,
+      "step": 3551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.818181818181818e-07,
+      "loss": 0.6678,
+      "step": 3552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.811562224183583e-07,
+      "loss": 1.389,
+      "step": 3553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.804942630185349e-07,
+      "loss": 0.107,
+      "step": 3554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.798323036187114e-07,
+      "loss": 0.485,
+      "step": 3555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.791703442188879e-07,
+      "loss": 0.4314,
+      "step": 3556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.785083848190644e-07,
+      "loss": 0.9884,
+      "step": 3557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.77846425419241e-07,
+      "loss": 1.7572,
+      "step": 3558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.771844660194175e-07,
+      "loss": 0.2353,
+      "step": 3559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.76522506619594e-07,
+      "loss": 2.5195,
+      "step": 3560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.758605472197705e-07,
+      "loss": 0.2221,
+      "step": 3561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.75198587819947e-07,
+      "loss": 0.8669,
+      "step": 3562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.745366284201235e-07,
+      "loss": 1.3225,
+      "step": 3563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.738746690203001e-07,
+      "loss": 0.5517,
+      "step": 3564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.732127096204766e-07,
+      "loss": 0.5861,
+      "step": 3565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.725507502206531e-07,
+      "loss": 0.7746,
+      "step": 3566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.718887908208296e-07,
+      "loss": 0.9146,
+      "step": 3567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.712268314210062e-07,
+      "loss": 0.8585,
+      "step": 3568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.705648720211827e-07,
+      "loss": 1.6955,
+      "step": 3569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.699029126213592e-07,
+      "loss": 1.0974,
+      "step": 3570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.692409532215357e-07,
+      "loss": 0.0826,
+      "step": 3571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.685789938217122e-07,
+      "loss": 0.4979,
+      "step": 3572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.679170344218887e-07,
+      "loss": 1.5227,
+      "step": 3573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.672550750220653e-07,
+      "loss": 0.774,
+      "step": 3574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.665931156222419e-07,
+      "loss": 0.8532,
+      "step": 3575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.659311562224184e-07,
+      "loss": 0.7326,
+      "step": 3576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.65269196822595e-07,
+      "loss": 2.9414,
+      "step": 3577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.646072374227715e-07,
+      "loss": 0.1018,
+      "step": 3578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.63945278022948e-07,
+      "loss": 1.4091,
+      "step": 3579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.632833186231245e-07,
+      "loss": 0.2761,
+      "step": 3580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.626213592233011e-07,
+      "loss": 1.2793,
+      "step": 3581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.619593998234775e-07,
+      "loss": 0.961,
+      "step": 3582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.61297440423654e-07,
+      "loss": 1.3,
+      "step": 3583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.606354810238306e-07,
+      "loss": 1.6091,
+      "step": 3584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.599735216240071e-07,
+      "loss": 2.0534,
+      "step": 3585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.593115622241836e-07,
+      "loss": 0.5701,
+      "step": 3586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.586496028243602e-07,
+      "loss": 0.5979,
+      "step": 3587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.579876434245367e-07,
+      "loss": 0.2583,
+      "step": 3588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.573256840247132e-07,
+      "loss": 1.0019,
+      "step": 3589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.566637246248897e-07,
+      "loss": 0.1327,
+      "step": 3590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.560017652250663e-07,
+      "loss": 1.9749,
+      "step": 3591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.553398058252427e-07,
+      "loss": 0.7131,
+      "step": 3592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.546778464254192e-07,
+      "loss": 0.6771,
+      "step": 3593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.540158870255958e-07,
+      "loss": 0.5966,
+      "step": 3594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.533539276257723e-07,
+      "loss": 0.5678,
+      "step": 3595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 6.526919682259488e-07,
+      "loss": 1.7645,
+      "step": 3596
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.520300088261254e-07,
+      "loss": 1.0676,
+      "step": 3597
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.513680494263019e-07,
+      "loss": 0.1218,
+      "step": 3598
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.507060900264784e-07,
+      "loss": 0.4616,
+      "step": 3599
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.500441306266549e-07,
+      "loss": 2.0455,
+      "step": 3600
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.493821712268315e-07,
+      "loss": 0.8101,
+      "step": 3601
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.48720211827008e-07,
+      "loss": 3.2163,
+      "step": 3602
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.480582524271844e-07,
+      "loss": 1.6276,
+      "step": 3603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.47396293027361e-07,
+      "loss": 1.3405,
+      "step": 3604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.467343336275375e-07,
+      "loss": 0.8972,
+      "step": 3605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.46072374227714e-07,
+      "loss": 0.6616,
+      "step": 3606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.454104148278906e-07,
+      "loss": 0.7795,
+      "step": 3607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.447484554280671e-07,
+      "loss": 0.8253,
+      "step": 3608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.440864960282436e-07,
+      "loss": 0.5088,
+      "step": 3609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.434245366284201e-07,
+      "loss": 0.8607,
+      "step": 3610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.427625772285967e-07,
+      "loss": 2.9174,
+      "step": 3611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.421006178287732e-07,
+      "loss": 2.3691,
+      "step": 3612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.414386584289497e-07,
+      "loss": 0.8871,
+      "step": 3613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.407766990291262e-07,
+      "loss": 1.6099,
+      "step": 3614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.401147396293027e-07,
+      "loss": 0.7803,
+      "step": 3615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.394527802294792e-07,
+      "loss": 0.26,
+      "step": 3616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.387908208296558e-07,
+      "loss": 2.0988,
+      "step": 3617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.381288614298323e-07,
+      "loss": 0.7022,
+      "step": 3618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.374669020300088e-07,
+      "loss": 2.3277,
+      "step": 3619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.368049426301853e-07,
+      "loss": 0.8856,
+      "step": 3620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.361429832303619e-07,
+      "loss": 1.3646,
+      "step": 3621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.354810238305384e-07,
+      "loss": 1.7455,
+      "step": 3622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.348190644307149e-07,
+      "loss": 0.3438,
+      "step": 3623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.341571050308914e-07,
+      "loss": 1.6285,
+      "step": 3624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.334951456310679e-07,
+      "loss": 0.7022,
+      "step": 3625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.328331862312444e-07,
+      "loss": 0.5749,
+      "step": 3626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.32171226831421e-07,
+      "loss": 1.8696,
+      "step": 3627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.315092674315975e-07,
+      "loss": 1.8661,
+      "step": 3628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.30847308031774e-07,
+      "loss": 2.0682,
+      "step": 3629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.301853486319505e-07,
+      "loss": 0.9419,
+      "step": 3630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.295233892321272e-07,
+      "loss": 0.2616,
+      "step": 3631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.288614298323037e-07,
+      "loss": 1.9282,
+      "step": 3632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.281994704324802e-07,
+      "loss": 0.6581,
+      "step": 3633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.275375110326568e-07,
+      "loss": 0.0637,
+      "step": 3634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.268755516328332e-07,
+      "loss": 1.4363,
+      "step": 3635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.262135922330097e-07,
+      "loss": 2.4259,
+      "step": 3636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.255516328331863e-07,
+      "loss": 0.5633,
+      "step": 3637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.248896734333628e-07,
+      "loss": 0.9804,
+      "step": 3638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.242277140335393e-07,
+      "loss": 1.0961,
+      "step": 3639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.235657546337158e-07,
+      "loss": 1.1792,
+      "step": 3640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.229037952338924e-07,
+      "loss": 1.2367,
+      "step": 3641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 6.222418358340689e-07,
+      "loss": 0.4877,
+      "step": 3642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.215798764342454e-07,
+      "loss": 1.4084,
+      "step": 3643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.20917917034422e-07,
+      "loss": 1.3185,
+      "step": 3644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.202559576345984e-07,
+      "loss": 0.9278,
+      "step": 3645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.195939982347749e-07,
+      "loss": 0.0311,
+      "step": 3646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.189320388349515e-07,
+      "loss": 2.9798,
+      "step": 3647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.18270079435128e-07,
+      "loss": 0.0492,
+      "step": 3648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.176081200353045e-07,
+      "loss": 2.1629,
+      "step": 3649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.16946160635481e-07,
+      "loss": 0.24,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.162842012356576e-07,
+      "loss": 1.0605,
+      "step": 3651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.156222418358341e-07,
+      "loss": 2.0718,
+      "step": 3652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.149602824360106e-07,
+      "loss": 1.3177,
+      "step": 3653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.142983230361872e-07,
+      "loss": 1.2637,
+      "step": 3654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.136363636363637e-07,
+      "loss": 1.3558,
+      "step": 3655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.129744042365401e-07,
+      "loss": 0.0465,
+      "step": 3656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.123124448367167e-07,
+      "loss": 1.0052,
+      "step": 3657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.116504854368932e-07,
+      "loss": 0.1108,
+      "step": 3658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.109885260370697e-07,
+      "loss": 0.1779,
+      "step": 3659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.103265666372463e-07,
+      "loss": 0.3807,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.096646072374228e-07,
+      "loss": 1.4158,
+      "step": 3661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.090026478375993e-07,
+      "loss": 0.9415,
+      "step": 3662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.083406884377758e-07,
+      "loss": 1.3862,
+      "step": 3663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.076787290379524e-07,
+      "loss": 0.4687,
+      "step": 3664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.070167696381289e-07,
+      "loss": 1.0822,
+      "step": 3665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.063548102383054e-07,
+      "loss": 1.2862,
+      "step": 3666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.056928508384819e-07,
+      "loss": 0.3119,
+      "step": 3667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.050308914386584e-07,
+      "loss": 2.6197,
+      "step": 3668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.043689320388349e-07,
+      "loss": 0.4884,
+      "step": 3669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.037069726390115e-07,
+      "loss": 0.6942,
+      "step": 3670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.03045013239188e-07,
+      "loss": 0.6881,
+      "step": 3671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.023830538393645e-07,
+      "loss": 1.3135,
+      "step": 3672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.01721094439541e-07,
+      "loss": 0.9102,
+      "step": 3673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.010591350397176e-07,
+      "loss": 0.3375,
+      "step": 3674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 6.003971756398941e-07,
+      "loss": 0.668,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.997352162400706e-07,
+      "loss": 2.1486,
+      "step": 3676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.990732568402471e-07,
+      "loss": 0.2467,
+      "step": 3677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.984112974404236e-07,
+      "loss": 1.4841,
+      "step": 3678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.977493380406001e-07,
+      "loss": 1.0362,
+      "step": 3679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.970873786407767e-07,
+      "loss": 1.1258,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.964254192409532e-07,
+      "loss": 1.6579,
+      "step": 3681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.957634598411297e-07,
+      "loss": 0.1674,
+      "step": 3682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.951015004413062e-07,
+      "loss": 1.1876,
+      "step": 3683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.944395410414829e-07,
+      "loss": 1.1065,
+      "step": 3684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.937775816416594e-07,
+      "loss": 0.6111,
+      "step": 3685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.931156222418359e-07,
+      "loss": 1.2517,
+      "step": 3686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.924536628420125e-07,
+      "loss": 1.2839,
+      "step": 3687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 5.917917034421889e-07,
+      "loss": 1.9447,
+      "step": 3688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.911297440423654e-07,
+      "loss": 0.528,
+      "step": 3689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.90467784642542e-07,
+      "loss": 0.2062,
+      "step": 3690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.898058252427185e-07,
+      "loss": 0.3783,
+      "step": 3691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.89143865842895e-07,
+      "loss": 2.0524,
+      "step": 3692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.884819064430715e-07,
+      "loss": 0.0525,
+      "step": 3693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.878199470432481e-07,
+      "loss": 1.4565,
+      "step": 3694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.871579876434246e-07,
+      "loss": 1.9453,
+      "step": 3695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.864960282436011e-07,
+      "loss": 1.6746,
+      "step": 3696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.858340688437777e-07,
+      "loss": 0.997,
+      "step": 3697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.851721094439541e-07,
+      "loss": 1.2008,
+      "step": 3698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.845101500441306e-07,
+      "loss": 0.9447,
+      "step": 3699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.838481906443072e-07,
+      "loss": 1.8709,
+      "step": 3700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.831862312444837e-07,
+      "loss": 0.3344,
+      "step": 3701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.825242718446602e-07,
+      "loss": 0.1974,
+      "step": 3702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.818623124448367e-07,
+      "loss": 0.2445,
+      "step": 3703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.812003530450133e-07,
+      "loss": 1.7732,
+      "step": 3704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.805383936451898e-07,
+      "loss": 2.0512,
+      "step": 3705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.798764342453663e-07,
+      "loss": 1.4278,
+      "step": 3706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.792144748455429e-07,
+      "loss": 0.3238,
+      "step": 3707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.785525154457194e-07,
+      "loss": 0.9092,
+      "step": 3708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.778905560458958e-07,
+      "loss": 0.0311,
+      "step": 3709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.772285966460724e-07,
+      "loss": 0.1589,
+      "step": 3710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.765666372462489e-07,
+      "loss": 0.7399,
+      "step": 3711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.759046778464254e-07,
+      "loss": 0.8563,
+      "step": 3712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.752427184466019e-07,
+      "loss": 1.0733,
+      "step": 3713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.745807590467785e-07,
+      "loss": 0.2489,
+      "step": 3714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.73918799646955e-07,
+      "loss": 0.8787,
+      "step": 3715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.732568402471315e-07,
+      "loss": 0.7628,
+      "step": 3716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.725948808473081e-07,
+      "loss": 1.6791,
+      "step": 3717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.719329214474846e-07,
+      "loss": 0.6672,
+      "step": 3718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.712709620476611e-07,
+      "loss": 2.0279,
+      "step": 3719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.706090026478376e-07,
+      "loss": 0.2323,
+      "step": 3720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.699470432480141e-07,
+      "loss": 1.3781,
+      "step": 3721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.692850838481906e-07,
+      "loss": 0.0624,
+      "step": 3722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.686231244483671e-07,
+      "loss": 1.4786,
+      "step": 3723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.679611650485437e-07,
+      "loss": 2.0863,
+      "step": 3724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.672992056487202e-07,
+      "loss": 0.1115,
+      "step": 3725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.666372462488967e-07,
+      "loss": 0.8348,
+      "step": 3726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.659752868490733e-07,
+      "loss": 0.1776,
+      "step": 3727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.653133274492498e-07,
+      "loss": 1.1667,
+      "step": 3728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.646513680494263e-07,
+      "loss": 0.636,
+      "step": 3729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.639894086496028e-07,
+      "loss": 1.2234,
+      "step": 3730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.633274492497793e-07,
+      "loss": 0.2836,
+      "step": 3731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.626654898499558e-07,
+      "loss": 0.9445,
+      "step": 3732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.620035304501324e-07,
+      "loss": 2.0655,
+      "step": 3733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 5.613415710503089e-07,
+      "loss": 1.6146,
+      "step": 3734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.606796116504854e-07,
+      "loss": 0.9525,
+      "step": 3735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.600176522506619e-07,
+      "loss": 1.6624,
+      "step": 3736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.593556928508386e-07,
+      "loss": 2.1066,
+      "step": 3737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.58693733451015e-07,
+      "loss": 0.1717,
+      "step": 3738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.580317740511916e-07,
+      "loss": 0.6215,
+      "step": 3739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.57369814651368e-07,
+      "loss": 0.4694,
+      "step": 3740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.567078552515446e-07,
+      "loss": 0.3054,
+      "step": 3741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.560458958517211e-07,
+      "loss": 1.075,
+      "step": 3742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.553839364518977e-07,
+      "loss": 0.0494,
+      "step": 3743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.547219770520742e-07,
+      "loss": 1.1786,
+      "step": 3744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.540600176522507e-07,
+      "loss": 0.774,
+      "step": 3745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.533980582524272e-07,
+      "loss": 1.1715,
+      "step": 3746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.527360988526038e-07,
+      "loss": 1.1356,
+      "step": 3747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.520741394527803e-07,
+      "loss": 0.1558,
+      "step": 3748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.514121800529568e-07,
+      "loss": 0.6266,
+      "step": 3749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.507502206531334e-07,
+      "loss": 1.3347,
+      "step": 3750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.500882612533098e-07,
+      "loss": 0.4315,
+      "step": 3751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.494263018534863e-07,
+      "loss": 1.3237,
+      "step": 3752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.487643424536629e-07,
+      "loss": 0.7665,
+      "step": 3753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.481023830538394e-07,
+      "loss": 2.1284,
+      "step": 3754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.474404236540159e-07,
+      "loss": 0.7228,
+      "step": 3755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.467784642541924e-07,
+      "loss": 0.4091,
+      "step": 3756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.46116504854369e-07,
+      "loss": 1.539,
+      "step": 3757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.454545454545455e-07,
+      "loss": 0.2073,
+      "step": 3758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.44792586054722e-07,
+      "loss": 1.1507,
+      "step": 3759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.441306266548986e-07,
+      "loss": 0.536,
+      "step": 3760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.434686672550751e-07,
+      "loss": 1.7116,
+      "step": 3761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.428067078552515e-07,
+      "loss": 3.6892,
+      "step": 3762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.421447484554281e-07,
+      "loss": 0.8787,
+      "step": 3763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.414827890556046e-07,
+      "loss": 1.0443,
+      "step": 3764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.408208296557811e-07,
+      "loss": 0.0242,
+      "step": 3765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.401588702559576e-07,
+      "loss": 2.5529,
+      "step": 3766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.394969108561342e-07,
+      "loss": 1.2577,
+      "step": 3767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.388349514563107e-07,
+      "loss": 0.8296,
+      "step": 3768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.381729920564872e-07,
+      "loss": 2.3079,
+      "step": 3769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.375110326566638e-07,
+      "loss": 0.6069,
+      "step": 3770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.368490732568403e-07,
+      "loss": 0.5689,
+      "step": 3771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.361871138570167e-07,
+      "loss": 0.7967,
+      "step": 3772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.355251544571933e-07,
+      "loss": 2.6583,
+      "step": 3773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.348631950573698e-07,
+      "loss": 0.5941,
+      "step": 3774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.342012356575463e-07,
+      "loss": 0.4206,
+      "step": 3775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.335392762577228e-07,
+      "loss": 0.6109,
+      "step": 3776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.328773168578994e-07,
+      "loss": 0.2963,
+      "step": 3777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.322153574580759e-07,
+      "loss": 0.3605,
+      "step": 3778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.315533980582524e-07,
+      "loss": 0.6898,
+      "step": 3779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 5.30891438658429e-07,
+      "loss": 1.3195,
+      "step": 3780
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.302294792586055e-07,
+      "loss": 0.2536,
+      "step": 3781
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.29567519858782e-07,
+      "loss": 1.9082,
+      "step": 3782
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.289055604589585e-07,
+      "loss": 1.1826,
+      "step": 3783
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.28243601059135e-07,
+      "loss": 0.2935,
+      "step": 3784
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.275816416593115e-07,
+      "loss": 0.4516,
+      "step": 3785
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.26919682259488e-07,
+      "loss": 1.4873,
+      "step": 3786
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.262577228596646e-07,
+      "loss": 1.3789,
+      "step": 3787
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.255957634598411e-07,
+      "loss": 0.9663,
+      "step": 3788
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.249338040600176e-07,
+      "loss": 0.4731,
+      "step": 3789
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.242718446601942e-07,
+      "loss": 1.0539,
+      "step": 3790
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.236098852603707e-07,
+      "loss": 1.459,
+      "step": 3791
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.229479258605472e-07,
+      "loss": 0.8897,
+      "step": 3792
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.222859664607238e-07,
+      "loss": 2.1709,
+      "step": 3793
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.216240070609003e-07,
+      "loss": 0.5178,
+      "step": 3794
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.209620476610768e-07,
+      "loss": 1.5831,
+      "step": 3795
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.203000882612533e-07,
+      "loss": 0.3917,
+      "step": 3796
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.196381288614299e-07,
+      "loss": 1.7297,
+      "step": 3797
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.189761694616064e-07,
+      "loss": 0.5545,
+      "step": 3798
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.183142100617829e-07,
+      "loss": 1.1499,
+      "step": 3799
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.176522506619595e-07,
+      "loss": 1.1576,
+      "step": 3800
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.16990291262136e-07,
+      "loss": 1.3369,
+      "step": 3801
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.163283318623125e-07,
+      "loss": 1.1468,
+      "step": 3802
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.15666372462489e-07,
+      "loss": 0.6181,
+      "step": 3803
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.150044130626655e-07,
+      "loss": 0.0678,
+      "step": 3804
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.14342453662842e-07,
+      "loss": 0.9942,
+      "step": 3805
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.136804942630186e-07,
+      "loss": 1.2324,
+      "step": 3806
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.130185348631951e-07,
+      "loss": 0.3923,
+      "step": 3807
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.123565754633716e-07,
+      "loss": 1.3967,
+      "step": 3808
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.116946160635481e-07,
+      "loss": 0.2946,
+      "step": 3809
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.110326566637247e-07,
+      "loss": 0.0904,
+      "step": 3810
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.103706972639012e-07,
+      "loss": 0.5443,
+      "step": 3811
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.097087378640777e-07,
+      "loss": 1.2368,
+      "step": 3812
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.090467784642542e-07,
+      "loss": 0.0503,
+      "step": 3813
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.083848190644308e-07,
+      "loss": 0.2406,
+      "step": 3814
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.077228596646072e-07,
+      "loss": 3.2109,
+      "step": 3815
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.070609002647838e-07,
+      "loss": 1.5843,
+      "step": 3816
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.063989408649603e-07,
+      "loss": 1.5647,
+      "step": 3817
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.057369814651368e-07,
+      "loss": 2.5017,
+      "step": 3818
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.050750220653133e-07,
+      "loss": 1.2147,
+      "step": 3819
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.044130626654899e-07,
+      "loss": 1.4537,
+      "step": 3820
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.037511032656664e-07,
+      "loss": 1.5299,
+      "step": 3821
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.030891438658429e-07,
+      "loss": 0.5187,
+      "step": 3822
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.024271844660194e-07,
+      "loss": 0.7837,
+      "step": 3823
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.01765225066196e-07,
+      "loss": 0.428,
+      "step": 3824
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 5.011032656663724e-07,
+      "loss": 1.7647,
+      "step": 3825
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.00441306266549e-07,
+      "loss": 0.8892,
+      "step": 3826
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.997793468667255e-07,
+      "loss": 1.1349,
+      "step": 3827
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.99117387466902e-07,
+      "loss": 0.6035,
+      "step": 3828
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.984554280670785e-07,
+      "loss": 0.4096,
+      "step": 3829
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.977934686672551e-07,
+      "loss": 0.204,
+      "step": 3830
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.971315092674316e-07,
+      "loss": 0.071,
+      "step": 3831
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.964695498676081e-07,
+      "loss": 2.0747,
+      "step": 3832
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.958075904677847e-07,
+      "loss": 1.7652,
+      "step": 3833
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.951456310679612e-07,
+      "loss": 1.1881,
+      "step": 3834
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.944836716681377e-07,
+      "loss": 0.95,
+      "step": 3835
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.938217122683142e-07,
+      "loss": 1.3865,
+      "step": 3836
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.931597528684907e-07,
+      "loss": 1.1647,
+      "step": 3837
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.924977934686672e-07,
+      "loss": 0.1039,
+      "step": 3838
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.918358340688437e-07,
+      "loss": 0.8201,
+      "step": 3839
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.911738746690203e-07,
+      "loss": 0.4812,
+      "step": 3840
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.905119152691968e-07,
+      "loss": 1.3015,
+      "step": 3841
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.898499558693733e-07,
+      "loss": 1.0258,
+      "step": 3842
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.891879964695499e-07,
+      "loss": 1.2566,
+      "step": 3843
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.885260370697264e-07,
+      "loss": 0.6461,
+      "step": 3844
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.87864077669903e-07,
+      "loss": 0.9991,
+      "step": 3845
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.872021182700794e-07,
+      "loss": 0.1834,
+      "step": 3846
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.86540158870256e-07,
+      "loss": 0.1746,
+      "step": 3847
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.858781994704324e-07,
+      "loss": 2.093,
+      "step": 3848
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.85216240070609e-07,
+      "loss": 0.1558,
+      "step": 3849
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.845542806707856e-07,
+      "loss": 0.7218,
+      "step": 3850
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.838923212709621e-07,
+      "loss": 0.6532,
+      "step": 3851
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.832303618711386e-07,
+      "loss": 1.45,
+      "step": 3852
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.825684024713152e-07,
+      "loss": 0.2631,
+      "step": 3853
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.819064430714917e-07,
+      "loss": 0.7955,
+      "step": 3854
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.812444836716682e-07,
+      "loss": 1.184,
+      "step": 3855
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.805825242718447e-07,
+      "loss": 1.3602,
+      "step": 3856
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.799205648720212e-07,
+      "loss": 0.9157,
+      "step": 3857
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.792586054721977e-07,
+      "loss": 0.9856,
+      "step": 3858
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.785966460723742e-07,
+      "loss": 2.1991,
+      "step": 3859
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.779346866725508e-07,
+      "loss": 0.9806,
+      "step": 3860
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.772727272727273e-07,
+      "loss": 2.6102,
+      "step": 3861
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7661076787290383e-07,
+      "loss": 1.9173,
+      "step": 3862
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7594880847308033e-07,
+      "loss": 0.6587,
+      "step": 3863
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.752868490732569e-07,
+      "loss": 2.2259,
+      "step": 3864
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.746248896734334e-07,
+      "loss": 0.8834,
+      "step": 3865
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7396293027360994e-07,
+      "loss": 0.9339,
+      "step": 3866
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7330097087378644e-07,
+      "loss": 0.6213,
+      "step": 3867
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.726390114739629e-07,
+      "loss": 1.4385,
+      "step": 3868
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.7197705207413944e-07,
+      "loss": 0.7812,
+      "step": 3869
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.71315092674316e-07,
+      "loss": 2.2329,
+      "step": 3870
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.706531332744925e-07,
+      "loss": 0.8361,
+      "step": 3871
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6999117387466905e-07,
+      "loss": 2.5767,
+      "step": 3872
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6932921447484555e-07,
+      "loss": 0.5955,
+      "step": 3873
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.686672550750221e-07,
+      "loss": 0.2334,
+      "step": 3874
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.680052956751986e-07,
+      "loss": 0.9074,
+      "step": 3875
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6734333627537516e-07,
+      "loss": 1.1643,
+      "step": 3876
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6668137687555166e-07,
+      "loss": 0.2866,
+      "step": 3877
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.660194174757281e-07,
+      "loss": 1.8881,
+      "step": 3878
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6535745807590466e-07,
+      "loss": 1.0405,
+      "step": 3879
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.646954986760812e-07,
+      "loss": 0.1402,
+      "step": 3880
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.640335392762577e-07,
+      "loss": 1.632,
+      "step": 3881
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6337157987643426e-07,
+      "loss": 0.7617,
+      "step": 3882
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6270962047661077e-07,
+      "loss": 2.6926,
+      "step": 3883
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.620476610767873e-07,
+      "loss": 2.6704,
+      "step": 3884
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.613857016769638e-07,
+      "loss": 0.3606,
+      "step": 3885
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.6072374227714037e-07,
+      "loss": 0.5554,
+      "step": 3886
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.600617828773169e-07,
+      "loss": 0.9888,
+      "step": 3887
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5939982347749343e-07,
+      "loss": 0.6047,
+      "step": 3888
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.587378640776699e-07,
+      "loss": 1.447,
+      "step": 3889
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5807590467784643e-07,
+      "loss": 0.8459,
+      "step": 3890
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5741394527802293e-07,
+      "loss": 2.0073,
+      "step": 3891
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.567519858781995e-07,
+      "loss": 1.6466,
+      "step": 3892
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.56090026478376e-07,
+      "loss": 2.4236,
+      "step": 3893
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5542806707855254e-07,
+      "loss": 0.7758,
+      "step": 3894
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5476610767872904e-07,
+      "loss": 1.042,
+      "step": 3895
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.541041482789056e-07,
+      "loss": 0.3923,
+      "step": 3896
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5344218887908214e-07,
+      "loss": 1.4703,
+      "step": 3897
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5278022947925865e-07,
+      "loss": 1.2887,
+      "step": 3898
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.521182700794352e-07,
+      "loss": 2.9382,
+      "step": 3899
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5145631067961165e-07,
+      "loss": 0.7184,
+      "step": 3900
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.5079435127978815e-07,
+      "loss": 1.3556,
+      "step": 3901
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.501323918799647e-07,
+      "loss": 0.0844,
+      "step": 3902
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.494704324801412e-07,
+      "loss": 1.3035,
+      "step": 3903
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4880847308031775e-07,
+      "loss": 1.2268,
+      "step": 3904
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4814651368049426e-07,
+      "loss": 1.8706,
+      "step": 3905
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.474845542806708e-07,
+      "loss": 1.2346,
+      "step": 3906
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4682259488084736e-07,
+      "loss": 1.379,
+      "step": 3907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4616063548102386e-07,
+      "loss": 1.6955,
+      "step": 3908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.454986760812004e-07,
+      "loss": 1.102,
+      "step": 3909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4483671668137686e-07,
+      "loss": 0.728,
+      "step": 3910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4417475728155337e-07,
+      "loss": 0.9744,
+      "step": 3911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.435127978817299e-07,
+      "loss": 0.635,
+      "step": 3912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.428508384819064e-07,
+      "loss": 0.066,
+      "step": 3913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4218887908208297e-07,
+      "loss": 0.3337,
+      "step": 3914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.415269196822595e-07,
+      "loss": 0.2714,
+      "step": 3915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.4086496028243603e-07,
+      "loss": 2.5456,
+      "step": 3916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.402030008826126e-07,
+      "loss": 2.2114,
+      "step": 3917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.395410414827891e-07,
+      "loss": 0.6971,
+      "step": 3918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3887908208296563e-07,
+      "loss": 0.8892,
+      "step": 3919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3821712268314214e-07,
+      "loss": 0.6156,
+      "step": 3920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.375551632833186e-07,
+      "loss": 0.7841,
+      "step": 3921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3689320388349514e-07,
+      "loss": 1.8485,
+      "step": 3922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3623124448367164e-07,
+      "loss": 1.3633,
+      "step": 3923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.355692850838482e-07,
+      "loss": 0.2031,
+      "step": 3924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3490732568402474e-07,
+      "loss": 1.8318,
+      "step": 3925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3424536628420125e-07,
+      "loss": 0.4041,
+      "step": 3926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.335834068843778e-07,
+      "loss": 1.5279,
+      "step": 3927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.329214474845543e-07,
+      "loss": 0.7883,
+      "step": 3928
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3225948808473085e-07,
+      "loss": 0.0598,
+      "step": 3929
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3159752868490735e-07,
+      "loss": 1.23,
+      "step": 3930
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.309355692850838e-07,
+      "loss": 0.6044,
+      "step": 3931
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.3027360988526035e-07,
+      "loss": 0.6999,
+      "step": 3932
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2961165048543686e-07,
+      "loss": 0.5532,
+      "step": 3933
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.289496910856134e-07,
+      "loss": 1.0542,
+      "step": 3934
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2828773168578996e-07,
+      "loss": 1.6695,
+      "step": 3935
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2762577228596646e-07,
+      "loss": 1.102,
+      "step": 3936
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.26963812886143e-07,
+      "loss": 0.2028,
+      "step": 3937
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.263018534863195e-07,
+      "loss": 1.0566,
+      "step": 3938
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2563989408649607e-07,
+      "loss": 2.0781,
+      "step": 3939
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2497793468667257e-07,
+      "loss": 1.2286,
+      "step": 3940
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.243159752868491e-07,
+      "loss": 2.4512,
+      "step": 3941
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2365401588702557e-07,
+      "loss": 0.5534,
+      "step": 3942
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.229920564872021e-07,
+      "loss": 1.0984,
+      "step": 3943
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2233009708737863e-07,
+      "loss": 2.124,
+      "step": 3944
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.216681376875552e-07,
+      "loss": 0.6068,
+      "step": 3945
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.210061782877317e-07,
+      "loss": 1.9725,
+      "step": 3946
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.2034421888790823e-07,
+      "loss": 1.5138,
+      "step": 3947
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1968225948808474e-07,
+      "loss": 0.9121,
+      "step": 3948
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.190203000882613e-07,
+      "loss": 0.1346,
+      "step": 3949
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.183583406884378e-07,
+      "loss": 0.1812,
+      "step": 3950
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1769638128861434e-07,
+      "loss": 0.8464,
+      "step": 3951
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.170344218887908e-07,
+      "loss": 1.9393,
+      "step": 3952
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1637246248896734e-07,
+      "loss": 1.3099,
+      "step": 3953
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1571050308914384e-07,
+      "loss": 0.4596,
+      "step": 3954
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.150485436893204e-07,
+      "loss": 1.2666,
+      "step": 3955
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.143865842894969e-07,
+      "loss": 1.8872,
+      "step": 3956
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1372462488967345e-07,
+      "loss": 0.1463,
+      "step": 3957
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1306266548984995e-07,
+      "loss": 1.4662,
+      "step": 3958
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.124007060900265e-07,
+      "loss": 0.8644,
+      "step": 3959
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.11738746690203e-07,
+      "loss": 1.6101,
+      "step": 3960
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.1107678729037956e-07,
+      "loss": 1.982,
+      "step": 3961
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.104148278905561e-07,
+      "loss": 2.2446,
+      "step": 3962
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.0975286849073256e-07,
+      "loss": 0.614,
+      "step": 3963
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0909090909090906e-07,
+      "loss": 0.7724,
+      "step": 3964
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.084289496910856e-07,
+      "loss": 0.2774,
+      "step": 3965
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.077669902912621e-07,
+      "loss": 0.489,
+      "step": 3966
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0710503089143867e-07,
+      "loss": 0.2602,
+      "step": 3967
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0644307149161517e-07,
+      "loss": 0.4634,
+      "step": 3968
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.057811120917917e-07,
+      "loss": 0.6926,
+      "step": 3969
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.051191526919683e-07,
+      "loss": 1.717,
+      "step": 3970
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.044571932921448e-07,
+      "loss": 3.1793,
+      "step": 3971
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0379523389232133e-07,
+      "loss": 0.7652,
+      "step": 3972
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0313327449249783e-07,
+      "loss": 3.8022,
+      "step": 3973
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.024713150926743e-07,
+      "loss": 1.8718,
+      "step": 3974
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0180935569285083e-07,
+      "loss": 1.249,
+      "step": 3975
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0114739629302733e-07,
+      "loss": 1.0647,
+      "step": 3976
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.004854368932039e-07,
+      "loss": 0.6051,
+      "step": 3977
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.998234774933804e-07,
+      "loss": 1.4599,
+      "step": 3978
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9916151809355694e-07,
+      "loss": 0.6534,
+      "step": 3979
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.984995586937335e-07,
+      "loss": 1.8405,
+      "step": 3980
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9783759929391e-07,
+      "loss": 1.4032,
+      "step": 3981
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9717563989408655e-07,
+      "loss": 0.2978,
+      "step": 3982
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9651368049426305e-07,
+      "loss": 0.726,
+      "step": 3983
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.958517210944395e-07,
+      "loss": 1.7584,
+      "step": 3984
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9518976169461605e-07,
+      "loss": 2.2894,
+      "step": 3985
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9452780229479255e-07,
+      "loss": 2.73,
+      "step": 3986
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.938658428949691e-07,
+      "loss": 0.4492,
+      "step": 3987
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.932038834951456e-07,
+      "loss": 0.8764,
+      "step": 3988
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9254192409532216e-07,
+      "loss": 1.2053,
+      "step": 3989
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.918799646954987e-07,
+      "loss": 2.4104,
+      "step": 3990
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.912180052956752e-07,
+      "loss": 1.2976,
+      "step": 3991
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9055604589585177e-07,
+      "loss": 1.0189,
+      "step": 3992
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8989408649602827e-07,
+      "loss": 1.3206,
+      "step": 3993
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.892321270962048e-07,
+      "loss": 1.3025,
+      "step": 3994
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8857016769638127e-07,
+      "loss": 1.6141,
+      "step": 3995
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8790820829655777e-07,
+      "loss": 1.9128,
+      "step": 3996
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.872462488967343e-07,
+      "loss": 0.6835,
+      "step": 3997
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.865842894969109e-07,
+      "loss": 0.9726,
+      "step": 3998
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.859223300970874e-07,
+      "loss": 1.4689,
+      "step": 3999
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8526037069726393e-07,
+      "loss": 0.4395,
+      "step": 4000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8459841129744043e-07,
+      "loss": 0.6767,
+      "step": 4001
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.83936451897617e-07,
+      "loss": 0.7838,
+      "step": 4002
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.832744924977935e-07,
+      "loss": 0.0594,
+      "step": 4003
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8261253309797004e-07,
+      "loss": 2.422,
+      "step": 4004
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.819505736981465e-07,
+      "loss": 0.0649,
+      "step": 4005
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.81288614298323e-07,
+      "loss": 1.4781,
+      "step": 4006
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.8062665489849954e-07,
+      "loss": 1.8495,
+      "step": 4007
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.799646954986761e-07,
+      "loss": 0.0703,
+      "step": 4008
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.793027360988526e-07,
+      "loss": 0.2761,
+      "step": 4009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7864077669902915e-07,
+      "loss": 2.1115,
+      "step": 4010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7797881729920565e-07,
+      "loss": 1.0079,
+      "step": 4011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.773168578993822e-07,
+      "loss": 1.7049,
+      "step": 4012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.766548984995587e-07,
+      "loss": 1.7239,
+      "step": 4013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7599293909973526e-07,
+      "loss": 0.9198,
+      "step": 4014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.753309796999118e-07,
+      "loss": 1.077,
+      "step": 4015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7466902030008826e-07,
+      "loss": 1.9739,
+      "step": 4016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.740070609002648e-07,
+      "loss": 2.5768,
+      "step": 4017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.733451015004413e-07,
+      "loss": 2.308,
+      "step": 4018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.726831421006178e-07,
+      "loss": 0.9663,
+      "step": 4019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7202118270079437e-07,
+      "loss": 0.1261,
+      "step": 4020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7135922330097087e-07,
+      "loss": 1.5659,
+      "step": 4021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.706972639011474e-07,
+      "loss": 1.7663,
+      "step": 4022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.700353045013239e-07,
+      "loss": 1.6697,
+      "step": 4023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.693733451015004e-07,
+      "loss": 1.4062,
+      "step": 4024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.68711385701677e-07,
+      "loss": 0.7609,
+      "step": 4025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.680494263018535e-07,
+      "loss": 1.5253,
+      "step": 4026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6738746690203003e-07,
+      "loss": 1.7232,
+      "step": 4027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6672550750220653e-07,
+      "loss": 0.8485,
+      "step": 4028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6606354810238303e-07,
+      "loss": 0.36,
+      "step": 4029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.654015887025596e-07,
+      "loss": 2.4979,
+      "step": 4030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.647396293027361e-07,
+      "loss": 0.3957,
+      "step": 4031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6407766990291264e-07,
+      "loss": 0.8945,
+      "step": 4032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6341571050308914e-07,
+      "loss": 0.6603,
+      "step": 4033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.627537511032657e-07,
+      "loss": 1.7622,
+      "step": 4034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.620917917034422e-07,
+      "loss": 0.6905,
+      "step": 4035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.614298323036187e-07,
+      "loss": 2.1984,
+      "step": 4036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6076787290379525e-07,
+      "loss": 1.0354,
+      "step": 4037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6010591350397175e-07,
+      "loss": 0.5732,
+      "step": 4038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.594439541041483e-07,
+      "loss": 2.1694,
+      "step": 4039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.587819947043248e-07,
+      "loss": 0.2533,
+      "step": 4040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.581200353045013e-07,
+      "loss": 3.2321,
+      "step": 4041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5745807590467786e-07,
+      "loss": 1.1028,
+      "step": 4042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.567961165048544e-07,
+      "loss": 1.3938,
+      "step": 4043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.561341571050309e-07,
+      "loss": 2.9778,
+      "step": 4044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.554721977052074e-07,
+      "loss": 0.2984,
+      "step": 4045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.548102383053839e-07,
+      "loss": 0.9657,
+      "step": 4046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5414827890556047e-07,
+      "loss": 2.4981,
+      "step": 4047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53486319505737e-07,
+      "loss": 0.3403,
+      "step": 4048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528243601059135e-07,
+      "loss": 1.1557,
+      "step": 4049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5216240070609e-07,
+      "loss": 0.5301,
+      "step": 4050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.515004413062665e-07,
+      "loss": 0.9428,
+      "step": 4051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.508384819064431e-07,
+      "loss": 0.7146,
+      "step": 4052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5017652250661963e-07,
+      "loss": 0.2675,
+      "step": 4053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4951456310679613e-07,
+      "loss": 0.9698,
+      "step": 4054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.488526037069727e-07,
+      "loss": 1.8829,
+      "step": 4055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4819064430714913e-07,
+      "loss": 0.2247,
+      "step": 4056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.475286849073257e-07,
+      "loss": 0.2943,
+      "step": 4057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4686672550750224e-07,
+      "loss": 0.3478,
+      "step": 4058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4620476610767874e-07,
+      "loss": 1.7855,
+      "step": 4059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.455428067078553e-07,
+      "loss": 0.1494,
+      "step": 4060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4488084730803174e-07,
+      "loss": 0.7663,
+      "step": 4061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.442188879082083e-07,
+      "loss": 1.0766,
+      "step": 4062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4355692850838485e-07,
+      "loss": 1.1486,
+      "step": 4063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4289496910856135e-07,
+      "loss": 2.5545,
+      "step": 4064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.422330097087379e-07,
+      "loss": 1.7328,
+      "step": 4065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4157105030891435e-07,
+      "loss": 1.7737,
+      "step": 4066
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.409090909090909e-07,
+      "loss": 1.9003,
+      "step": 4067
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4024713150926746e-07,
+      "loss": 1.6227,
+      "step": 4068
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3958517210944396e-07,
+      "loss": 0.0868,
+      "step": 4069
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.389232127096205e-07,
+      "loss": 1.3234,
+      "step": 4070
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.38261253309797e-07,
+      "loss": 2.7861,
+      "step": 4071
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.375992939099735e-07,
+      "loss": 0.5368,
+      "step": 4072
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3693733451015007e-07,
+      "loss": 0.581,
+      "step": 4073
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3627537511032657e-07,
+      "loss": 1.5351,
+      "step": 4074
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.356134157105031e-07,
+      "loss": 0.8896,
+      "step": 4075
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.349514563106796e-07,
+      "loss": 0.2702,
+      "step": 4076
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.342894969108561e-07,
+      "loss": 2.3948,
+      "step": 4077
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.336275375110327e-07,
+      "loss": 1.7237,
+      "step": 4078
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.329655781112092e-07,
+      "loss": 2.6971,
+      "step": 4079
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3230361871138573e-07,
+      "loss": 0.5627,
+      "step": 4080
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3164165931156223e-07,
+      "loss": 0.9472,
+      "step": 4081
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3097969991173873e-07,
+      "loss": 1.0466,
+      "step": 4082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.303177405119153e-07,
+      "loss": 3.6504,
+      "step": 4083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.296557811120918e-07,
+      "loss": 2.7878,
+      "step": 4084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2899382171226834e-07,
+      "loss": 1.4688,
+      "step": 4085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2833186231244484e-07,
+      "loss": 0.5797,
+      "step": 4086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2766990291262134e-07,
+      "loss": 0.2787,
+      "step": 4087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.270079435127979e-07,
+      "loss": 0.8634,
+      "step": 4088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.263459841129744e-07,
+      "loss": 0.6491,
+      "step": 4089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2568402471315095e-07,
+      "loss": 2.9725,
+      "step": 4090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2502206531332745e-07,
+      "loss": 2.0117,
+      "step": 4091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.24360105913504e-07,
+      "loss": 1.4007,
+      "step": 4092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.236981465136805e-07,
+      "loss": 1.0485,
+      "step": 4093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.23036187113857e-07,
+      "loss": 0.8021,
+      "step": 4094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2237422771403356e-07,
+      "loss": 2.1009,
+      "step": 4095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2171226831421006e-07,
+      "loss": 0.5596,
+      "step": 4096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.210503089143866e-07,
+      "loss": 0.3866,
+      "step": 4097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.203883495145631e-07,
+      "loss": 1.246,
+      "step": 4098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.197263901147396e-07,
+      "loss": 1.6793,
+      "step": 4099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1906443071491616e-07,
+      "loss": 0.3435,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1840247131509267e-07,
+      "loss": 0.7427,
+      "step": 4101
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.177405119152692e-07,
+      "loss": 0.5245,
+      "step": 4102
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.170785525154457e-07,
+      "loss": 0.5209,
+      "step": 4103
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.164165931156222e-07,
+      "loss": 1.2043,
+      "step": 4104
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.157546337157988e-07,
+      "loss": 0.5263,
+      "step": 4105
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.150926743159753e-07,
+      "loss": 1.4152,
+      "step": 4106
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1443071491615183e-07,
+      "loss": 0.7906,
+      "step": 4107
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.137687555163284e-07,
+      "loss": 0.4167,
+      "step": 4108
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1310679611650483e-07,
+      "loss": 1.2146,
+      "step": 4109
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.124448367166814e-07,
+      "loss": 1.4057,
+      "step": 4110
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.117828773168579e-07,
+      "loss": 1.1443,
+      "step": 4111
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.1112091791703444e-07,
+      "loss": 1.3488,
+      "step": 4112
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.10458958517211e-07,
+      "loss": 0.6456,
+      "step": 4113
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0979699911738744e-07,
+      "loss": 1.7281,
+      "step": 4114
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.09135039717564e-07,
+      "loss": 0.301,
+      "step": 4115
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.084730803177405e-07,
+      "loss": 0.85,
+      "step": 4116
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0781112091791705e-07,
+      "loss": 1.1564,
+      "step": 4117
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.071491615180936e-07,
+      "loss": 0.7383,
+      "step": 4118
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0648720211827005e-07,
+      "loss": 1.1587,
+      "step": 4119
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.058252427184466e-07,
+      "loss": 2.3156,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0516328331862315e-07,
+      "loss": 1.2044,
+      "step": 4121
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0450132391879966e-07,
+      "loss": 0.7561,
+      "step": 4122
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.038393645189762e-07,
+      "loss": 0.1651,
+      "step": 4123
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.031774051191527e-07,
+      "loss": 1.7241,
+      "step": 4124
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.025154457193292e-07,
+      "loss": 1.3149,
+      "step": 4125
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0185348631950576e-07,
+      "loss": 1.447,
+      "step": 4126
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.0119152691968226e-07,
+      "loss": 2.4119,
+      "step": 4127
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.005295675198588e-07,
+      "loss": 0.3275,
+      "step": 4128
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.998676081200353e-07,
+      "loss": 3.3991,
+      "step": 4129
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.992056487202118e-07,
+      "loss": 0.3275,
+      "step": 4130
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9854368932038837e-07,
+      "loss": 0.9681,
+      "step": 4131
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9788172992056487e-07,
+      "loss": 0.9811,
+      "step": 4132
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9721977052074143e-07,
+      "loss": 0.8111,
+      "step": 4133
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9655781112091793e-07,
+      "loss": 1.1282,
+      "step": 4134
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9589585172109443e-07,
+      "loss": 2.6049,
+      "step": 4135
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.95233892321271e-07,
+      "loss": 0.723,
+      "step": 4136
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.945719329214475e-07,
+      "loss": 0.6545,
+      "step": 4137
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9390997352162404e-07,
+      "loss": 0.867,
+      "step": 4138
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9324801412180054e-07,
+      "loss": 0.7772,
+      "step": 4139
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9258605472197704e-07,
+      "loss": 0.1077,
+      "step": 4140
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.919240953221536e-07,
+      "loss": 3.0716,
+      "step": 4141
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.912621359223301e-07,
+      "loss": 0.7007,
+      "step": 4142
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.9060017652250664e-07,
+      "loss": 1.2761,
+      "step": 4143
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8993821712268315e-07,
+      "loss": 2.1502,
+      "step": 4144
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.892762577228597e-07,
+      "loss": 1.7894,
+      "step": 4145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.886142983230362e-07,
+      "loss": 1.6121,
+      "step": 4146
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.879523389232127e-07,
+      "loss": 0.9365,
+      "step": 4147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8729037952338925e-07,
+      "loss": 1.1321,
+      "step": 4148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8662842012356575e-07,
+      "loss": 1.7191,
+      "step": 4149
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.859664607237423e-07,
+      "loss": 1.3698,
+      "step": 4150
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.853045013239188e-07,
+      "loss": 1.4865,
+      "step": 4151
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.846425419240953e-07,
+      "loss": 0.5771,
+      "step": 4152
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8398058252427186e-07,
+      "loss": 1.0255,
+      "step": 4153
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8331862312444836e-07,
+      "loss": 0.1679,
+      "step": 4154
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.826566637246249e-07,
+      "loss": 0.99,
+      "step": 4155
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.819947043248014e-07,
+      "loss": 0.1394,
+      "step": 4156
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.813327449249779e-07,
+      "loss": 1.4945,
+      "step": 4157
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8067078552515447e-07,
+      "loss": 0.7456,
+      "step": 4158
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.8000882612533097e-07,
+      "loss": 1.207,
+      "step": 4159
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.793468667255075e-07,
+      "loss": 0.8811,
+      "step": 4160
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.78684907325684e-07,
+      "loss": 1.4702,
+      "step": 4161
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7802294792586053e-07,
+      "loss": 0.5237,
+      "step": 4162
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.773609885260371e-07,
+      "loss": 0.4839,
+      "step": 4163
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.766990291262136e-07,
+      "loss": 0.0757,
+      "step": 4164
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7603706972639013e-07,
+      "loss": 0.6792,
+      "step": 4165
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.753751103265667e-07,
+      "loss": 0.674,
+      "step": 4166
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7471315092674314e-07,
+      "loss": 1.1894,
+      "step": 4167
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.740511915269197e-07,
+      "loss": 0.769,
+      "step": 4168
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.733892321270962e-07,
+      "loss": 1.8565,
+      "step": 4169
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7272727272727274e-07,
+      "loss": 0.7997,
+      "step": 4170
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.720653133274493e-07,
+      "loss": 1.1874,
+      "step": 4171
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.7140335392762575e-07,
+      "loss": 0.9804,
+      "step": 4172
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.707413945278023e-07,
+      "loss": 0.496,
+      "step": 4173
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.700794351279788e-07,
+      "loss": 1.0723,
+      "step": 4174
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6941747572815535e-07,
+      "loss": 2.9111,
+      "step": 4175
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.687555163283319e-07,
+      "loss": 0.1675,
+      "step": 4176
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6809355692850835e-07,
+      "loss": 0.3947,
+      "step": 4177
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.674315975286849e-07,
+      "loss": 1.0821,
+      "step": 4178
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.667696381288614e-07,
+      "loss": 1.2614,
+      "step": 4179
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6610767872903796e-07,
+      "loss": 0.9404,
+      "step": 4180
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.654457193292145e-07,
+      "loss": 0.4053,
+      "step": 4181
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.64783759929391e-07,
+      "loss": 2.1193,
+      "step": 4182
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.641218005295675e-07,
+      "loss": 0.6958,
+      "step": 4183
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.63459841129744e-07,
+      "loss": 1.2305,
+      "step": 4184
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.6279788172992057e-07,
+      "loss": 0.698,
+      "step": 4185
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.621359223300971e-07,
+      "loss": 1.0865,
+      "step": 4186
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.614739629302736e-07,
+      "loss": 1.2749,
+      "step": 4187
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.608120035304501e-07,
+      "loss": 1.1577,
+      "step": 4188
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.601500441306266e-07,
+      "loss": 0.0334,
+      "step": 4189
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.594880847308032e-07,
+      "loss": 0.8234,
+      "step": 4190
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5882612533097973e-07,
+      "loss": 0.3114,
+      "step": 4191
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.5816416593115623e-07,
+      "loss": 0.2027,
+      "step": 4192
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5750220653133273e-07,
+      "loss": 0.3702,
+      "step": 4193
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.568402471315093e-07,
+      "loss": 1.2388,
+      "step": 4194
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.561782877316858e-07,
+      "loss": 2.6305,
+      "step": 4195
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5551632833186234e-07,
+      "loss": 1.2968,
+      "step": 4196
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5485436893203884e-07,
+      "loss": 0.1256,
+      "step": 4197
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.541924095322154e-07,
+      "loss": 0.5973,
+      "step": 4198
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.535304501323919e-07,
+      "loss": 2.2236,
+      "step": 4199
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.528684907325684e-07,
+      "loss": 1.7837,
+      "step": 4200
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5220653133274495e-07,
+      "loss": 0.7553,
+      "step": 4201
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.5154457193292145e-07,
+      "loss": 0.4791,
+      "step": 4202
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.50882612533098e-07,
+      "loss": 0.8356,
+      "step": 4203
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.502206531332745e-07,
+      "loss": 1.6646,
+      "step": 4204
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.49558693733451e-07,
+      "loss": 0.9779,
+      "step": 4205
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4889673433362756e-07,
+      "loss": 1.1925,
+      "step": 4206
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4823477493380406e-07,
+      "loss": 0.2389,
+      "step": 4207
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.475728155339806e-07,
+      "loss": 1.1507,
+      "step": 4208
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.469108561341571e-07,
+      "loss": 2.1099,
+      "step": 4209
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.462488967343336e-07,
+      "loss": 1.8174,
+      "step": 4210
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4558693733451017e-07,
+      "loss": 1.5347,
+      "step": 4211
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4492497793468667e-07,
+      "loss": 1.2262,
+      "step": 4212
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.442630185348632e-07,
+      "loss": 2.5149,
+      "step": 4213
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.436010591350397e-07,
+      "loss": 1.1678,
+      "step": 4214
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.429390997352162e-07,
+      "loss": 0.0664,
+      "step": 4215
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.422771403353928e-07,
+      "loss": 0.7046,
+      "step": 4216
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.416151809355693e-07,
+      "loss": 0.5994,
+      "step": 4217
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4095322153574583e-07,
+      "loss": 2.2221,
+      "step": 4218
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.4029126213592233e-07,
+      "loss": 0.7053,
+      "step": 4219
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3962930273609883e-07,
+      "loss": 1.2192,
+      "step": 4220
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.389673433362754e-07,
+      "loss": 1.3612,
+      "step": 4221
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3830538393645191e-07,
+      "loss": 0.4435,
+      "step": 4222
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3764342453662844e-07,
+      "loss": 0.642,
+      "step": 4223
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3698146513680497e-07,
+      "loss": 2.7668,
+      "step": 4224
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3631950573698144e-07,
+      "loss": 0.5045,
+      "step": 4225
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.35657546337158e-07,
+      "loss": 0.5744,
+      "step": 4226
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3499558693733452e-07,
+      "loss": 0.4202,
+      "step": 4227
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3433362753751105e-07,
+      "loss": 1.5213,
+      "step": 4228
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3367166813768758e-07,
+      "loss": 1.9991,
+      "step": 4229
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3300970873786405e-07,
+      "loss": 1.6829,
+      "step": 4230
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.323477493380406e-07,
+      "loss": 1.9376,
+      "step": 4231
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3168578993821713e-07,
+      "loss": 0.855,
+      "step": 4232
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3102383053839366e-07,
+      "loss": 1.1519,
+      "step": 4233
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.3036187113857019e-07,
+      "loss": 0.6516,
+      "step": 4234
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2969991173874671e-07,
+      "loss": 0.0802,
+      "step": 4235
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2903795233892321e-07,
+      "loss": 0.4794,
+      "step": 4236
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2837599293909974e-07,
+      "loss": 0.8804,
+      "step": 4237
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.2771403353927627e-07,
+      "loss": 1.2455,
+      "step": 4238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.270520741394528e-07,
+      "loss": 0.1192,
+      "step": 4239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2639011473962932e-07,
+      "loss": 2.6958,
+      "step": 4240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2572815533980582e-07,
+      "loss": 2.0319,
+      "step": 4241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2506619593998235e-07,
+      "loss": 0.5192,
+      "step": 4242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2440423654015888e-07,
+      "loss": 0.6054,
+      "step": 4243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.237422771403354e-07,
+      "loss": 0.1722,
+      "step": 4244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2308031774051193e-07,
+      "loss": 0.7071,
+      "step": 4245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2241835834068843e-07,
+      "loss": 2.7614,
+      "step": 4246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2175639894086496e-07,
+      "loss": 0.9433,
+      "step": 4247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2109443954104149e-07,
+      "loss": 1.1369,
+      "step": 4248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2043248014121801e-07,
+      "loss": 1.553,
+      "step": 4249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1977052074139454e-07,
+      "loss": 0.4239,
+      "step": 4250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1910856134157107e-07,
+      "loss": 0.1583,
+      "step": 4251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1844660194174757e-07,
+      "loss": 1.8729,
+      "step": 4252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.177846425419241e-07,
+      "loss": 0.9694,
+      "step": 4253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1712268314210062e-07,
+      "loss": 0.6902,
+      "step": 4254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1646072374227715e-07,
+      "loss": 0.6347,
+      "step": 4255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1579876434245368e-07,
+      "loss": 0.8546,
+      "step": 4256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1513680494263018e-07,
+      "loss": 0.0441,
+      "step": 4257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.144748455428067e-07,
+      "loss": 0.2103,
+      "step": 4258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1381288614298323e-07,
+      "loss": 0.3209,
+      "step": 4259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1315092674315976e-07,
+      "loss": 0.62,
+      "step": 4260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1248896734333629e-07,
+      "loss": 2.6284,
+      "step": 4261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1182700794351279e-07,
+      "loss": 0.6041,
+      "step": 4262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1116504854368931e-07,
+      "loss": 1.9926,
+      "step": 4263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1050308914386584e-07,
+      "loss": 0.1317,
+      "step": 4264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0984112974404237e-07,
+      "loss": 1.4704,
+      "step": 4265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.091791703442189e-07,
+      "loss": 0.5356,
+      "step": 4266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.085172109443954e-07,
+      "loss": 0.7251,
+      "step": 4267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0785525154457192e-07,
+      "loss": 0.9902,
+      "step": 4268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0719329214474845e-07,
+      "loss": 2.7147,
+      "step": 4269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0653133274492498e-07,
+      "loss": 1.0985,
+      "step": 4270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.058693733451015e-07,
+      "loss": 2.1797,
+      "step": 4271
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0520741394527806e-07,
+      "loss": 0.9955,
+      "step": 4272
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0454545454545453e-07,
+      "loss": 1.5819,
+      "step": 4273
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0388349514563106e-07,
+      "loss": 1.0013,
+      "step": 4274
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0322153574580759e-07,
+      "loss": 1.2931,
+      "step": 4275
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0255957634598414e-07,
+      "loss": 0.8571,
+      "step": 4276
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0189761694616067e-07,
+      "loss": 0.2878,
+      "step": 4277
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0123565754633714e-07,
+      "loss": 1.0592,
+      "step": 4278
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.0057369814651367e-07,
+      "loss": 1.2426,
+      "step": 4279
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.999117387466902e-07,
+      "loss": 0.2545,
+      "step": 4280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9924977934686675e-07,
+      "loss": 0.5772,
+      "step": 4281
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9858781994704328e-07,
+      "loss": 1.6107,
+      "step": 4282
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9792586054721975e-07,
+      "loss": 0.6486,
+      "step": 4283
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9726390114739628e-07,
+      "loss": 1.6475,
+      "step": 4284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.966019417475728e-07,
+      "loss": 0.5294,
+      "step": 4285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9593998234774936e-07,
+      "loss": 0.6809,
+      "step": 4286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9527802294792588e-07,
+      "loss": 0.8537,
+      "step": 4287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.946160635481024e-07,
+      "loss": 0.3079,
+      "step": 4288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9395410414827889e-07,
+      "loss": 0.9761,
+      "step": 4289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9329214474845544e-07,
+      "loss": 0.9333,
+      "step": 4290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9263018534863197e-07,
+      "loss": 0.3245,
+      "step": 4291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.919682259488085e-07,
+      "loss": 2.6946,
+      "step": 4292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9130626654898502e-07,
+      "loss": 3.1366,
+      "step": 4293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.906443071491615e-07,
+      "loss": 0.7362,
+      "step": 4294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8998234774933805e-07,
+      "loss": 0.5465,
+      "step": 4295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8932038834951458e-07,
+      "loss": 1.8042,
+      "step": 4296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.886584289496911e-07,
+      "loss": 1.1065,
+      "step": 4297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8799646954986763e-07,
+      "loss": 2.0746,
+      "step": 4298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8733451015004413e-07,
+      "loss": 2.4885,
+      "step": 4299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8667255075022066e-07,
+      "loss": 0.7149,
+      "step": 4300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8601059135039718e-07,
+      "loss": 1.2316,
+      "step": 4301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.853486319505737e-07,
+      "loss": 0.8459,
+      "step": 4302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.846866725507502e-07,
+      "loss": 0.5736,
+      "step": 4303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8402471315092674e-07,
+      "loss": 0.5189,
+      "step": 4304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8336275375110327e-07,
+      "loss": 0.2312,
+      "step": 4305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.827007943512798e-07,
+      "loss": 0.9144,
+      "step": 4306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8203883495145632e-07,
+      "loss": 0.9686,
+      "step": 4307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8137687555163285e-07,
+      "loss": 1.561,
+      "step": 4308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8071491615180935e-07,
+      "loss": 0.3912,
+      "step": 4309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8005295675198587e-07,
+      "loss": 2.1404,
+      "step": 4310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.793909973521624e-07,
+      "loss": 1.1703,
+      "step": 4311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7872903795233893e-07,
+      "loss": 0.7885,
+      "step": 4312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7806707855251546e-07,
+      "loss": 0.4898,
+      "step": 4313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7740511915269196e-07,
+      "loss": 0.4055,
+      "step": 4314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.767431597528685e-07,
+      "loss": 0.1038,
+      "step": 4315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.76081200353045e-07,
+      "loss": 0.6272,
+      "step": 4316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7541924095322154e-07,
+      "loss": 1.2514,
+      "step": 4317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7475728155339807e-07,
+      "loss": 0.7365,
+      "step": 4318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7409532215357457e-07,
+      "loss": 3.026,
+      "step": 4319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7343336275375112e-07,
+      "loss": 0.7645,
+      "step": 4320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7277140335392765e-07,
+      "loss": 1.519,
+      "step": 4321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7210944395410415e-07,
+      "loss": 0.3157,
+      "step": 4322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7144748455428067e-07,
+      "loss": 0.9628,
+      "step": 4323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7078552515445717e-07,
+      "loss": 2.3726,
+      "step": 4324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7012356575463373e-07,
+      "loss": 1.1824,
+      "step": 4325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6946160635481026e-07,
+      "loss": 0.8136,
+      "step": 4326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6879964695498676e-07,
+      "loss": 1.087,
+      "step": 4327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6813768755516328e-07,
+      "loss": 1.1779,
+      "step": 4328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.674757281553398e-07,
+      "loss": 1.4344,
+      "step": 4329
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6681376875551634e-07,
+      "loss": 0.9078,
+      "step": 4330
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6615180935569286e-07,
+      "loss": 1.7308,
+      "step": 4331
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6548984995586937e-07,
+      "loss": 0.4875,
+      "step": 4332
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.648278905560459e-07,
+      "loss": 1.8051,
+      "step": 4333
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6416593115622242e-07,
+      "loss": 0.6127,
+      "step": 4334
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6350397175639895e-07,
+      "loss": 0.8808,
+      "step": 4335
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6284201235657547e-07,
+      "loss": 1.3206,
+      "step": 4336
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.62180052956752e-07,
+      "loss": 0.5143,
+      "step": 4337
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.615180935569285e-07,
+      "loss": 0.9691,
+      "step": 4338
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6085613415710503e-07,
+      "loss": 1.1273,
+      "step": 4339
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6019417475728156e-07,
+      "loss": 0.5152,
+      "step": 4340
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5953221535745808e-07,
+      "loss": 0.2383,
+      "step": 4341
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.588702559576346e-07,
+      "loss": 0.6493,
+      "step": 4342
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.582082965578111e-07,
+      "loss": 0.6844,
+      "step": 4343
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5754633715798764e-07,
+      "loss": 2.4172,
+      "step": 4344
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.568843777581642e-07,
+      "loss": 2.9455,
+      "step": 4345
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.562224183583407e-07,
+      "loss": 0.4469,
+      "step": 4346
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5556045895851722e-07,
+      "loss": 4.2178,
+      "step": 4347
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5489849955869372e-07,
+      "loss": 0.4111,
+      "step": 4348
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5423654015887025e-07,
+      "loss": 2.5591,
+      "step": 4349
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.535745807590468e-07,
+      "loss": 0.5706,
+      "step": 4350
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.529126213592233e-07,
+      "loss": 0.8537,
+      "step": 4351
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5225066195939983e-07,
+      "loss": 1.1545,
+      "step": 4352
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5158870255957635e-07,
+      "loss": 0.3998,
+      "step": 4353
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5092674315975288e-07,
+      "loss": 1.6502,
+      "step": 4354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.502647837599294e-07,
+      "loss": 1.7172,
+      "step": 4355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.496028243601059e-07,
+      "loss": 0.7272,
+      "step": 4356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4894086496028244e-07,
+      "loss": 0.6076,
+      "step": 4357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4827890556045896e-07,
+      "loss": 1.0667,
+      "step": 4358
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.476169461606355e-07,
+      "loss": 0.881,
+      "step": 4359
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4695498676081202e-07,
+      "loss": 0.4839,
+      "step": 4360
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4629302736098852e-07,
+      "loss": 1.019,
+      "step": 4361
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4563106796116505e-07,
+      "loss": 2.8778,
+      "step": 4362
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4496910856134157e-07,
+      "loss": 1.3559,
+      "step": 4363
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.443071491615181e-07,
+      "loss": 0.8984,
+      "step": 4364
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4364518976169463e-07,
+      "loss": 0.2879,
+      "step": 4365
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4298323036187115e-07,
+      "loss": 2.6009,
+      "step": 4366
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4232127096204765e-07,
+      "loss": 2.7448,
+      "step": 4367
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4165931156222418e-07,
+      "loss": 1.228,
+      "step": 4368
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.409973521624007e-07,
+      "loss": 2.7097,
+      "step": 4369
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4033539276257724e-07,
+      "loss": 0.4972,
+      "step": 4370
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3967343336275376e-07,
+      "loss": 2.5052,
+      "step": 4371
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3901147396293026e-07,
+      "loss": 0.7035,
+      "step": 4372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.383495145631068e-07,
+      "loss": 0.9375,
+      "step": 4373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3768755516328334e-07,
+      "loss": 2.1446,
+      "step": 4374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3702559576345984e-07,
+      "loss": 1.5273,
+      "step": 4375
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3636363636363637e-07,
+      "loss": 2.7294,
+      "step": 4376
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3570167696381287e-07,
+      "loss": 1.8782,
+      "step": 4377
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.350397175639894e-07,
+      "loss": 1.1133,
+      "step": 4378
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3437775816416595e-07,
+      "loss": 0.3089,
+      "step": 4379
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3371579876434245e-07,
+      "loss": 4.0504,
+      "step": 4380
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3305383936451898e-07,
+      "loss": 0.1361,
+      "step": 4381
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.323918799646955e-07,
+      "loss": 0.3357,
+      "step": 4382
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.31729920564872e-07,
+      "loss": 0.3159,
+      "step": 4383
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3106796116504856e-07,
+      "loss": 0.7642,
+      "step": 4384
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.3040600176522506e-07,
+      "loss": 0.6774,
+      "step": 4385
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.297440423654016e-07,
+      "loss": 3.0899,
+      "step": 4386
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2908208296557812e-07,
+      "loss": 0.5639,
+      "step": 4387
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2842012356575464e-07,
+      "loss": 0.7493,
+      "step": 4388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2775816416593117e-07,
+      "loss": 0.7034,
+      "step": 4389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.270962047661077e-07,
+      "loss": 2.4495,
+      "step": 4390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.264342453662842e-07,
+      "loss": 2.4403,
+      "step": 4391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2577228596646073e-07,
+      "loss": 0.7328,
+      "step": 4392
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2511032656663725e-07,
+      "loss": 1.3661,
+      "step": 4393
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2444836716681378e-07,
+      "loss": 2.1654,
+      "step": 4394
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.237864077669903e-07,
+      "loss": 1.181,
+      "step": 4395
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.231244483671668e-07,
+      "loss": 0.6822,
+      "step": 4396
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2246248896734333e-07,
+      "loss": 1.4862,
+      "step": 4397
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2180052956751986e-07,
+      "loss": 2.1598,
+      "step": 4398
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.211385701676964e-07,
+      "loss": 0.5146,
+      "step": 4399
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.2047661076787292e-07,
+      "loss": 0.4872,
+      "step": 4400
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1981465136804942e-07,
+      "loss": 1.174,
+      "step": 4401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1915269196822596e-07,
+      "loss": 1.3841,
+      "step": 4402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1849073256840248e-07,
+      "loss": 0.4517,
+      "step": 4403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.17828773168579e-07,
+      "loss": 0.4619,
+      "step": 4404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1716681376875553e-07,
+      "loss": 0.1424,
+      "step": 4405
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1650485436893203e-07,
+      "loss": 0.9999,
+      "step": 4406
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1584289496910857e-07,
+      "loss": 1.8081,
+      "step": 4407
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1518093556928509e-07,
+      "loss": 0.7801,
+      "step": 4408
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1451897616946161e-07,
+      "loss": 0.8681,
+      "step": 4409
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1385701676963813e-07,
+      "loss": 0.3243,
+      "step": 4410
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1319505736981466e-07,
+      "loss": 0.5447,
+      "step": 4411
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1253309796999118e-07,
+      "loss": 0.6815,
+      "step": 4412
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.118711385701677e-07,
+      "loss": 1.0077,
+      "step": 4413
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1120917917034422e-07,
+      "loss": 0.049,
+      "step": 4414
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1054721977052074e-07,
+      "loss": 1.9083,
+      "step": 4415
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0988526037069727e-07,
+      "loss": 0.9803,
+      "step": 4416
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0922330097087378e-07,
+      "loss": 1.0146,
+      "step": 4417
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0856134157105031e-07,
+      "loss": 0.0355,
+      "step": 4418
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0789938217122684e-07,
+      "loss": 1.1522,
+      "step": 4419
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0723742277140335e-07,
+      "loss": 2.0594,
+      "step": 4420
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0657546337157988e-07,
+      "loss": 0.8356,
+      "step": 4421
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0591350397175639e-07,
+      "loss": 1.8986,
+      "step": 4422
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0525154457193292e-07,
+      "loss": 0.8682,
+      "step": 4423
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0458958517210945e-07,
+      "loss": 2.5032,
+      "step": 4424
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0392762577228596e-07,
+      "loss": 1.7367,
+      "step": 4425
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0326566637246249e-07,
+      "loss": 0.7344,
+      "step": 4426
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0260370697263903e-07,
+      "loss": 1.0793,
+      "step": 4427
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0194174757281553e-07,
+      "loss": 0.6449,
+      "step": 4428
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0127978817299207e-07,
+      "loss": 1.5129,
+      "step": 4429
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.0061782877316857e-07,
+      "loss": 0.5106,
+      "step": 4430
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.99558693733451e-08,
+      "loss": 0.745,
+      "step": 4431
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.929390997352164e-08,
+      "loss": 1.6619,
+      "step": 4432
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.863195057369814e-08,
+      "loss": 1.0356,
+      "step": 4433
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.796999117387468e-08,
+      "loss": 2.0991,
+      "step": 4434
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.73080317740512e-08,
+      "loss": 0.9875,
+      "step": 4435
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.664607237422772e-08,
+      "loss": 1.4075,
+      "step": 4436
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.598411297440425e-08,
+      "loss": 1.326,
+      "step": 4437
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.532215357458075e-08,
+      "loss": 0.1356,
+      "step": 4438
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.466019417475729e-08,
+      "loss": 0.5124,
+      "step": 4439
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.399823477493381e-08,
+      "loss": 1.9711,
+      "step": 4440
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.333627537511033e-08,
+      "loss": 0.146,
+      "step": 4441
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.267431597528686e-08,
+      "loss": 0.5061,
+      "step": 4442
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.201235657546337e-08,
+      "loss": 0.3908,
+      "step": 4443
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.13503971756399e-08,
+      "loss": 0.529,
+      "step": 4444
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.068843777581642e-08,
+      "loss": 0.4619,
+      "step": 4445
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.002647837599294e-08,
+      "loss": 0.7746,
+      "step": 4446
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.936451897616946e-08,
+      "loss": 1.4584,
+      "step": 4447
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.870255957634598e-08,
+      "loss": 1.1267,
+      "step": 4448
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.80406001765225e-08,
+      "loss": 1.8584,
+      "step": 4449
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.737864077669903e-08,
+      "loss": 1.6208,
+      "step": 4450
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.671668137687556e-08,
+      "loss": 0.062,
+      "step": 4451
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.605472197705207e-08,
+      "loss": 1.0169,
+      "step": 4452
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.539276257722859e-08,
+      "loss": 1.4918,
+      "step": 4453
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.473080317740513e-08,
+      "loss": 0.5752,
+      "step": 4454
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.406884377758164e-08,
+      "loss": 1.554,
+      "step": 4455
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.340688437775817e-08,
+      "loss": 1.2951,
+      "step": 4456
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.274492497793468e-08,
+      "loss": 0.2593,
+      "step": 4457
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.208296557811121e-08,
+      "loss": 1.2456,
+      "step": 4458
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.142100617828774e-08,
+      "loss": 0.8784,
+      "step": 4459
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.075904677846425e-08,
+      "loss": 0.9659,
+      "step": 4460
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.009708737864078e-08,
+      "loss": 1.1587,
+      "step": 4461
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.94351279788173e-08,
+      "loss": 0.5179,
+      "step": 4462
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.877316857899382e-08,
+      "loss": 1.3103,
+      "step": 4463
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.811120917917035e-08,
+      "loss": 0.5273,
+      "step": 4464
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.744924977934686e-08,
+      "loss": 1.1582,
+      "step": 4465
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.67872903795234e-08,
+      "loss": 0.1237,
+      "step": 4466
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.612533097969991e-08,
+      "loss": 2.9938,
+      "step": 4467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.546337157987644e-08,
+      "loss": 0.6405,
+      "step": 4468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.480141218005295e-08,
+      "loss": 2.0653,
+      "step": 4469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.413945278022948e-08,
+      "loss": 0.7656,
+      "step": 4470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.347749338040601e-08,
+      "loss": 0.5379,
+      "step": 4471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.281553398058252e-08,
+      "loss": 0.7572,
+      "step": 4472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.215357458075905e-08,
+      "loss": 2.0024,
+      "step": 4473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.149161518093558e-08,
+      "loss": 0.4885,
+      "step": 4474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.082965578111209e-08,
+      "loss": 2.0631,
+      "step": 4475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.016769638128862e-08,
+      "loss": 1.324,
+      "step": 4476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.950573698146513e-08,
+      "loss": 0.4016,
+      "step": 4477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.884377758164167e-08,
+      "loss": 2.0363,
+      "step": 4478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.818181818181819e-08,
+      "loss": 2.7577,
+      "step": 4479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.75198587819947e-08,
+      "loss": 1.8891,
+      "step": 4480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.685789938217123e-08,
+      "loss": 0.8588,
+      "step": 4481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.619593998234775e-08,
+      "loss": 1.4666,
+      "step": 4482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.553398058252428e-08,
+      "loss": 1.9601,
+      "step": 4483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.48720211827008e-08,
+      "loss": 2.6614,
+      "step": 4484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.421006178287732e-08,
+      "loss": 1.9321,
+      "step": 4485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.354810238305385e-08,
+      "loss": 0.2351,
+      "step": 4486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.288614298323036e-08,
+      "loss": 0.0865,
+      "step": 4487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.222418358340689e-08,
+      "loss": 0.9122,
+      "step": 4488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.15622241835834e-08,
+      "loss": 0.5662,
+      "step": 4489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.090026478375993e-08,
+      "loss": 1.0753,
+      "step": 4490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.023830538393646e-08,
+      "loss": 1.3893,
+      "step": 4491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.957634598411298e-08,
+      "loss": 0.923,
+      "step": 4492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.89143865842895e-08,
+      "loss": 0.5529,
+      "step": 4493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.825242718446601e-08,
+      "loss": 0.179,
+      "step": 4494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.7590467784642547e-08,
+      "loss": 0.8091,
+      "step": 4495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.692850838481907e-08,
+      "loss": 2.4461,
+      "step": 4496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.626654898499559e-08,
+      "loss": 1.2965,
+      "step": 4497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.560458958517211e-08,
+      "loss": 0.9293,
+      "step": 4498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4942630185348635e-08,
+      "loss": 1.4897,
+      "step": 4499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.4280670785525156e-08,
+      "loss": 0.7659,
+      "step": 4500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.3618711385701676e-08,
+      "loss": 0.2933,
+      "step": 4501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2956751985878197e-08,
+      "loss": 0.6523,
+      "step": 4502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.2294792586054724e-08,
+      "loss": 0.4103,
+      "step": 4503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.1632833186231244e-08,
+      "loss": 0.5141,
+      "step": 4504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0970873786407765e-08,
+      "loss": 1.8225,
+      "step": 4505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.0308914386584285e-08,
+      "loss": 1.0278,
+      "step": 4506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.964695498676082e-08,
+      "loss": 0.7572,
+      "step": 4507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.898499558693734e-08,
+      "loss": 0.9667,
+      "step": 4508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.832303618711386e-08,
+      "loss": 1.3372,
+      "step": 4509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.7661076787290374e-08,
+      "loss": 1.7623,
+      "step": 4510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.699911738746691e-08,
+      "loss": 1.2159,
+      "step": 4511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.633715798764343e-08,
+      "loss": 1.9319,
+      "step": 4512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.567519858781995e-08,
+      "loss": 0.8652,
+      "step": 4513
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.501323918799647e-08,
+      "loss": 0.5691,
+      "step": 4514
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.435127978817299e-08,
+      "loss": 0.9194,
+      "step": 4515
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3689320388349516e-08,
+      "loss": 1.4019,
+      "step": 4516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.302736098852604e-08,
+      "loss": 1.2834,
+      "step": 4517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2365401588702564e-08,
+      "loss": 1.2056,
+      "step": 4518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1703442188879084e-08,
+      "loss": 0.1506,
+      "step": 4519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1041482789055605e-08,
+      "loss": 1.4003,
+      "step": 4520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0379523389232125e-08,
+      "loss": 1.4358,
+      "step": 4521
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.971756398940865e-08,
+      "loss": 0.8942,
+      "step": 4522
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.905560458958517e-08,
+      "loss": 1.1667,
+      "step": 4523
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.83936451897617e-08,
+      "loss": 0.8446,
+      "step": 4524
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.773168578993822e-08,
+      "loss": 0.5079,
+      "step": 4525
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.706972639011474e-08,
+      "loss": 0.4894,
+      "step": 4526
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.640776699029126e-08,
+      "loss": 0.2482,
+      "step": 4527
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.574580759046779e-08,
+      "loss": 0.994,
+      "step": 4528
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.508384819064431e-08,
+      "loss": 1.4817,
+      "step": 4529
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4421888790820836e-08,
+      "loss": 2.3474,
+      "step": 4530
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.375992939099735e-08,
+      "loss": 1.9729,
+      "step": 4531
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.309796999117388e-08,
+      "loss": 2.2385,
+      "step": 4532
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.24360105913504e-08,
+      "loss": 0.4284,
+      "step": 4533
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1774051191526925e-08,
+      "loss": 1.3663,
+      "step": 4534
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1112091791703445e-08,
+      "loss": 2.9327,
+      "step": 4535
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0450132391879966e-08,
+      "loss": 1.2529,
+      "step": 4536
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.978817299205649e-08,
+      "loss": 1.1424,
+      "step": 4537
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9126213592233006e-08,
+      "loss": 1.0073,
+      "step": 4538
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8464254192409534e-08,
+      "loss": 0.6373,
+      "step": 4539
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7802294792586054e-08,
+      "loss": 2.5601,
+      "step": 4540
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7140335392762578e-08,
+      "loss": 0.8347,
+      "step": 4541
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6478375992939098e-08,
+      "loss": 0.5534,
+      "step": 4542
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5816416593115622e-08,
+      "loss": 0.5646,
+      "step": 4543
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5154457193292143e-08,
+      "loss": 1.7236,
+      "step": 4544
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.449249779346867e-08,
+      "loss": 1.7282,
+      "step": 4545
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3830538393645187e-08,
+      "loss": 0.571,
+      "step": 4546
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3168578993821714e-08,
+      "loss": 1.5619,
+      "step": 4547
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2506619593998234e-08,
+      "loss": 1.3946,
+      "step": 4548
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1844660194174758e-08,
+      "loss": 1.0872,
+      "step": 4549
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1182700794351282e-08,
+      "loss": 0.3038,
+      "step": 4550
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0520741394527802e-08,
+      "loss": 0.3308,
+      "step": 4551
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9858781994704326e-08,
+      "loss": 2.1442,
+      "step": 4552
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.919682259488085e-08,
+      "loss": 0.4981,
+      "step": 4553
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.853486319505737e-08,
+      "loss": 1.1284,
+      "step": 4554
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7872903795233894e-08,
+      "loss": 1.8334,
+      "step": 4555
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7210944395410418e-08,
+      "loss": 1.8604,
+      "step": 4556
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.654898499558694e-08,
+      "loss": 1.3527,
+      "step": 4557
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5887025595763462e-08,
+      "loss": 0.6707,
+      "step": 4558
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5225066195939983e-08,
+      "loss": 2.3716,
+      "step": 4559
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4563106796116503e-08,
+      "loss": 0.22,
+      "step": 4560
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3901147396293027e-08,
+      "loss": 0.0682,
+      "step": 4561
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3239187996469549e-08,
+      "loss": 0.4294,
+      "step": 4562
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2577228596646071e-08,
+      "loss": 4.7249,
+      "step": 4563
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1915269196822593e-08,
+      "loss": 2.0491,
+      "step": 4564
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1253309796999117e-08,
+      "loss": 3.0196,
+      "step": 4565
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0591350397175641e-08,
+      "loss": 0.7399,
+      "step": 4566
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.929390997352163e-09,
+      "loss": 0.1829,
+      "step": 4567
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.267431597528685e-09,
+      "loss": 1.4211,
+      "step": 4568
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.605472197705209e-09,
+      "loss": 0.9945,
+      "step": 4569
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.943512797881731e-09,
+      "loss": 1.1294,
+      "step": 4570
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.281553398058252e-09,
+      "loss": 0.1688,
+      "step": 4571
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.6195939982347746e-09,
+      "loss": 0.1012,
+      "step": 4572
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.957634598411297e-09,
+      "loss": 0.8709,
+      "step": 4573
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.2956751985878205e-09,
+      "loss": 1.711,
+      "step": 4574
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.633715798764343e-09,
+      "loss": 0.424,
+      "step": 4575
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9717563989408656e-09,
+      "loss": 0.487,
+      "step": 4576
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3097969991173873e-09,
+      "loss": 0.8599,
+      "step": 4577
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6478375992939102e-09,
+      "loss": 0.3661,
+      "step": 4578
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9858781994704328e-09,
+      "loss": 0.4797,
+      "step": 4579
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3239187996469551e-09,
+      "loss": 2.2528,
+      "step": 4580
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.619593998234776e-10,
+      "loss": 0.2098,
+      "step": 4581
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.0617,
+      "step": 4582
+    },
+    {
+      "epoch": 1.0,
+      "eval_accuracy": 0.7610474631751227,
+      "eval_loss": 0.9528456330299377,
+      "eval_runtime": 338.4539,
+      "eval_samples_per_second": 18.053,
+      "eval_steps_per_second": 4.515,
+      "step": 4582
+    }
+  ],
+  "max_steps": 4582,
+  "num_train_epochs": 1,
+  "total_flos": 6.320964198749562e+17,
+  "trial_name": null,
+  "trial_params": null
+}