diff --git "a/RTL-GeLM-7B/trainer_state.json" "b/RTL-GeLM-7B/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/RTL-GeLM-7B/trainer_state.json"
@@ -0,0 +1,15748 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "eval_steps": 500,
+  "global_step": 2620,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2658227848101266e-07,
+      "loss": 8.0625,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5316455696202533e-07,
+      "loss": 8.0547,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.79746835443038e-07,
+      "loss": 8.0625,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.063291139240507e-07,
+      "loss": 8.4531,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.329113924050634e-07,
+      "loss": 7.9844,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.59493670886076e-07,
+      "loss": 8.0781,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.860759493670887e-07,
+      "loss": 8.0938,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0126582278481013e-06,
+      "loss": 7.9688,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.139240506329114e-06,
+      "loss": 7.9531,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2658227848101267e-06,
+      "loss": 7.9453,
+      "step": 10
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3924050632911392e-06,
+      "loss": 7.9062,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.518987341772152e-06,
+      "loss": 7.5703,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6455696202531647e-06,
+      "loss": 7.5156,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7721518987341774e-06,
+      "loss": 7.3906,
+      "step": 14
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8987341772151901e-06,
+      "loss": 7.2109,
+      "step": 15
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0253164556962026e-06,
+      "loss": 7.0156,
+      "step": 16
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.1518987341772153e-06,
+      "loss": 6.7969,
+      "step": 17
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.278481012658228e-06,
+      "loss": 6.5703,
+      "step": 18
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4050632911392408e-06,
+      "loss": 6.3125,
+      "step": 19
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.5316455696202535e-06,
+      "loss": 6.1016,
+      "step": 20
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.6582278481012658e-06,
+      "loss": 6.0234,
+      "step": 21
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.7848101265822785e-06,
+      "loss": 5.9922,
+      "step": 22
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.9113924050632912e-06,
+      "loss": 5.6953,
+      "step": 23
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.037974683544304e-06,
+      "loss": 5.5781,
+      "step": 24
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.164556962025317e-06,
+      "loss": 5.5156,
+      "step": 25
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.2911392405063294e-06,
+      "loss": 5.3125,
+      "step": 26
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.417721518987342e-06,
+      "loss": 5.1953,
+      "step": 27
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.544303797468355e-06,
+      "loss": 5.0234,
+      "step": 28
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6708860759493675e-06,
+      "loss": 4.9766,
+      "step": 29
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.7974683544303802e-06,
+      "loss": 4.8672,
+      "step": 30
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.924050632911393e-06,
+      "loss": 4.8672,
+      "step": 31
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.050632911392405e-06,
+      "loss": 4.6406,
+      "step": 32
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.177215189873418e-06,
+      "loss": 4.5859,
+      "step": 33
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.303797468354431e-06,
+      "loss": 4.7031,
+      "step": 34
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.430379746835443e-06,
+      "loss": 4.3203,
+      "step": 35
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.556962025316456e-06,
+      "loss": 4.5703,
+      "step": 36
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.683544303797468e-06,
+      "loss": 4.6016,
+      "step": 37
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8101265822784815e-06,
+      "loss": 4.5391,
+      "step": 38
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.936708860759495e-06,
+      "loss": 4.3672,
+      "step": 39
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.063291139240507e-06,
+      "loss": 4.4297,
+      "step": 40
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.189873417721519e-06,
+      "loss": 4.3594,
+      "step": 41
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.3164556962025316e-06,
+      "loss": 4.1016,
+      "step": 42
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.443037974683545e-06,
+      "loss": 4.2266,
+      "step": 43
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.569620253164557e-06,
+      "loss": 4.1484,
+      "step": 44
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.69620253164557e-06,
+      "loss": 4.3398,
+      "step": 45
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.8227848101265824e-06,
+      "loss": 4.2344,
+      "step": 46
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.949367088607595e-06,
+      "loss": 4.2539,
+      "step": 47
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.075949367088608e-06,
+      "loss": 4.1641,
+      "step": 48
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.20253164556962e-06,
+      "loss": 4.3359,
+      "step": 49
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.329113924050634e-06,
+      "loss": 4.1172,
+      "step": 50
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.4556962025316464e-06,
+      "loss": 4.3125,
+      "step": 51
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.582278481012659e-06,
+      "loss": 4.1016,
+      "step": 52
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.708860759493672e-06,
+      "loss": 4.3906,
+      "step": 53
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.835443037974684e-06,
+      "loss": 4.082,
+      "step": 54
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.962025316455697e-06,
+      "loss": 4.2188,
+      "step": 55
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.08860759493671e-06,
+      "loss": 4.1953,
+      "step": 56
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.215189873417722e-06,
+      "loss": 4.1719,
+      "step": 57
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.341772151898735e-06,
+      "loss": 4.1953,
+      "step": 58
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.468354430379747e-06,
+      "loss": 4.3477,
+      "step": 59
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.5949367088607605e-06,
+      "loss": 4.1328,
+      "step": 60
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.721518987341773e-06,
+      "loss": 4.1406,
+      "step": 61
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.848101265822786e-06,
+      "loss": 4.1094,
+      "step": 62
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.974683544303799e-06,
+      "loss": 4.4297,
+      "step": 63
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.10126582278481e-06,
+      "loss": 4.1875,
+      "step": 64
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.227848101265824e-06,
+      "loss": 4.3438,
+      "step": 65
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.354430379746837e-06,
+      "loss": 4.25,
+      "step": 66
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.481012658227848e-06,
+      "loss": 4.0586,
+      "step": 67
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.607594936708861e-06,
+      "loss": 4.1797,
+      "step": 68
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.734177215189874e-06,
+      "loss": 4.1953,
+      "step": 69
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.860759493670886e-06,
+      "loss": 3.9336,
+      "step": 70
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 8.987341772151899e-06,
+      "loss": 4.1953,
+      "step": 71
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.113924050632912e-06,
+      "loss": 4.1797,
+      "step": 72
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.240506329113925e-06,
+      "loss": 4.0938,
+      "step": 73
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.367088607594937e-06,
+      "loss": 4.2109,
+      "step": 74
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.49367088607595e-06,
+      "loss": 3.9883,
+      "step": 75
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620253164556963e-06,
+      "loss": 4.1094,
+      "step": 76
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.746835443037975e-06,
+      "loss": 4.2383,
+      "step": 77
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.87341772151899e-06,
+      "loss": 4.1875,
+      "step": 78
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1e-05,
+      "loss": 4.3672,
+      "step": 79
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.999996178530757e-06,
+      "loss": 4.1172,
+      "step": 80
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.999984714128867e-06,
+      "loss": 4.2344,
+      "step": 81
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.999965606811854e-06,
+      "loss": 4.1367,
+      "step": 82
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.999938856608927e-06,
+      "loss": 3.918,
+      "step": 83
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.999904463560974e-06,
+      "loss": 4.2266,
+      "step": 84
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.99986242772057e-06,
+      "loss": 4.2109,
+      "step": 85
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.999812749151968e-06,
+      "loss": 3.8516,
+      "step": 86
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.999755427931108e-06,
+      "loss": 4.25,
+      "step": 87
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.99969046414561e-06,
+      "loss": 3.8398,
+      "step": 88
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.999617857894777e-06,
+      "loss": 4.1953,
+      "step": 89
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.999537609289592e-06,
+      "loss": 4.25,
+      "step": 90
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.999449718452724e-06,
+      "loss": 3.8867,
+      "step": 91
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.999354185518525e-06,
+      "loss": 3.9805,
+      "step": 92
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.99925101063302e-06,
+      "loss": 3.9883,
+      "step": 93
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.999140193953923e-06,
+      "loss": 4.1562,
+      "step": 94
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.999021735650626e-06,
+      "loss": 4.0859,
+      "step": 95
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.998895635904205e-06,
+      "loss": 4.1641,
+      "step": 96
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.998761894907414e-06,
+      "loss": 3.75,
+      "step": 97
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.998620512864686e-06,
+      "loss": 4.1406,
+      "step": 98
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.998471489992138e-06,
+      "loss": 3.8789,
+      "step": 99
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.998314826517564e-06,
+      "loss": 4.1641,
+      "step": 100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.998150522680437e-06,
+      "loss": 3.9688,
+      "step": 101
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.997978578731908e-06,
+      "loss": 3.8438,
+      "step": 102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.997798994934812e-06,
+      "loss": 4.0312,
+      "step": 103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.997611771563661e-06,
+      "loss": 3.8672,
+      "step": 104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.997416908904634e-06,
+      "loss": 3.8711,
+      "step": 105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.997214407255602e-06,
+      "loss": 4.1367,
+      "step": 106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.997004266926105e-06,
+      "loss": 4.0469,
+      "step": 107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.99678648823736e-06,
+      "loss": 3.9844,
+      "step": 108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.996561071522264e-06,
+      "loss": 3.8945,
+      "step": 109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.996328017125381e-06,
+      "loss": 4.1719,
+      "step": 110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.996087325402959e-06,
+      "loss": 3.7891,
+      "step": 111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.995838996722916e-06,
+      "loss": 3.8594,
+      "step": 112
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.995583031464841e-06,
+      "loss": 3.8125,
+      "step": 113
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.995319430020004e-06,
+      "loss": 4.1172,
+      "step": 114
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.995048192791341e-06,
+      "loss": 4.0195,
+      "step": 115
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.99476932019346e-06,
+      "loss": 4.0664,
+      "step": 116
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.994482812652645e-06,
+      "loss": 4.0938,
+      "step": 117
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.994188670606845e-06,
+      "loss": 4.0469,
+      "step": 118
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.993886894505687e-06,
+      "loss": 3.8047,
+      "step": 119
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.993577484810457e-06,
+      "loss": 4.0625,
+      "step": 120
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.993260441994116e-06,
+      "loss": 4.0039,
+      "step": 121
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.992935766541295e-06,
+      "loss": 3.7656,
+      "step": 122
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.992603458948282e-06,
+      "loss": 3.9766,
+      "step": 123
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.992263519723046e-06,
+      "loss": 3.7422,
+      "step": 124
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.99191594938521e-06,
+      "loss": 4.1953,
+      "step": 125
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.991560748466068e-06,
+      "loss": 4.0039,
+      "step": 126
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.99119791750857e-06,
+      "loss": 3.9453,
+      "step": 127
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.990827457067342e-06,
+      "loss": 3.875,
+      "step": 128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.990449367708662e-06,
+      "loss": 4.1016,
+      "step": 129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.990063650010473e-06,
+      "loss": 3.9727,
+      "step": 130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.989670304562378e-06,
+      "loss": 3.9922,
+      "step": 131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.98926933196564e-06,
+      "loss": 3.7188,
+      "step": 132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.988860732833183e-06,
+      "loss": 4.375,
+      "step": 133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.988444507789584e-06,
+      "loss": 3.6367,
+      "step": 134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.988020657471078e-06,
+      "loss": 3.8633,
+      "step": 135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.987589182525562e-06,
+      "loss": 3.8555,
+      "step": 136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.98715008361258e-06,
+      "loss": 3.9688,
+      "step": 137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.986703361403334e-06,
+      "loss": 3.8008,
+      "step": 138
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.98624901658068e-06,
+      "loss": 3.9062,
+      "step": 139
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.98578704983912e-06,
+      "loss": 3.7344,
+      "step": 140
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.985317461884815e-06,
+      "loss": 3.7617,
+      "step": 141
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.984840253435569e-06,
+      "loss": 3.9844,
+      "step": 142
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.984355425220835e-06,
+      "loss": 4.0859,
+      "step": 143
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.983862977981717e-06,
+      "loss": 3.75,
+      "step": 144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.983362912470967e-06,
+      "loss": 3.9492,
+      "step": 145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.982855229452976e-06,
+      "loss": 3.9375,
+      "step": 146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.982339929703782e-06,
+      "loss": 3.8516,
+      "step": 147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.981817014011066e-06,
+      "loss": 3.8477,
+      "step": 148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.98128648317415e-06,
+      "loss": 3.8516,
+      "step": 149
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.980748338003999e-06,
+      "loss": 3.8008,
+      "step": 150
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.980202579323212e-06,
+      "loss": 3.8867,
+      "step": 151
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.97964920796603e-06,
+      "loss": 3.918,
+      "step": 152
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.979088224778336e-06,
+      "loss": 3.5586,
+      "step": 153
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.978519630617632e-06,
+      "loss": 3.9609,
+      "step": 154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.977943426353067e-06,
+      "loss": 3.5508,
+      "step": 155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.977359612865424e-06,
+      "loss": 3.8711,
+      "step": 156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.976768191047109e-06,
+      "loss": 3.7422,
+      "step": 157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.976169161802164e-06,
+      "loss": 3.8477,
+      "step": 158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.975562526046257e-06,
+      "loss": 3.582,
+      "step": 159
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.974948284706684e-06,
+      "loss": 3.8164,
+      "step": 160
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.974326438722367e-06,
+      "loss": 3.7188,
+      "step": 161
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.973696989043852e-06,
+      "loss": 4.1992,
+      "step": 162
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.973059936633308e-06,
+      "loss": 3.8047,
+      "step": 163
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.972415282464524e-06,
+      "loss": 3.7305,
+      "step": 164
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.971763027522913e-06,
+      "loss": 3.8906,
+      "step": 165
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.971103172805504e-06,
+      "loss": 3.5859,
+      "step": 166
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.97043571932094e-06,
+      "loss": 3.9258,
+      "step": 167
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.969760668089485e-06,
+      "loss": 3.8047,
+      "step": 168
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.969078020143013e-06,
+      "loss": 3.9531,
+      "step": 169
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.968387776525009e-06,
+      "loss": 4.0352,
+      "step": 170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.967689938290574e-06,
+      "loss": 3.7422,
+      "step": 171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.966984506506415e-06,
+      "loss": 3.8125,
+      "step": 172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.966271482250845e-06,
+      "loss": 3.668,
+      "step": 173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.965550866613782e-06,
+      "loss": 3.7031,
+      "step": 174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.964822660696753e-06,
+      "loss": 3.668,
+      "step": 175
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.964086865612886e-06,
+      "loss": 3.9883,
+      "step": 176
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.963343482486907e-06,
+      "loss": 3.5977,
+      "step": 177
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.96259251245514e-06,
+      "loss": 3.7891,
+      "step": 178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.96183395666551e-06,
+      "loss": 3.8945,
+      "step": 179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.961067816277537e-06,
+      "loss": 3.6172,
+      "step": 180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.960294092462332e-06,
+      "loss": 3.4766,
+      "step": 181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.959512786402603e-06,
+      "loss": 3.7305,
+      "step": 182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.958723899292642e-06,
+      "loss": 3.7852,
+      "step": 183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.957927432338332e-06,
+      "loss": 3.7188,
+      "step": 184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.957123386757144e-06,
+      "loss": 3.7031,
+      "step": 185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.956311763778131e-06,
+      "loss": 3.5898,
+      "step": 186
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.95549256464193e-06,
+      "loss": 3.8477,
+      "step": 187
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.954665790600762e-06,
+      "loss": 3.4219,
+      "step": 188
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.953831442918418e-06,
+      "loss": 3.7852,
+      "step": 189
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.952989522870275e-06,
+      "loss": 3.7305,
+      "step": 190
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.952140031743282e-06,
+      "loss": 3.75,
+      "step": 191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.951282970835959e-06,
+      "loss": 3.6523,
+      "step": 192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.950418341458398e-06,
+      "loss": 3.5859,
+      "step": 193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.949546144932265e-06,
+      "loss": 3.707,
+      "step": 194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.948666382590785e-06,
+      "loss": 3.6094,
+      "step": 195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.947779055778752e-06,
+      "loss": 3.7227,
+      "step": 196
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.946884165852526e-06,
+      "loss": 3.7617,
+      "step": 197
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.945981714180021e-06,
+      "loss": 3.6797,
+      "step": 198
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.945071702140716e-06,
+      "loss": 3.4102,
+      "step": 199
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.944154131125643e-06,
+      "loss": 3.9375,
+      "step": 200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.943229002537391e-06,
+      "loss": 3.9297,
+      "step": 201
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.9422963177901e-06,
+      "loss": 3.8242,
+      "step": 202
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.941356078309457e-06,
+      "loss": 3.7305,
+      "step": 203
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.940408285532705e-06,
+      "loss": 3.6289,
+      "step": 204
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.939452940908627e-06,
+      "loss": 3.9258,
+      "step": 205
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.93849004589755e-06,
+      "loss": 3.8594,
+      "step": 206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.937519601971343e-06,
+      "loss": 3.625,
+      "step": 207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.936541610613417e-06,
+      "loss": 3.6602,
+      "step": 208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.935556073318718e-06,
+      "loss": 3.6562,
+      "step": 209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.934562991593723e-06,
+      "loss": 3.7539,
+      "step": 210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.933562366956445e-06,
+      "loss": 3.7578,
+      "step": 211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.932554200936428e-06,
+      "loss": 3.5078,
+      "step": 212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.931538495074744e-06,
+      "loss": 3.5352,
+      "step": 213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.930515250923985e-06,
+      "loss": 3.6289,
+      "step": 214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.92948447004827e-06,
+      "loss": 3.4375,
+      "step": 215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.928446154023238e-06,
+      "loss": 3.6797,
+      "step": 216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.927400304436048e-06,
+      "loss": 3.4609,
+      "step": 217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.92634692288537e-06,
+      "loss": 3.6094,
+      "step": 218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.925286010981394e-06,
+      "loss": 3.6562,
+      "step": 219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.924217570345814e-06,
+      "loss": 3.6094,
+      "step": 220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.923141602611835e-06,
+      "loss": 3.6641,
+      "step": 221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.922058109424168e-06,
+      "loss": 3.4648,
+      "step": 222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.920967092439028e-06,
+      "loss": 3.5938,
+      "step": 223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.919868553324131e-06,
+      "loss": 3.6484,
+      "step": 224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.918762493758688e-06,
+      "loss": 3.3789,
+      "step": 225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.917648915433413e-06,
+      "loss": 3.5078,
+      "step": 226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.916527820050504e-06,
+      "loss": 3.9688,
+      "step": 227
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.915399209323653e-06,
+      "loss": 3.5938,
+      "step": 228
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.914263084978041e-06,
+      "loss": 3.5469,
+      "step": 229
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.913119448750338e-06,
+      "loss": 3.4453,
+      "step": 230
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.911968302388685e-06,
+      "loss": 3.5898,
+      "step": 231
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.910809647652716e-06,
+      "loss": 3.5312,
+      "step": 232
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.909643486313533e-06,
+      "loss": 3.5859,
+      "step": 233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.90846982015372e-06,
+      "loss": 3.4023,
+      "step": 234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.907288650967324e-06,
+      "loss": 3.9297,
+      "step": 235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.906099980559869e-06,
+      "loss": 3.7305,
+      "step": 236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.904903810748338e-06,
+      "loss": 3.7266,
+      "step": 237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.903700143361185e-06,
+      "loss": 3.332,
+      "step": 238
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.902488980238322e-06,
+      "loss": 3.5469,
+      "step": 239
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.901270323231114e-06,
+      "loss": 4.0391,
+      "step": 240
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.900044174202389e-06,
+      "loss": 3.8672,
+      "step": 241
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.898810535026421e-06,
+      "loss": 3.5938,
+      "step": 242
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.897569407588936e-06,
+      "loss": 3.8359,
+      "step": 243
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.896320793787106e-06,
+      "loss": 3.4844,
+      "step": 244
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.895064695529549e-06,
+      "loss": 4.0312,
+      "step": 245
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.893801114736318e-06,
+      "loss": 3.7656,
+      "step": 246
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.892530053338909e-06,
+      "loss": 3.7461,
+      "step": 247
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.891251513280249e-06,
+      "loss": 3.4648,
+      "step": 248
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.889965496514702e-06,
+      "loss": 3.4531,
+      "step": 249
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.888672005008054e-06,
+      "loss": 3.3945,
+      "step": 250
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.887371040737523e-06,
+      "loss": 3.5938,
+      "step": 251
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.886062605691744e-06,
+      "loss": 3.4961,
+      "step": 252
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.884746701870778e-06,
+      "loss": 3.4688,
+      "step": 253
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.883423331286096e-06,
+      "loss": 3.2773,
+      "step": 254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.882092495960589e-06,
+      "loss": 3.8945,
+      "step": 255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.880754197928553e-06,
+      "loss": 3.4883,
+      "step": 256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.879408439235697e-06,
+      "loss": 3.5625,
+      "step": 257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.878055221939127e-06,
+      "loss": 3.7734,
+      "step": 258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.876694548107358e-06,
+      "loss": 3.6836,
+      "step": 259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.875326419820296e-06,
+      "loss": 3.7617,
+      "step": 260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.873950839169248e-06,
+      "loss": 3.7383,
+      "step": 261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.87256780825691e-06,
+      "loss": 4.0312,
+      "step": 262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.871177329197363e-06,
+      "loss": 3.5938,
+      "step": 263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.869779404116078e-06,
+      "loss": 3.4062,
+      "step": 264
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.868374035149906e-06,
+      "loss": 3.3945,
+      "step": 265
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.866961224447076e-06,
+      "loss": 3.6445,
+      "step": 266
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.865540974167194e-06,
+      "loss": 3.6367,
+      "step": 267
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.864113286481237e-06,
+      "loss": 3.7266,
+      "step": 268
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.86267816357155e-06,
+      "loss": 3.5352,
+      "step": 269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.861235607631847e-06,
+      "loss": 3.5352,
+      "step": 270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859785620867197e-06,
+      "loss": 3.5469,
+      "step": 271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858328205494035e-06,
+      "loss": 3.3516,
+      "step": 272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.856863363740145e-06,
+      "loss": 3.8477,
+      "step": 273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855391097844671e-06,
+      "loss": 3.3281,
+      "step": 274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853911410058097e-06,
+      "loss": 3.5273,
+      "step": 275
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.852424302642256e-06,
+      "loss": 3.625,
+      "step": 276
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.850929777870324e-06,
+      "loss": 3.5391,
+      "step": 277
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.84942783802681e-06,
+      "loss": 3.3359,
+      "step": 278
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.847918485407562e-06,
+      "loss": 3.5156,
+      "step": 279
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.846401722319759e-06,
+      "loss": 3.3516,
+      "step": 280
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.844877551081905e-06,
+      "loss": 3.5312,
+      "step": 281
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.843345974023833e-06,
+      "loss": 3.2656,
+      "step": 282
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.841806993486686e-06,
+      "loss": 3.8047,
+      "step": 283
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.840260611822938e-06,
+      "loss": 3.2695,
+      "step": 284
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.838706831396363e-06,
+      "loss": 3.6562,
+      "step": 285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.837145654582055e-06,
+      "loss": 3.5938,
+      "step": 286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.835577083766408e-06,
+      "loss": 3.6914,
+      "step": 287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.83400112134712e-06,
+      "loss": 3.5,
+      "step": 288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.832417769733185e-06,
+      "loss": 3.8594,
+      "step": 289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.830827031344901e-06,
+      "loss": 3.168,
+      "step": 290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.829228908613846e-06,
+      "loss": 3.7812,
+      "step": 291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.827623403982893e-06,
+      "loss": 3.7188,
+      "step": 292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.826010519906195e-06,
+      "loss": 3.7383,
+      "step": 293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.824390258849187e-06,
+      "loss": 3.2383,
+      "step": 294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.822762623288584e-06,
+      "loss": 3.7578,
+      "step": 295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.821127615712365e-06,
+      "loss": 3.6836,
+      "step": 296
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.819485238619785e-06,
+      "loss": 3.6367,
+      "step": 297
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.81783549452136e-06,
+      "loss": 3.5078,
+      "step": 298
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.816178385938867e-06,
+      "loss": 3.4375,
+      "step": 299
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.814513915405346e-06,
+      "loss": 3.2852,
+      "step": 300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.812842085465086e-06,
+      "loss": 3.8047,
+      "step": 301
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.81116289867362e-06,
+      "loss": 3.3477,
+      "step": 302
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.809476357597738e-06,
+      "loss": 3.4453,
+      "step": 303
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.807782464815463e-06,
+      "loss": 3.6133,
+      "step": 304
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.80608122291606e-06,
+      "loss": 3.6367,
+      "step": 305
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.804372634500025e-06,
+      "loss": 3.4727,
+      "step": 306
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.802656702179088e-06,
+      "loss": 3.5391,
+      "step": 307
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.800933428576201e-06,
+      "loss": 3.5156,
+      "step": 308
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.799202816325534e-06,
+      "loss": 3.5859,
+      "step": 309
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 3.6523,
+      "step": 310
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.795719586473663e-06,
+      "loss": 3.3828,
+      "step": 311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.793966974196877e-06,
+      "loss": 3.1953,
+      "step": 312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.792207033921152e-06,
+      "loss": 3.375,
+      "step": 313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.79043976833671e-06,
+      "loss": 3.3555,
+      "step": 314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.788665180144974e-06,
+      "loss": 3.3203,
+      "step": 315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.786883272058554e-06,
+      "loss": 3.5508,
+      "step": 316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.785094046801256e-06,
+      "loss": 3.4922,
+      "step": 317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.783297507108065e-06,
+      "loss": 3.7539,
+      "step": 318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.781493655725149e-06,
+      "loss": 3.5977,
+      "step": 319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.779682495409858e-06,
+      "loss": 3.5859,
+      "step": 320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.777864028930705e-06,
+      "loss": 3.2539,
+      "step": 321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.776038259067375e-06,
+      "loss": 3.793,
+      "step": 322
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.774205188610721e-06,
+      "loss": 3.5039,
+      "step": 323
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.77236482036275e-06,
+      "loss": 3.4727,
+      "step": 324
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.770517157136625e-06,
+      "loss": 3.1953,
+      "step": 325
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.768662201756663e-06,
+      "loss": 3.625,
+      "step": 326
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.766799957058324e-06,
+      "loss": 3.4961,
+      "step": 327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.764930425888216e-06,
+      "loss": 3.4453,
+      "step": 328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.76305361110408e-06,
+      "loss": 3.4062,
+      "step": 329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.761169515574787e-06,
+      "loss": 3.8398,
+      "step": 330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.759278142180348e-06,
+      "loss": 3.4297,
+      "step": 331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.757379493811892e-06,
+      "loss": 3.6797,
+      "step": 332
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.75547357337167e-06,
+      "loss": 3.4883,
+      "step": 333
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.753560383773046e-06,
+      "loss": 3.4531,
+      "step": 334
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.751639927940501e-06,
+      "loss": 3.6328,
+      "step": 335
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.749712208809617e-06,
+      "loss": 3.5312,
+      "step": 336
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.747777229327085e-06,
+      "loss": 3.4648,
+      "step": 337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.745834992450688e-06,
+      "loss": 3.5547,
+      "step": 338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.743885501149309e-06,
+      "loss": 3.3281,
+      "step": 339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.741928758402912e-06,
+      "loss": 3.625,
+      "step": 340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.739964767202552e-06,
+      "loss": 3.7148,
+      "step": 341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.737993530550362e-06,
+      "loss": 3.3008,
+      "step": 342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.736015051459551e-06,
+      "loss": 3.5625,
+      "step": 343
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.734029332954396e-06,
+      "loss": 3.418,
+      "step": 344
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.732036378070243e-06,
+      "loss": 3.6602,
+      "step": 345
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.730036189853497e-06,
+      "loss": 3.6172,
+      "step": 346
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.728028771361625e-06,
+      "loss": 3.5938,
+      "step": 347
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.726014125663136e-06,
+      "loss": 3.6328,
+      "step": 348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.723992255837596e-06,
+      "loss": 3.4922,
+      "step": 349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.721963164975613e-06,
+      "loss": 3.4414,
+      "step": 350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.719926856178824e-06,
+      "loss": 3.707,
+      "step": 351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.717883332559911e-06,
+      "loss": 3.4023,
+      "step": 352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.715832597242576e-06,
+      "loss": 3.6289,
+      "step": 353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.71377465336155e-06,
+      "loss": 3.6172,
+      "step": 354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.71170950406258e-06,
+      "loss": 3.3984,
+      "step": 355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.709637152502426e-06,
+      "loss": 3.5547,
+      "step": 356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.707557601848863e-06,
+      "loss": 3.4141,
+      "step": 357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.705470855280661e-06,
+      "loss": 3.3008,
+      "step": 358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.703376915987601e-06,
+      "loss": 3.2891,
+      "step": 359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.701275787170447e-06,
+      "loss": 3.4961,
+      "step": 360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.699167472040965e-06,
+      "loss": 3.6953,
+      "step": 361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.697051973821896e-06,
+      "loss": 3.3047,
+      "step": 362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.694929295746963e-06,
+      "loss": 3.9414,
+      "step": 363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.692799441060868e-06,
+      "loss": 3.5703,
+      "step": 364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.69066241301928e-06,
+      "loss": 3.4297,
+      "step": 365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.688518214888836e-06,
+      "loss": 3.4844,
+      "step": 366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.686366849947126e-06,
+      "loss": 3.3828,
+      "step": 367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.684208321482705e-06,
+      "loss": 3.4453,
+      "step": 368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.682042632795069e-06,
+      "loss": 3.332,
+      "step": 369
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.679869787194664e-06,
+      "loss": 3.5195,
+      "step": 370
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.67768978800288e-06,
+      "loss": 3.3594,
+      "step": 371
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.67550263855203e-06,
+      "loss": 3.5117,
+      "step": 372
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.673308342185366e-06,
+      "loss": 3.4609,
+      "step": 373
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.671106902257065e-06,
+      "loss": 3.5234,
+      "step": 374
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.668898322132218e-06,
+      "loss": 3.6562,
+      "step": 375
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.666682605186834e-06,
+      "loss": 3.4062,
+      "step": 376
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.664459754807832e-06,
+      "loss": 3.7031,
+      "step": 377
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.662229774393033e-06,
+      "loss": 3.8438,
+      "step": 378
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.659992667351157e-06,
+      "loss": 3.3086,
+      "step": 379
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.657748437101819e-06,
+      "loss": 3.6094,
+      "step": 380
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.655497087075523e-06,
+      "loss": 3.4336,
+      "step": 381
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.653238620713652e-06,
+      "loss": 3.4609,
+      "step": 382
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.65097304146847e-06,
+      "loss": 3.3438,
+      "step": 383
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.648700352803117e-06,
+      "loss": 3.5469,
+      "step": 384
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.646420558191596e-06,
+      "loss": 3.4258,
+      "step": 385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.64413366111877e-06,
+      "loss": 3.4531,
+      "step": 386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.641839665080363e-06,
+      "loss": 3.3281,
+      "step": 387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.639538573582952e-06,
+      "loss": 3.5,
+      "step": 388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.637230390143953e-06,
+      "loss": 3.5312,
+      "step": 389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.63491511829163e-06,
+      "loss": 3.2852,
+      "step": 390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.632592761565078e-06,
+      "loss": 3.8398,
+      "step": 391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.630263323514221e-06,
+      "loss": 3.625,
+      "step": 392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.627926807699813e-06,
+      "loss": 3.1289,
+      "step": 393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.625583217693419e-06,
+      "loss": 3.5312,
+      "step": 394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.623232557077427e-06,
+      "loss": 3.3672,
+      "step": 395
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.620874829445023e-06,
+      "loss": 3.4922,
+      "step": 396
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.618510038400203e-06,
+      "loss": 3.7539,
+      "step": 397
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.616138187557758e-06,
+      "loss": 3.5234,
+      "step": 398
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.613759280543269e-06,
+      "loss": 3.1953,
+      "step": 399
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.611373320993104e-06,
+      "loss": 3.4805,
+      "step": 400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.60898031255441e-06,
+      "loss": 3.4883,
+      "step": 401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.606580258885115e-06,
+      "loss": 3.5977,
+      "step": 402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.604173163653906e-06,
+      "loss": 3.2031,
+      "step": 403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.60175903054024e-06,
+      "loss": 3.2852,
+      "step": 404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.599337863234335e-06,
+      "loss": 3.4062,
+      "step": 405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.596909665437155e-06,
+      "loss": 3.5078,
+      "step": 406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.594474440860413e-06,
+      "loss": 3.4922,
+      "step": 407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.592032193226564e-06,
+      "loss": 3.6172,
+      "step": 408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.589582926268798e-06,
+      "loss": 3.4961,
+      "step": 409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.587126643731033e-06,
+      "loss": 3.3242,
+      "step": 410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.584663349367913e-06,
+      "loss": 3.5508,
+      "step": 411
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.582193046944799e-06,
+      "loss": 3.7539,
+      "step": 412
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.579715740237768e-06,
+      "loss": 3.7852,
+      "step": 413
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.577231433033595e-06,
+      "loss": 3.5898,
+      "step": 414
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.574740129129767e-06,
+      "loss": 3.3477,
+      "step": 415
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.572241832334457e-06,
+      "loss": 3.582,
+      "step": 416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.56973654646653e-06,
+      "loss": 3.4375,
+      "step": 417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.567224275355538e-06,
+      "loss": 3.3125,
+      "step": 418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.564705022841708e-06,
+      "loss": 3.5,
+      "step": 419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.562178792775936e-06,
+      "loss": 3.3438,
+      "step": 420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.559645589019787e-06,
+      "loss": 3.4531,
+      "step": 421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.557105415445485e-06,
+      "loss": 3.4531,
+      "step": 422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.554558275935907e-06,
+      "loss": 3.3086,
+      "step": 423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.552004174384582e-06,
+      "loss": 3.5078,
+      "step": 424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549443114695677e-06,
+      "loss": 3.2422,
+      "step": 425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.546875100783995e-06,
+      "loss": 3.2188,
+      "step": 426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.544300136574973e-06,
+      "loss": 3.3242,
+      "step": 427
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.541718226004667e-06,
+      "loss": 3.4102,
+      "step": 428
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.539129373019755e-06,
+      "loss": 3.2305,
+      "step": 429
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.536533581577526e-06,
+      "loss": 3.4688,
+      "step": 430
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.533930855645872e-06,
+      "loss": 3.2422,
+      "step": 431
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.531321199203293e-06,
+      "loss": 3.6172,
+      "step": 432
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.528704616238875e-06,
+      "loss": 3.7539,
+      "step": 433
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.526081110752292e-06,
+      "loss": 3.1055,
+      "step": 434
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.523450686753808e-06,
+      "loss": 3.25,
+      "step": 435
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.520813348264252e-06,
+      "loss": 3.3555,
+      "step": 436
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.518169099315028e-06,
+      "loss": 3.2344,
+      "step": 437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.515517943948104e-06,
+      "loss": 3.5234,
+      "step": 438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.512859886216003e-06,
+      "loss": 3.5898,
+      "step": 439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.5101949301818e-06,
+      "loss": 3.6016,
+      "step": 440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.507523079919112e-06,
+      "loss": 2.9219,
+      "step": 441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.504844339512096e-06,
+      "loss": 3.543,
+      "step": 442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.502158713055444e-06,
+      "loss": 2.9844,
+      "step": 443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.499466204654371e-06,
+      "loss": 3.625,
+      "step": 444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.496766818424612e-06,
+      "loss": 3.4414,
+      "step": 445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.494060558492416e-06,
+      "loss": 3.6523,
+      "step": 446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.491347428994536e-06,
+      "loss": 3.7305,
+      "step": 447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.488627434078232e-06,
+      "loss": 3.5859,
+      "step": 448
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.485900577901252e-06,
+      "loss": 3.4727,
+      "step": 449
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.483166864631837e-06,
+      "loss": 3.2656,
+      "step": 450
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.480426298448706e-06,
+      "loss": 3.3633,
+      "step": 451
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.477678883541055e-06,
+      "loss": 3.3789,
+      "step": 452
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.47492462410855e-06,
+      "loss": 3.4023,
+      "step": 453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.472163524361317e-06,
+      "loss": 3.4102,
+      "step": 454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.469395588519938e-06,
+      "loss": 3.6719,
+      "step": 455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.466620820815446e-06,
+      "loss": 3.6172,
+      "step": 456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.46383922548932e-06,
+      "loss": 3.4336,
+      "step": 457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.461050806793468e-06,
+      "loss": 3.5156,
+      "step": 458
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.458255568990234e-06,
+      "loss": 3.5742,
+      "step": 459
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.455453516352385e-06,
+      "loss": 3.25,
+      "step": 460
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.452644653163105e-06,
+      "loss": 3.8438,
+      "step": 461
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.449828983715985e-06,
+      "loss": 3.6562,
+      "step": 462
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.447006512315025e-06,
+      "loss": 3.6172,
+      "step": 463
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.444177243274619e-06,
+      "loss": 3.2578,
+      "step": 464
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.441341180919552e-06,
+      "loss": 3.2695,
+      "step": 465
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.438498329584995e-06,
+      "loss": 3.4375,
+      "step": 466
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.435648693616496e-06,
+      "loss": 3.6953,
+      "step": 467
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.432792277369975e-06,
+      "loss": 3.1523,
+      "step": 468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.429929085211711e-06,
+      "loss": 3.5977,
+      "step": 469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.427059121518346e-06,
+      "loss": 3.8594,
+      "step": 470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.424182390676872e-06,
+      "loss": 3.2031,
+      "step": 471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.421298897084622e-06,
+      "loss": 3.3945,
+      "step": 472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.418408645149273e-06,
+      "loss": 3.1484,
+      "step": 473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.415511639288825e-06,
+      "loss": 3.8711,
+      "step": 474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.412607883931608e-06,
+      "loss": 3.4102,
+      "step": 475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.409697383516263e-06,
+      "loss": 3.2383,
+      "step": 476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.40678014249175e-06,
+      "loss": 3.3867,
+      "step": 477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.403856165317322e-06,
+      "loss": 3.1836,
+      "step": 478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.40092545646254e-06,
+      "loss": 3.4531,
+      "step": 479
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.397988020407246e-06,
+      "loss": 3.2266,
+      "step": 480
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.39504386164157e-06,
+      "loss": 3.375,
+      "step": 481
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.392092984665919e-06,
+      "loss": 3.5234,
+      "step": 482
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.389135393990962e-06,
+      "loss": 3.7383,
+      "step": 483
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.38617109413764e-06,
+      "loss": 3.1797,
+      "step": 484
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.383200089637143e-06,
+      "loss": 3.6328,
+      "step": 485
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.380222385030916e-06,
+      "loss": 3.375,
+      "step": 486
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.377237984870636e-06,
+      "loss": 3.3203,
+      "step": 487
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.374246893718222e-06,
+      "loss": 3.3828,
+      "step": 488
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.37124911614582e-06,
+      "loss": 3.4883,
+      "step": 489
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.3682446567358e-06,
+      "loss": 3.4844,
+      "step": 490
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.365233520080732e-06,
+      "loss": 3.6016,
+      "step": 491
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.362215710783411e-06,
+      "loss": 3.6367,
+      "step": 492
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.359191233456821e-06,
+      "loss": 3.2734,
+      "step": 493
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.356160092724139e-06,
+      "loss": 3.375,
+      "step": 494
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.35312229321873e-06,
+      "loss": 3.125,
+      "step": 495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.350077839584139e-06,
+      "loss": 3.4023,
+      "step": 496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.347026736474077e-06,
+      "loss": 3.5781,
+      "step": 497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.343968988552427e-06,
+      "loss": 3.1289,
+      "step": 498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.34090460049322e-06,
+      "loss": 3.5469,
+      "step": 499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.337833576980647e-06,
+      "loss": 3.2188,
+      "step": 500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.334755922709032e-06,
+      "loss": 3.4258,
+      "step": 501
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.33167164238284e-06,
+      "loss": 3.0156,
+      "step": 502
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.328580740716668e-06,
+      "loss": 2.9883,
+      "step": 503
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.325483222435225e-06,
+      "loss": 3.4688,
+      "step": 504
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.322379092273345e-06,
+      "loss": 3.5156,
+      "step": 505
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.319268354975958e-06,
+      "loss": 3.3125,
+      "step": 506
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.316151015298103e-06,
+      "loss": 3.0,
+      "step": 507
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.313027078004903e-06,
+      "loss": 3.2656,
+      "step": 508
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.309896547871576e-06,
+      "loss": 3.2539,
+      "step": 509
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.306759429683404e-06,
+      "loss": 3.5938,
+      "step": 510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.303615728235753e-06,
+      "loss": 3.7305,
+      "step": 511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.300465448334044e-06,
+      "loss": 3.3164,
+      "step": 512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.297308594793757e-06,
+      "loss": 3.4414,
+      "step": 513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.29414517244042e-06,
+      "loss": 3.6055,
+      "step": 514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.2909751861096e-06,
+      "loss": 3.3086,
+      "step": 515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.287798640646899e-06,
+      "loss": 3.4531,
+      "step": 516
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.284615540907947e-06,
+      "loss": 3.3672,
+      "step": 517
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.281425891758392e-06,
+      "loss": 3.1445,
+      "step": 518
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.278229698073889e-06,
+      "loss": 3.2969,
+      "step": 519
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.275026964740101e-06,
+      "loss": 3.3125,
+      "step": 520
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.271817696652688e-06,
+      "loss": 3.2461,
+      "step": 521
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.2686018987173e-06,
+      "loss": 3.207,
+      "step": 522
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.265379575849561e-06,
+      "loss": 3.4375,
+      "step": 523
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.262150732975078e-06,
+      "loss": 3.4609,
+      "step": 524
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.258915375029419e-06,
+      "loss": 3.293,
+      "step": 525
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.255673506958114e-06,
+      "loss": 3.3125,
+      "step": 526
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.252425133716639e-06,
+      "loss": 3.3828,
+      "step": 527
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.249170260270422e-06,
+      "loss": 3.4336,
+      "step": 528
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.245908891594818e-06,
+      "loss": 3.3008,
+      "step": 529
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.242641032675118e-06,
+      "loss": 3.5117,
+      "step": 530
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.23936668850653e-06,
+      "loss": 3.4297,
+      "step": 531
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.236085864094178e-06,
+      "loss": 2.9727,
+      "step": 532
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.232798564453087e-06,
+      "loss": 3.3711,
+      "step": 533
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.229504794608182e-06,
+      "loss": 3.207,
+      "step": 534
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.226204559594284e-06,
+      "loss": 3.4609,
+      "step": 535
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.22289786445609e-06,
+      "loss": 3.2461,
+      "step": 536
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.219584714248168e-06,
+      "loss": 3.1719,
+      "step": 537
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.216265114034964e-06,
+      "loss": 3.207,
+      "step": 538
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.212939068890778e-06,
+      "loss": 3.6172,
+      "step": 539
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.209606583899759e-06,
+      "loss": 3.3555,
+      "step": 540
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 3.332,
+      "step": 541
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.202922314763049e-06,
+      "loss": 3.2812,
+      "step": 542
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.199570540834847e-06,
+      "loss": 3.3008,
+      "step": 543
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.19621234749478e-06,
+      "loss": 3.1055,
+      "step": 544
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.192847739876142e-06,
+      "loss": 3.0859,
+      "step": 545
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.189476723122034e-06,
+      "loss": 3.2305,
+      "step": 546
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.186099302385345e-06,
+      "loss": 3.3945,
+      "step": 547
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.182715482828764e-06,
+      "loss": 3.4219,
+      "step": 548
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.17932526962475e-06,
+      "loss": 3.6133,
+      "step": 549
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.175928667955547e-06,
+      "loss": 3.3047,
+      "step": 550
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.172525683013154e-06,
+      "loss": 3.4844,
+      "step": 551
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.169116319999336e-06,
+      "loss": 3.1758,
+      "step": 552
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.1657005841256e-06,
+      "loss": 3.3984,
+      "step": 553
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.162278480613203e-06,
+      "loss": 3.4492,
+      "step": 554
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.158850014693123e-06,
+      "loss": 3.3828,
+      "step": 555
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.155415191606074e-06,
+      "loss": 3.457,
+      "step": 556
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.151974016602486e-06,
+      "loss": 3.1953,
+      "step": 557
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.148526494942496e-06,
+      "loss": 3.4414,
+      "step": 558
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.145072631895942e-06,
+      "loss": 3.3906,
+      "step": 559
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.141612432742357e-06,
+      "loss": 3.25,
+      "step": 560
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.138145902770959e-06,
+      "loss": 3.625,
+      "step": 561
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.134673047280644e-06,
+      "loss": 3.2812,
+      "step": 562
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.131193871579975e-06,
+      "loss": 3.1914,
+      "step": 563
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.127708380987177e-06,
+      "loss": 3.457,
+      "step": 564
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.12421658083013e-06,
+      "loss": 3.2773,
+      "step": 565
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.120718476446354e-06,
+      "loss": 3.5508,
+      "step": 566
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.11721407318301e-06,
+      "loss": 3.2305,
+      "step": 567
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.113703376396885e-06,
+      "loss": 3.2188,
+      "step": 568
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.110186391454389e-06,
+      "loss": 3.1211,
+      "step": 569
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.106663123731538e-06,
+      "loss": 3.4453,
+      "step": 570
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.103133578613959e-06,
+      "loss": 3.0547,
+      "step": 571
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.09959776149687e-06,
+      "loss": 3.0352,
+      "step": 572
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.096055677785078e-06,
+      "loss": 3.1602,
+      "step": 573
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.092507332892968e-06,
+      "loss": 3.125,
+      "step": 574
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.088952732244497e-06,
+      "loss": 3.0,
+      "step": 575
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.085391881273182e-06,
+      "loss": 3.2656,
+      "step": 576
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.081824785422099e-06,
+      "loss": 3.3555,
+      "step": 577
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.078251450143866e-06,
+      "loss": 2.9062,
+      "step": 578
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.074671880900637e-06,
+      "loss": 3.293,
+      "step": 579
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.071086083164099e-06,
+      "loss": 3.4766,
+      "step": 580
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.06749406241546e-06,
+      "loss": 3.3555,
+      "step": 581
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.063895824145435e-06,
+      "loss": 3.0195,
+      "step": 582
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.060291373854252e-06,
+      "loss": 3.043,
+      "step": 583
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.056680717051623e-06,
+      "loss": 3.2266,
+      "step": 584
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.05306385925676e-06,
+      "loss": 3.2812,
+      "step": 585
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.04944080599834e-06,
+      "loss": 3.0859,
+      "step": 586
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.045811562814526e-06,
+      "loss": 3.3906,
+      "step": 587
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.042176135252927e-06,
+      "loss": 3.2539,
+      "step": 588
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.038534528870618e-06,
+      "loss": 3.6016,
+      "step": 589
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.034886749234112e-06,
+      "loss": 3.2852,
+      "step": 590
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.03123280191936e-06,
+      "loss": 3.125,
+      "step": 591
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.027572692511739e-06,
+      "loss": 3.6562,
+      "step": 592
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.02390642660605e-06,
+      "loss": 3.3477,
+      "step": 593
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.020234009806504e-06,
+      "loss": 3.1328,
+      "step": 594
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.016555447726704e-06,
+      "loss": 3.375,
+      "step": 595
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.012870745989663e-06,
+      "loss": 3.4648,
+      "step": 596
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.009179910227767e-06,
+      "loss": 3.3281,
+      "step": 597
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.005482946082784e-06,
+      "loss": 3.1797,
+      "step": 598
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.001779859205846e-06,
+      "loss": 3.0312,
+      "step": 599
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.998070655257447e-06,
+      "loss": 3.3789,
+      "step": 600
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.994355339907429e-06,
+      "loss": 3.6094,
+      "step": 601
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.990633918834978e-06,
+      "loss": 3.3594,
+      "step": 602
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.986906397728617e-06,
+      "loss": 3.2656,
+      "step": 603
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.98317278228618e-06,
+      "loss": 3.082,
+      "step": 604
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.979433078214834e-06,
+      "loss": 2.8984,
+      "step": 605
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.975687291231041e-06,
+      "loss": 3.4883,
+      "step": 606
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.971935427060563e-06,
+      "loss": 3.1875,
+      "step": 607
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.968177491438457e-06,
+      "loss": 3.293,
+      "step": 608
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.964413490109056e-06,
+      "loss": 3.1016,
+      "step": 609
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.960643428825965e-06,
+      "loss": 3.0938,
+      "step": 610
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.956867313352055e-06,
+      "loss": 3.3594,
+      "step": 611
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.95308514945945e-06,
+      "loss": 3.6602,
+      "step": 612
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.949296942929515e-06,
+      "loss": 3.3594,
+      "step": 613
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.94550269955286e-06,
+      "loss": 3.2773,
+      "step": 614
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.94170242512932e-06,
+      "loss": 3.1875,
+      "step": 615
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.937896125467941e-06,
+      "loss": 3.4375,
+      "step": 616
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.934083806386995e-06,
+      "loss": 3.0898,
+      "step": 617
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.930265473713939e-06,
+      "loss": 2.9688,
+      "step": 618
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.92644113328543e-06,
+      "loss": 3.4805,
+      "step": 619
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.922610790947312e-06,
+      "loss": 3.0898,
+      "step": 620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.918774452554595e-06,
+      "loss": 3.3164,
+      "step": 621
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.91493212397146e-06,
+      "loss": 3.1328,
+      "step": 622
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.911083811071244e-06,
+      "loss": 3.0273,
+      "step": 623
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.90722951973643e-06,
+      "loss": 3.3203,
+      "step": 624
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.90336925585864e-06,
+      "loss": 3.0938,
+      "step": 625
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.899503025338628e-06,
+      "loss": 3.668,
+      "step": 626
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.895630834086264e-06,
+      "loss": 3.1953,
+      "step": 627
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.891752688020532e-06,
+      "loss": 3.2148,
+      "step": 628
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.88786859306952e-06,
+      "loss": 3.2969,
+      "step": 629
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.883978555170405e-06,
+      "loss": 3.0312,
+      "step": 630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.880082580269454e-06,
+      "loss": 2.7461,
+      "step": 631
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.876180674322006e-06,
+      "loss": 2.9297,
+      "step": 632
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.872272843292465e-06,
+      "loss": 3.4258,
+      "step": 633
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.868359093154292e-06,
+      "loss": 3.2734,
+      "step": 634
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.864439429890001e-06,
+      "loss": 3.3555,
+      "step": 635
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.86051385949114e-06,
+      "loss": 3.0078,
+      "step": 636
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.856582387958286e-06,
+      "loss": 3.1992,
+      "step": 637
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.852645021301041e-06,
+      "loss": 3.1523,
+      "step": 638
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.84870176553801e-06,
+      "loss": 3.6172,
+      "step": 639
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.84475262669681e-06,
+      "loss": 3.1016,
+      "step": 640
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.840797610814046e-06,
+      "loss": 3.4453,
+      "step": 641
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.836836723935304e-06,
+      "loss": 3.4023,
+      "step": 642
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.832869972115148e-06,
+      "loss": 2.9883,
+      "step": 643
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.828897361417106e-06,
+      "loss": 3.1758,
+      "step": 644
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.82491889791366e-06,
+      "loss": 3.0195,
+      "step": 645
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.820934587686247e-06,
+      "loss": 2.9805,
+      "step": 646
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 8.816944436825227e-06,
+      "loss": 3.0781,
+      "step": 647
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.812948451429898e-06,
+      "loss": 3.1445,
+      "step": 648
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.808946637608476e-06,
+      "loss": 3.5547,
+      "step": 649
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.804939001478084e-06,
+      "loss": 2.9922,
+      "step": 650
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.800925549164742e-06,
+      "loss": 3.1992,
+      "step": 651
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 8.796906286803366e-06,
+      "loss": 3.5625,
+      "step": 652
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.792881220537752e-06,
+      "loss": 2.9844,
+      "step": 653
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.788850356520567e-06,
+      "loss": 2.8125,
+      "step": 654
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.784813700913336e-06,
+      "loss": 3.0352,
+      "step": 655
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.780771259886447e-06,
+      "loss": 2.9805,
+      "step": 656
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 8.776723039619122e-06,
+      "loss": 3.3555,
+      "step": 657
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.77266904629942e-06,
+      "loss": 3.0547,
+      "step": 658
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.768609286124229e-06,
+      "loss": 3.1016,
+      "step": 659
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.764543765299245e-06,
+      "loss": 3.2266,
+      "step": 660
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.760472490038974e-06,
+      "loss": 2.9844,
+      "step": 661
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 8.756395466566718e-06,
+      "loss": 3.1016,
+      "step": 662
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.752312701114565e-06,
+      "loss": 3.3945,
+      "step": 663
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.748224199923378e-06,
+      "loss": 2.7188,
+      "step": 664
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.744129969242792e-06,
+      "loss": 3.4492,
+      "step": 665
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.740030015331198e-06,
+      "loss": 3.1602,
+      "step": 666
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.735924344455732e-06,
+      "loss": 3.0117,
+      "step": 667
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 8.731812962892277e-06,
+      "loss": 2.7773,
+      "step": 668
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.727695876925434e-06,
+      "loss": 3.3203,
+      "step": 669
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.723573092848535e-06,
+      "loss": 3.5078,
+      "step": 670
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.719444616963613e-06,
+      "loss": 3.1289,
+      "step": 671
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.715310455581409e-06,
+      "loss": 3.2656,
+      "step": 672
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 8.71117061502135e-06,
+      "loss": 3.1758,
+      "step": 673
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.707025101611546e-06,
+      "loss": 3.3438,
+      "step": 674
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.702873921688777e-06,
+      "loss": 3.2461,
+      "step": 675
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.698717081598484e-06,
+      "loss": 2.9648,
+      "step": 676
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.694554587694763e-06,
+      "loss": 3.1289,
+      "step": 677
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 8.690386446340353e-06,
+      "loss": 2.9297,
+      "step": 678
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.68621266390662e-06,
+      "loss": 3.3125,
+      "step": 679
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.68203324677356e-06,
+      "loss": 2.8281,
+      "step": 680
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.677848201329775e-06,
+      "loss": 3.4922,
+      "step": 681
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.673657533972477e-06,
+      "loss": 3.168,
+      "step": 682
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 8.669461251107466e-06,
+      "loss": 2.9414,
+      "step": 683
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.665259359149132e-06,
+      "loss": 3.3906,
+      "step": 684
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.661051864520434e-06,
+      "loss": 3.0156,
+      "step": 685
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.656838773652891e-06,
+      "loss": 2.8398,
+      "step": 686
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.65262009298659e-06,
+      "loss": 3.125,
+      "step": 687
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.64839582897015e-06,
+      "loss": 3.7773,
+      "step": 688
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 8.644165988060731e-06,
+      "loss": 3.2734,
+      "step": 689
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.639930576724014e-06,
+      "loss": 3.1094,
+      "step": 690
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.635689601434197e-06,
+      "loss": 3.1133,
+      "step": 691
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.631443068673983e-06,
+      "loss": 3.1797,
+      "step": 692
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.627190984934568e-06,
+      "loss": 3.1641,
+      "step": 693
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 8.622933356715639e-06,
+      "loss": 3.1914,
+      "step": 694
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.61867019052535e-06,
+      "loss": 3.082,
+      "step": 695
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.614401492880328e-06,
+      "loss": 2.957,
+      "step": 696
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.61012727030565e-06,
+      "loss": 3.4766,
+      "step": 697
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.60584752933484e-06,
+      "loss": 3.3828,
+      "step": 698
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 8.601562276509855e-06,
+      "loss": 3.1016,
+      "step": 699
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.597271518381083e-06,
+      "loss": 3.1172,
+      "step": 700
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.592975261507323e-06,
+      "loss": 3.2148,
+      "step": 701
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.588673512455781e-06,
+      "loss": 3.0312,
+      "step": 702
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.584366277802057e-06,
+      "loss": 3.0391,
+      "step": 703
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 8.580053564130137e-06,
+      "loss": 2.9961,
+      "step": 704
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.575735378032382e-06,
+      "loss": 3.4609,
+      "step": 705
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.571411726109518e-06,
+      "loss": 3.1055,
+      "step": 706
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.567082614970628e-06,
+      "loss": 3.1172,
+      "step": 707
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.562748051233135e-06,
+      "loss": 2.9141,
+      "step": 708
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.558408041522801e-06,
+      "loss": 2.8945,
+      "step": 709
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 8.554062592473713e-06,
+      "loss": 3.8281,
+      "step": 710
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.549711710728269e-06,
+      "loss": 3.1562,
+      "step": 711
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.545355402937174e-06,
+      "loss": 3.0898,
+      "step": 712
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.540993675759427e-06,
+      "loss": 3.168,
+      "step": 713
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.536626535862309e-06,
+      "loss": 3.4688,
+      "step": 714
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 8.532253989921377e-06,
+      "loss": 3.1367,
+      "step": 715
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.527876044620453e-06,
+      "loss": 3.3125,
+      "step": 716
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.523492706651607e-06,
+      "loss": 3.0898,
+      "step": 717
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.519103982715158e-06,
+      "loss": 2.8984,
+      "step": 718
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.514709879519654e-06,
+      "loss": 3.2188,
+      "step": 719
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 8.510310403781868e-06,
+      "loss": 2.7266,
+      "step": 720
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.505905562226784e-06,
+      "loss": 3.293,
+      "step": 721
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.501495361587589e-06,
+      "loss": 3.0586,
+      "step": 722
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.497079808605659e-06,
+      "loss": 2.7109,
+      "step": 723
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.492658910030558e-06,
+      "loss": 3.3008,
+      "step": 724
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 8.488232672620016e-06,
+      "loss": 2.8281,
+      "step": 725
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.483801103139923e-06,
+      "loss": 3.6719,
+      "step": 726
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.479364208364322e-06,
+      "loss": 2.9492,
+      "step": 727
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.474921995075399e-06,
+      "loss": 3.0781,
+      "step": 728
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.470474470063462e-06,
+      "loss": 2.8594,
+      "step": 729
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 8.466021640126946e-06,
+      "loss": 3.0234,
+      "step": 730
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.461563512072392e-06,
+      "loss": 3.1914,
+      "step": 731
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.457100092714438e-06,
+      "loss": 3.0273,
+      "step": 732
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.452631388875814e-06,
+      "loss": 3.207,
+      "step": 733
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.448157407387324e-06,
+      "loss": 2.9883,
+      "step": 734
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.443678155087841e-06,
+      "loss": 2.9531,
+      "step": 735
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 8.439193638824296e-06,
+      "loss": 3.1172,
+      "step": 736
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.434703865451666e-06,
+      "loss": 3.1094,
+      "step": 737
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.43020884183296e-06,
+      "loss": 3.0039,
+      "step": 738
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.425708574839221e-06,
+      "loss": 3.1484,
+      "step": 739
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.421203071349498e-06,
+      "loss": 3.2109,
+      "step": 740
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 8.416692338250849e-06,
+      "loss": 2.8945,
+      "step": 741
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.412176382438327e-06,
+      "loss": 2.957,
+      "step": 742
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.407655210814962e-06,
+      "loss": 3.3789,
+      "step": 743
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.403128830291767e-06,
+      "loss": 3.2383,
+      "step": 744
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.398597247787706e-06,
+      "loss": 3.3359,
+      "step": 745
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 8.394060470229704e-06,
+      "loss": 3.0938,
+      "step": 746
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.389518504552623e-06,
+      "loss": 2.8242,
+      "step": 747
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.384971357699255e-06,
+      "loss": 2.9492,
+      "step": 748
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.380419036620311e-06,
+      "loss": 3.082,
+      "step": 749
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.375861548274417e-06,
+      "loss": 3.3516,
+      "step": 750
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 8.371298899628091e-06,
+      "loss": 2.9492,
+      "step": 751
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.366731097655742e-06,
+      "loss": 3.0156,
+      "step": 752
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.362158149339657e-06,
+      "loss": 3.2109,
+      "step": 753
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.357580061669985e-06,
+      "loss": 3.0195,
+      "step": 754
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.35299684164474e-06,
+      "loss": 2.9531,
+      "step": 755
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.348408496269771e-06,
+      "loss": 3.1641,
+      "step": 756
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 8.343815032558768e-06,
+      "loss": 2.957,
+      "step": 757
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.339216457533244e-06,
+      "loss": 3.1602,
+      "step": 758
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.334612778222524e-06,
+      "loss": 2.8242,
+      "step": 759
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.330004001663734e-06,
+      "loss": 3.0781,
+      "step": 760
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.325390134901794e-06,
+      "loss": 3.2148,
+      "step": 761
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 8.320771184989404e-06,
+      "loss": 3.3359,
+      "step": 762
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.316147158987036e-06,
+      "loss": 3.1875,
+      "step": 763
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.311518063962915e-06,
+      "loss": 3.293,
+      "step": 764
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.306883906993022e-06,
+      "loss": 2.957,
+      "step": 765
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.30224469516107e-06,
+      "loss": 2.8164,
+      "step": 766
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 8.297600435558506e-06,
+      "loss": 3.0938,
+      "step": 767
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.292951135284481e-06,
+      "loss": 2.8477,
+      "step": 768
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.288296801445863e-06,
+      "loss": 2.5312,
+      "step": 769
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.28363744115721e-06,
+      "loss": 3.2891,
+      "step": 770
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.278973061540758e-06,
+      "loss": 3.043,
+      "step": 771
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 2.9492,
+      "step": 772
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.269629272851786e-06,
+      "loss": 2.957,
+      "step": 773
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.264949878062063e-06,
+      "loss": 3.2305,
+      "step": 774
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.260265492510121e-06,
+      "loss": 3.2305,
+      "step": 775
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.25557612335646e-06,
+      "loss": 3.125,
+      "step": 776
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.250881777769187e-06,
+      "loss": 2.7031,
+      "step": 777
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 8.246182462924022e-06,
+      "loss": 3.0312,
+      "step": 778
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.24147818600428e-06,
+      "loss": 2.625,
+      "step": 779
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.236768954200861e-06,
+      "loss": 3.0234,
+      "step": 780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.232054774712239e-06,
+      "loss": 3.2227,
+      "step": 781
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.227335654744451e-06,
+      "loss": 2.8867,
+      "step": 782
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 8.222611601511084e-06,
+      "loss": 3.0469,
+      "step": 783
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.217882622233268e-06,
+      "loss": 2.793,
+      "step": 784
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.213148724139665e-06,
+      "loss": 2.9453,
+      "step": 785
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.20840991446645e-06,
+      "loss": 3.0234,
+      "step": 786
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.203666200457312e-06,
+      "loss": 3.1094,
+      "step": 787
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 8.198917589363433e-06,
+      "loss": 2.7227,
+      "step": 788
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.19416408844348e-06,
+      "loss": 3.0195,
+      "step": 789
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.1894057049636e-06,
+      "loss": 2.9023,
+      "step": 790
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.184642446197396e-06,
+      "loss": 3.2461,
+      "step": 791
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.179874319425926e-06,
+      "loss": 3.1211,
+      "step": 792
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 8.175101331937692e-06,
+      "loss": 2.9922,
+      "step": 793
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.170323491028625e-06,
+      "loss": 2.6875,
+      "step": 794
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.16554080400207e-06,
+      "loss": 3.1211,
+      "step": 795
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.160753278168787e-06,
+      "loss": 3.0156,
+      "step": 796
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.155960920846925e-06,
+      "loss": 3.2773,
+      "step": 797
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.151163739362029e-06,
+      "loss": 3.082,
+      "step": 798
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 8.146361741047007e-06,
+      "loss": 3.1328,
+      "step": 799
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.141554933242135e-06,
+      "loss": 3.0469,
+      "step": 800
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.136743323295041e-06,
+      "loss": 3.1328,
+      "step": 801
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.131926918560692e-06,
+      "loss": 2.9453,
+      "step": 802
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.127105726401387e-06,
+      "loss": 3.1875,
+      "step": 803
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 8.122279754186738e-06,
+      "loss": 3.1641,
+      "step": 804
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.117449009293668e-06,
+      "loss": 3.2109,
+      "step": 805
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.112613499106397e-06,
+      "loss": 2.8984,
+      "step": 806
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.10777323101642e-06,
+      "loss": 2.6992,
+      "step": 807
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.102928212422519e-06,
+      "loss": 3.125,
+      "step": 808
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 8.098078450730723e-06,
+      "loss": 3.0,
+      "step": 809
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.093223953354324e-06,
+      "loss": 2.9531,
+      "step": 810
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.08836472771384e-06,
+      "loss": 3.1641,
+      "step": 811
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.08350078123703e-06,
+      "loss": 2.8359,
+      "step": 812
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.07863212135886e-06,
+      "loss": 3.0352,
+      "step": 813
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.073758755521506e-06,
+      "loss": 3.2344,
+      "step": 814
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.06888069117433e-06,
+      "loss": 3.0156,
+      "step": 815
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.063997935773885e-06,
+      "loss": 3.0391,
+      "step": 816
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.059110496783891e-06,
+      "loss": 2.8984,
+      "step": 817
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.054218381675225e-06,
+      "loss": 2.9844,
+      "step": 818
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.049321597925914e-06,
+      "loss": 3.3398,
+      "step": 819
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.044420153021123e-06,
+      "loss": 3.0312,
+      "step": 820
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.03951405445314e-06,
+      "loss": 3.457,
+      "step": 821
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.034603309721369e-06,
+      "loss": 3.125,
+      "step": 822
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.02968792633231e-06,
+      "loss": 3.1328,
+      "step": 823
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.024767911799558e-06,
+      "loss": 3.0703,
+      "step": 824
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.019843273643789e-06,
+      "loss": 3.1211,
+      "step": 825
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.014914019392743e-06,
+      "loss": 3.207,
+      "step": 826
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.009980156581218e-06,
+      "loss": 3.0664,
+      "step": 827
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.005041692751055e-06,
+      "loss": 3.1445,
+      "step": 828
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.00009863545113e-06,
+      "loss": 3.4531,
+      "step": 829
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.99515099223734e-06,
+      "loss": 3.1172,
+      "step": 830
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.99019877067259e-06,
+      "loss": 2.9805,
+      "step": 831
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.985241978326786e-06,
+      "loss": 2.9648,
+      "step": 832
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.98028062277682e-06,
+      "loss": 2.9727,
+      "step": 833
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.975314711606558e-06,
+      "loss": 2.5312,
+      "step": 834
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.970344252406832e-06,
+      "loss": 2.6758,
+      "step": 835
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.965369252775423e-06,
+      "loss": 3.3281,
+      "step": 836
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.960389720317056e-06,
+      "loss": 2.7188,
+      "step": 837
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.955405662643384e-06,
+      "loss": 3.2422,
+      "step": 838
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.950417087372972e-06,
+      "loss": 2.8281,
+      "step": 839
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.945424002131298e-06,
+      "loss": 3.1133,
+      "step": 840
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.940426414550732e-06,
+      "loss": 3.0469,
+      "step": 841
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.935424332270523e-06,
+      "loss": 2.9414,
+      "step": 842
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.93041776293679e-06,
+      "loss": 2.8828,
+      "step": 843
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.925406714202518e-06,
+      "loss": 3.2695,
+      "step": 844
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.920391193727533e-06,
+      "loss": 2.7344,
+      "step": 845
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.915371209178496e-06,
+      "loss": 2.9492,
+      "step": 846
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.910346768228893e-06,
+      "loss": 2.8008,
+      "step": 847
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.905317878559026e-06,
+      "loss": 3.1484,
+      "step": 848
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.900284547855992e-06,
+      "loss": 3.0195,
+      "step": 849
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.895246783813679e-06,
+      "loss": 3.0,
+      "step": 850
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.89020459413275e-06,
+      "loss": 3.3398,
+      "step": 851
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.885157986520635e-06,
+      "loss": 2.668,
+      "step": 852
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.880106968691516e-06,
+      "loss": 2.875,
+      "step": 853
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.875051548366317e-06,
+      "loss": 2.8789,
+      "step": 854
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.869991733272693e-06,
+      "loss": 2.8516,
+      "step": 855
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.864927531145012e-06,
+      "loss": 2.7383,
+      "step": 856
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.859858949724351e-06,
+      "loss": 2.8984,
+      "step": 857
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.854785996758484e-06,
+      "loss": 2.7773,
+      "step": 858
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.849708680001862e-06,
+      "loss": 3.043,
+      "step": 859
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.844627007215613e-06,
+      "loss": 2.7852,
+      "step": 860
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.839540986167514e-06,
+      "loss": 3.0898,
+      "step": 861
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.834450624631998e-06,
+      "loss": 2.8984,
+      "step": 862
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.829355930390126e-06,
+      "loss": 2.8633,
+      "step": 863
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.824256911229588e-06,
+      "loss": 2.875,
+      "step": 864
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.81915357494468e-06,
+      "loss": 2.6992,
+      "step": 865
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.814045929336301e-06,
+      "loss": 3.0117,
+      "step": 866
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.808933982211932e-06,
+      "loss": 2.8945,
+      "step": 867
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.803817741385636e-06,
+      "loss": 2.8945,
+      "step": 868
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.798697214678033e-06,
+      "loss": 2.832,
+      "step": 869
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.7935724099163e-06,
+      "loss": 2.7812,
+      "step": 870
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.788443334934148e-06,
+      "loss": 3.1484,
+      "step": 871
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.783309997571819e-06,
+      "loss": 2.8086,
+      "step": 872
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.778172405676067e-06,
+      "loss": 3.0039,
+      "step": 873
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.773030567100158e-06,
+      "loss": 2.8789,
+      "step": 874
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.767884489703837e-06,
+      "loss": 3.2109,
+      "step": 875
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.762734181353336e-06,
+      "loss": 2.8633,
+      "step": 876
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.757579649921354e-06,
+      "loss": 2.8203,
+      "step": 877
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.752420903287043e-06,
+      "loss": 2.6367,
+      "step": 878
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.747257949336003e-06,
+      "loss": 3.0195,
+      "step": 879
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.742090795960258e-06,
+      "loss": 3.2109,
+      "step": 880
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.736919451058257e-06,
+      "loss": 3.0547,
+      "step": 881
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.731743922534854e-06,
+      "loss": 2.7539,
+      "step": 882
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.726564218301298e-06,
+      "loss": 2.7266,
+      "step": 883
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.721380346275221e-06,
+      "loss": 2.7656,
+      "step": 884
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.716192314380626e-06,
+      "loss": 2.7773,
+      "step": 885
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.711000130547875e-06,
+      "loss": 2.7578,
+      "step": 886
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.705803802713678e-06,
+      "loss": 2.6055,
+      "step": 887
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.700603338821074e-06,
+      "loss": 3.1484,
+      "step": 888
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.695398746819431e-06,
+      "loss": 3.0352,
+      "step": 889
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.690190034664423e-06,
+      "loss": 3.0312,
+      "step": 890
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.684977210318024e-06,
+      "loss": 3.1328,
+      "step": 891
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.679760281748492e-06,
+      "loss": 2.625,
+      "step": 892
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.674539256930364e-06,
+      "loss": 3.125,
+      "step": 893
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.669314143844429e-06,
+      "loss": 2.8516,
+      "step": 894
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.664084950477732e-06,
+      "loss": 2.5078,
+      "step": 895
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.658851684823554e-06,
+      "loss": 3.0859,
+      "step": 896
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.653614354881403e-06,
+      "loss": 3.3047,
+      "step": 897
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.648372968656995e-06,
+      "loss": 2.9766,
+      "step": 898
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.643127534162246e-06,
+      "loss": 3.0039,
+      "step": 899
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.637878059415267e-06,
+      "loss": 2.9297,
+      "step": 900
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.632624552440337e-06,
+      "loss": 2.7734,
+      "step": 901
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.627367021267906e-06,
+      "loss": 3.2031,
+      "step": 902
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.622105473934569e-06,
+      "loss": 3.332,
+      "step": 903
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.616839918483061e-06,
+      "loss": 2.7109,
+      "step": 904
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.611570362962247e-06,
+      "loss": 2.6719,
+      "step": 905
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.606296815427106e-06,
+      "loss": 2.6562,
+      "step": 906
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.601019283938718e-06,
+      "loss": 2.4531,
+      "step": 907
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.595737776564249e-06,
+      "loss": 3.1211,
+      "step": 908
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.590452301376949e-06,
+      "loss": 2.9375,
+      "step": 909
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.58516286645613e-06,
+      "loss": 3.0273,
+      "step": 910
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.579869479887158e-06,
+      "loss": 2.7734,
+      "step": 911
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.574572149761437e-06,
+      "loss": 3.0039,
+      "step": 912
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.569270884176401e-06,
+      "loss": 2.5977,
+      "step": 913
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.5639656912355e-06,
+      "loss": 2.582,
+      "step": 914
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.5586565790481855e-06,
+      "loss": 2.7695,
+      "step": 915
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.553343555729904e-06,
+      "loss": 2.6133,
+      "step": 916
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.548026629402074e-06,
+      "loss": 3.0664,
+      "step": 917
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.542705808192085e-06,
+      "loss": 2.7891,
+      "step": 918
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.5373811002332785e-06,
+      "loss": 2.9141,
+      "step": 919
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.532052513664939e-06,
+      "loss": 2.8359,
+      "step": 920
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.526720056632277e-06,
+      "loss": 2.5977,
+      "step": 921
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.521383737286423e-06,
+      "loss": 2.5703,
+      "step": 922
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.516043563784405e-06,
+      "loss": 2.8164,
+      "step": 923
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.510699544289151e-06,
+      "loss": 2.9766,
+      "step": 924
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.505351686969458e-06,
+      "loss": 2.8008,
+      "step": 925
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 2.7344,
+      "step": 926
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.494644491561299e-06,
+      "loss": 3.1602,
+      "step": 927
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.4892851698397174e-06,
+      "loss": 2.832,
+      "step": 928
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.483922043027448e-06,
+      "loss": 2.5781,
+      "step": 929
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.478555119322505e-06,
+      "loss": 2.4961,
+      "step": 930
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4731844069286965e-06,
+      "loss": 2.918,
+      "step": 931
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.467809914055631e-06,
+      "loss": 2.6758,
+      "step": 932
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.462431648918689e-06,
+      "loss": 2.9258,
+      "step": 933
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.457049619739023e-06,
+      "loss": 2.7969,
+      "step": 934
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.451663834743537e-06,
+      "loss": 2.8555,
+      "step": 935
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.446274302164873e-06,
+      "loss": 2.8203,
+      "step": 936
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.440881030241407e-06,
+      "loss": 2.5938,
+      "step": 937
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.435484027217226e-06,
+      "loss": 2.5156,
+      "step": 938
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.430083301342124e-06,
+      "loss": 2.8281,
+      "step": 939
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.424678860871584e-06,
+      "loss": 2.3047,
+      "step": 940
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.419270714066765e-06,
+      "loss": 3.1602,
+      "step": 941
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.413858869194497e-06,
+      "loss": 2.6953,
+      "step": 942
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.408443334527257e-06,
+      "loss": 2.6367,
+      "step": 943
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.4030241183431675e-06,
+      "loss": 2.7852,
+      "step": 944
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.39760122892597e-06,
+      "loss": 2.8359,
+      "step": 945
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.392174674565032e-06,
+      "loss": 2.9141,
+      "step": 946
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.3867444635553165e-06,
+      "loss": 2.625,
+      "step": 947
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.381310604197375e-06,
+      "loss": 2.7305,
+      "step": 948
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.3758731047973416e-06,
+      "loss": 2.668,
+      "step": 949
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.370431973666909e-06,
+      "loss": 2.7539,
+      "step": 950
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.364987219123323e-06,
+      "loss": 2.5859,
+      "step": 951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3595388494893684e-06,
+      "loss": 2.6602,
+      "step": 952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.354086873093357e-06,
+      "loss": 2.5781,
+      "step": 953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.3486312982691134e-06,
+      "loss": 3.0078,
+      "step": 954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.343172133355958e-06,
+      "loss": 2.9727,
+      "step": 955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.337709386698709e-06,
+      "loss": 2.8672,
+      "step": 956
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.332243066647651e-06,
+      "loss": 2.832,
+      "step": 957
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.326773181558532e-06,
+      "loss": 2.8594,
+      "step": 958
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.321299739792553e-06,
+      "loss": 2.8418,
+      "step": 959
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.31582274971635e-06,
+      "loss": 2.5469,
+      "step": 960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.310342219701981e-06,
+      "loss": 2.7656,
+      "step": 961
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.304858158126917e-06,
+      "loss": 2.5195,
+      "step": 962
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.299370573374032e-06,
+      "loss": 2.9023,
+      "step": 963
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.293879473831573e-06,
+      "loss": 2.3438,
+      "step": 964
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.28838486789317e-06,
+      "loss": 3.1133,
+      "step": 965
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 7.282886763957812e-06,
+      "loss": 2.8711,
+      "step": 966
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.2773851704298304e-06,
+      "loss": 2.4727,
+      "step": 967
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.271880095718895e-06,
+      "loss": 2.8125,
+      "step": 968
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.266371548239997e-06,
+      "loss": 3.0039,
+      "step": 969
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.260859536413429e-06,
+      "loss": 2.5703,
+      "step": 970
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.255344068664789e-06,
+      "loss": 3.0117,
+      "step": 971
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 7.249825153424952e-06,
+      "loss": 3.1758,
+      "step": 972
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.244302799130064e-06,
+      "loss": 2.832,
+      "step": 973
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.238777014221528e-06,
+      "loss": 2.7422,
+      "step": 974
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.233247807145989e-06,
+      "loss": 2.6484,
+      "step": 975
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.227715186355326e-06,
+      "loss": 3.0039,
+      "step": 976
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 7.222179160306637e-06,
+      "loss": 2.7539,
+      "step": 977
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.216639737462221e-06,
+      "loss": 2.7383,
+      "step": 978
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.211096926289572e-06,
+      "loss": 2.6055,
+      "step": 979
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.205550735261364e-06,
+      "loss": 3.3047,
+      "step": 980
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.200001172855436e-06,
+      "loss": 2.6797,
+      "step": 981
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 7.19444824755478e-06,
+      "loss": 2.5312,
+      "step": 982
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.1888919678475285e-06,
+      "loss": 3.2344,
+      "step": 983
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.183332342226945e-06,
+      "loss": 2.9141,
+      "step": 984
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.1777693791914025e-06,
+      "loss": 2.7812,
+      "step": 985
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.1722030872443804e-06,
+      "loss": 2.7344,
+      "step": 986
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 7.1666334748944424e-06,
+      "loss": 2.957,
+      "step": 987
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.161060550655227e-06,
+      "loss": 2.5391,
+      "step": 988
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.155484323045442e-06,
+      "loss": 2.6875,
+      "step": 989
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.149904800588838e-06,
+      "loss": 2.8633,
+      "step": 990
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.144321991814205e-06,
+      "loss": 2.5781,
+      "step": 991
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 7.138735905255355e-06,
+      "loss": 2.6836,
+      "step": 992
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.133146549451113e-06,
+      "loss": 2.793,
+      "step": 993
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.1275539329452984e-06,
+      "loss": 2.6484,
+      "step": 994
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.121958064286714e-06,
+      "loss": 2.6133,
+      "step": 995
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.11635895202914e-06,
+      "loss": 2.6094,
+      "step": 996
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.110756604731307e-06,
+      "loss": 2.7227,
+      "step": 997
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 7.105151030956895e-06,
+      "loss": 2.793,
+      "step": 998
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.099542239274516e-06,
+      "loss": 2.8633,
+      "step": 999
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.0939302382577e-06,
+      "loss": 3.1523,
+      "step": 1000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.088315036484882e-06,
+      "loss": 2.5625,
+      "step": 1001
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.082696642539391e-06,
+      "loss": 2.5703,
+      "step": 1002
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 2.9141,
+      "step": 1003
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.071450312488084e-06,
+      "loss": 3.0703,
+      "step": 1004
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.0658223935732715e-06,
+      "loss": 2.7891,
+      "step": 1005
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.060191316867764e-06,
+      "loss": 2.5723,
+      "step": 1006
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.0545570909791525e-06,
+      "loss": 2.6211,
+      "step": 1007
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 7.048919724519848e-06,
+      "loss": 2.6211,
+      "step": 1008
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.0432792261070605e-06,
+      "loss": 2.6641,
+      "step": 1009
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.037635604362786e-06,
+      "loss": 2.7266,
+      "step": 1010
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.031988867913793e-06,
+      "loss": 2.8672,
+      "step": 1011
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.026339025391615e-06,
+      "loss": 2.8633,
+      "step": 1012
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 7.020686085432532e-06,
+      "loss": 2.8125,
+      "step": 1013
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.015030056677559e-06,
+      "loss": 2.7539,
+      "step": 1014
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.00937094777243e-06,
+      "loss": 3.0703,
+      "step": 1015
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 7.00370876736759e-06,
+      "loss": 2.8281,
+      "step": 1016
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.9980435241181785e-06,
+      "loss": 2.6562,
+      "step": 1017
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.992375226684016e-06,
+      "loss": 2.707,
+      "step": 1018
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.986703883729595e-06,
+      "loss": 2.6719,
+      "step": 1019
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.981029503924059e-06,
+      "loss": 2.5703,
+      "step": 1020
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.975352095941194e-06,
+      "loss": 2.6797,
+      "step": 1021
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.969671668459414e-06,
+      "loss": 2.8711,
+      "step": 1022
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.963988230161757e-06,
+      "loss": 2.7656,
+      "step": 1023
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.958301789735853e-06,
+      "loss": 2.8281,
+      "step": 1024
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.952612355873922e-06,
+      "loss": 2.5195,
+      "step": 1025
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.946919937272768e-06,
+      "loss": 2.4766,
+      "step": 1026
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.941224542633748e-06,
+      "loss": 2.6484,
+      "step": 1027
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.935526180662776e-06,
+      "loss": 2.7383,
+      "step": 1028
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.929824860070295e-06,
+      "loss": 2.7383,
+      "step": 1029
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.924120589571275e-06,
+      "loss": 2.5508,
+      "step": 1030
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.918413377885193e-06,
+      "loss": 2.6562,
+      "step": 1031
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.91270323373602e-06,
+      "loss": 2.582,
+      "step": 1032
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.906990165852218e-06,
+      "loss": 2.8359,
+      "step": 1033
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.901274182966709e-06,
+      "loss": 2.5156,
+      "step": 1034
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.895555293816871e-06,
+      "loss": 2.502,
+      "step": 1035
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.889833507144534e-06,
+      "loss": 2.7109,
+      "step": 1036
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.884108831695946e-06,
+      "loss": 2.7695,
+      "step": 1037
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.878381276221777e-06,
+      "loss": 2.457,
+      "step": 1038
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.872650849477098e-06,
+      "loss": 2.4219,
+      "step": 1039
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.866917560221369e-06,
+      "loss": 2.6602,
+      "step": 1040
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.861181417218425e-06,
+      "loss": 2.625,
+      "step": 1041
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.855442429236463e-06,
+      "loss": 2.5195,
+      "step": 1042
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.849700605048031e-06,
+      "loss": 2.793,
+      "step": 1043
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.843955953430008e-06,
+      "loss": 2.5938,
+      "step": 1044
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.838208483163601e-06,
+      "loss": 2.707,
+      "step": 1045
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.832458203034321e-06,
+      "loss": 2.0371,
+      "step": 1046
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.8267051218319766e-06,
+      "loss": 3.0781,
+      "step": 1047
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.820949248350653e-06,
+      "loss": 2.3164,
+      "step": 1048
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.815190591388713e-06,
+      "loss": 2.8516,
+      "step": 1049
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.809429159748766e-06,
+      "loss": 2.8086,
+      "step": 1050
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.803664962237665e-06,
+      "loss": 2.5977,
+      "step": 1051
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.797898007666493e-06,
+      "loss": 2.6797,
+      "step": 1052
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.792128304850543e-06,
+      "loss": 2.7734,
+      "step": 1053
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.786355862609315e-06,
+      "loss": 2.543,
+      "step": 1054
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.780580689766492e-06,
+      "loss": 2.9766,
+      "step": 1055
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.774802795149932e-06,
+      "loss": 2.5547,
+      "step": 1056
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.769022187591654e-06,
+      "loss": 2.7344,
+      "step": 1057
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.7632388759278225e-06,
+      "loss": 3.0664,
+      "step": 1058
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.757452868998737e-06,
+      "loss": 2.5859,
+      "step": 1059
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.751664175648817e-06,
+      "loss": 2.8828,
+      "step": 1060
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.745872804726588e-06,
+      "loss": 2.7695,
+      "step": 1061
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.74007876508467e-06,
+      "loss": 2.7734,
+      "step": 1062
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.734282065579757e-06,
+      "loss": 2.4023,
+      "step": 1063
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.728482715072615e-06,
+      "loss": 2.9414,
+      "step": 1064
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.722680722428058e-06,
+      "loss": 2.6523,
+      "step": 1065
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.716876096514944e-06,
+      "loss": 3.1523,
+      "step": 1066
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.711068846206149e-06,
+      "loss": 2.3672,
+      "step": 1067
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.705258980378565e-06,
+      "loss": 2.457,
+      "step": 1068
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.699446507913083e-06,
+      "loss": 2.4688,
+      "step": 1069
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.693631437694579e-06,
+      "loss": 2.6914,
+      "step": 1070
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.687813778611893e-06,
+      "loss": 2.457,
+      "step": 1071
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.681993539557828e-06,
+      "loss": 2.6445,
+      "step": 1072
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.676170729429132e-06,
+      "loss": 2.4805,
+      "step": 1073
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.670345357126481e-06,
+      "loss": 2.5352,
+      "step": 1074
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.664517431554466e-06,
+      "loss": 2.7617,
+      "step": 1075
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.658686961621582e-06,
+      "loss": 2.9062,
+      "step": 1076
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.652853956240215e-06,
+      "loss": 2.7383,
+      "step": 1077
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.6470184243266235e-06,
+      "loss": 2.8086,
+      "step": 1078
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.641180374800931e-06,
+      "loss": 2.6387,
+      "step": 1079
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.635339816587109e-06,
+      "loss": 2.8984,
+      "step": 1080
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.6294967586129614e-06,
+      "loss": 2.9062,
+      "step": 1081
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.623651209810116e-06,
+      "loss": 2.8281,
+      "step": 1082
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.617803179114005e-06,
+      "loss": 2.7617,
+      "step": 1083
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.611952675463859e-06,
+      "loss": 2.4922,
+      "step": 1084
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.606099707802682e-06,
+      "loss": 2.4609,
+      "step": 1085
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.600244285077251e-06,
+      "loss": 2.4102,
+      "step": 1086
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.594386416238095e-06,
+      "loss": 2.8906,
+      "step": 1087
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.588526110239475e-06,
+      "loss": 2.543,
+      "step": 1088
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.582663376039387e-06,
+      "loss": 2.6914,
+      "step": 1089
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.576798222599532e-06,
+      "loss": 2.6367,
+      "step": 1090
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.570930658885314e-06,
+      "loss": 2.5742,
+      "step": 1091
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.565060693865815e-06,
+      "loss": 2.6133,
+      "step": 1092
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.559188336513794e-06,
+      "loss": 2.6016,
+      "step": 1093
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.553313595805666e-06,
+      "loss": 2.7109,
+      "step": 1094
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.547436480721482e-06,
+      "loss": 2.4648,
+      "step": 1095
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.5415570002449315e-06,
+      "loss": 2.8555,
+      "step": 1096
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.535675163363315e-06,
+      "loss": 2.7188,
+      "step": 1097
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.5297909790675375e-06,
+      "loss": 2.7461,
+      "step": 1098
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.523904456352089e-06,
+      "loss": 2.4766,
+      "step": 1099
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.518015604215037e-06,
+      "loss": 2.7891,
+      "step": 1100
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.512124431658006e-06,
+      "loss": 2.5742,
+      "step": 1101
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.5062309476861714e-06,
+      "loss": 2.707,
+      "step": 1102
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.500335161308242e-06,
+      "loss": 2.8438,
+      "step": 1103
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.494437081536442e-06,
+      "loss": 2.6406,
+      "step": 1104
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.488536717386504e-06,
+      "loss": 2.8906,
+      "step": 1105
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.4826340778776534e-06,
+      "loss": 2.6992,
+      "step": 1106
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.47672917203259e-06,
+      "loss": 2.6719,
+      "step": 1107
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.470822008877482e-06,
+      "loss": 2.5391,
+      "step": 1108
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.464912597441947e-06,
+      "loss": 2.5586,
+      "step": 1109
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.4590009467590366e-06,
+      "loss": 2.8125,
+      "step": 1110
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.453087065865229e-06,
+      "loss": 2.4668,
+      "step": 1111
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.447170963800408e-06,
+      "loss": 2.3438,
+      "step": 1112
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.4412526496078555e-06,
+      "loss": 2.9453,
+      "step": 1113
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.435332132334233e-06,
+      "loss": 2.3477,
+      "step": 1114
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.4294094210295725e-06,
+      "loss": 2.4102,
+      "step": 1115
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.423484524747255e-06,
+      "loss": 2.6562,
+      "step": 1116
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.4175574525440065e-06,
+      "loss": 2.5586,
+      "step": 1117
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.411628213479875e-06,
+      "loss": 3.0273,
+      "step": 1118
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.405696816618222e-06,
+      "loss": 2.7656,
+      "step": 1119
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.3997632710257095e-06,
+      "loss": 2.7852,
+      "step": 1120
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.393827585772281e-06,
+      "loss": 2.5273,
+      "step": 1121
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.3878897699311525e-06,
+      "loss": 2.3828,
+      "step": 1122
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.381949832578796e-06,
+      "loss": 2.4277,
+      "step": 1123
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.376007782794926e-06,
+      "loss": 2.3867,
+      "step": 1124
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.370063629662488e-06,
+      "loss": 2.625,
+      "step": 1125
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.364117382267642e-06,
+      "loss": 2.2578,
+      "step": 1126
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.358169049699746e-06,
+      "loss": 2.5781,
+      "step": 1127
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.352218641051349e-06,
+      "loss": 2.2188,
+      "step": 1128
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.346266165418173e-06,
+      "loss": 2.4746,
+      "step": 1129
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.340311631899099e-06,
+      "loss": 2.4844,
+      "step": 1130
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.334355049596155e-06,
+      "loss": 2.4531,
+      "step": 1131
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.328396427614496e-06,
+      "loss": 2.6094,
+      "step": 1132
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.3224357750624014e-06,
+      "loss": 2.7852,
+      "step": 1133
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.316473101051251e-06,
+      "loss": 2.3906,
+      "step": 1134
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.310508414695511e-06,
+      "loss": 2.707,
+      "step": 1135
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.304541725112734e-06,
+      "loss": 2.3711,
+      "step": 1136
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.298573041423523e-06,
+      "loss": 2.9453,
+      "step": 1137
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.292602372751536e-06,
+      "loss": 2.7891,
+      "step": 1138
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.2866297282234654e-06,
+      "loss": 2.5391,
+      "step": 1139
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.280655116969022e-06,
+      "loss": 2.6836,
+      "step": 1140
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.274678548120921e-06,
+      "loss": 2.4648,
+      "step": 1141
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.268700030814871e-06,
+      "loss": 2.8633,
+      "step": 1142
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.262719574189564e-06,
+      "loss": 2.832,
+      "step": 1143
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.25673718738665e-06,
+      "loss": 2.5234,
+      "step": 1144
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.2507528795507326e-06,
+      "loss": 2.4414,
+      "step": 1145
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.244766659829351e-06,
+      "loss": 2.6406,
+      "step": 1146
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.238778537372966e-06,
+      "loss": 2.2812,
+      "step": 1147
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.232788521334948e-06,
+      "loss": 2.9844,
+      "step": 1148
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.226796620871563e-06,
+      "loss": 2.5156,
+      "step": 1149
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.2208028451419575e-06,
+      "loss": 2.9609,
+      "step": 1150
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.214807203308141e-06,
+      "loss": 2.2344,
+      "step": 1151
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.208809704534977e-06,
+      "loss": 2.6719,
+      "step": 1152
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.2028103579901725e-06,
+      "loss": 2.7773,
+      "step": 1153
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.196809172844252e-06,
+      "loss": 3.0547,
+      "step": 1154
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.1908061582705506e-06,
+      "loss": 2.4727,
+      "step": 1155
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.184801323445208e-06,
+      "loss": 2.3164,
+      "step": 1156
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.178794677547138e-06,
+      "loss": 2.3418,
+      "step": 1157
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.1727862297580245e-06,
+      "loss": 2.8828,
+      "step": 1158
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.166775989262309e-06,
+      "loss": 2.9141,
+      "step": 1159
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.160763965247171e-06,
+      "loss": 2.6016,
+      "step": 1160
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.154750166902516e-06,
+      "loss": 2.8164,
+      "step": 1161
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.1487346034209605e-06,
+      "loss": 2.5625,
+      "step": 1162
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.142717283997823e-06,
+      "loss": 2.8516,
+      "step": 1163
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.136698217831106e-06,
+      "loss": 2.5781,
+      "step": 1164
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.130677414121474e-06,
+      "loss": 2.3516,
+      "step": 1165
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.124654882072258e-06,
+      "loss": 2.7539,
+      "step": 1166
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.118630630889426e-06,
+      "loss": 2.6758,
+      "step": 1167
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.112604669781572e-06,
+      "loss": 2.4258,
+      "step": 1168
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.10657700795991e-06,
+      "loss": 2.1992,
+      "step": 1169
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.100547654638245e-06,
+      "loss": 2.084,
+      "step": 1170
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.094516619032975e-06,
+      "loss": 2.5039,
+      "step": 1171
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.088483910363065e-06,
+      "loss": 2.7773,
+      "step": 1172
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.082449537850043e-06,
+      "loss": 2.4082,
+      "step": 1173
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.076413510717973e-06,
+      "loss": 2.7031,
+      "step": 1174
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.070375838193453e-06,
+      "loss": 2.5938,
+      "step": 1175
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.064336529505594e-06,
+      "loss": 2.832,
+      "step": 1176
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.058295593886013e-06,
+      "loss": 2.3086,
+      "step": 1177
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.052253040568804e-06,
+      "loss": 2.3477,
+      "step": 1178
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.046208878790543e-06,
+      "loss": 2.4141,
+      "step": 1179
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.040163117790261e-06,
+      "loss": 2.2656,
+      "step": 1180
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.034115766809434e-06,
+      "loss": 2.7188,
+      "step": 1181
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.028066835091968e-06,
+      "loss": 3.1406,
+      "step": 1182
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.022016331884185e-06,
+      "loss": 2.957,
+      "step": 1183
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.01596426643481e-06,
+      "loss": 2.5215,
+      "step": 1184
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.009910647994956e-06,
+      "loss": 2.291,
+      "step": 1185
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.003855485818111e-06,
+      "loss": 2.6797,
+      "step": 1186
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.997798789160121e-06,
+      "loss": 2.7422,
+      "step": 1187
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.991740567279174e-06,
+      "loss": 2.1875,
+      "step": 1188
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.985680829435798e-06,
+      "loss": 2.5039,
+      "step": 1189
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.979619584892834e-06,
+      "loss": 2.5938,
+      "step": 1190
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.973556842915422e-06,
+      "loss": 2.4023,
+      "step": 1191
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 5.967492612770999e-06,
+      "loss": 2.5078,
+      "step": 1192
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.961426903729271e-06,
+      "loss": 2.3398,
+      "step": 1193
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.955359725062203e-06,
+      "loss": 2.5664,
+      "step": 1194
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.9492910860440135e-06,
+      "loss": 2.4141,
+      "step": 1195
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.94322099595115e-06,
+      "loss": 2.3086,
+      "step": 1196
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 5.937149464062275e-06,
+      "loss": 2.7734,
+      "step": 1197
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.931076499658258e-06,
+      "loss": 2.3809,
+      "step": 1198
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.925002112022158e-06,
+      "loss": 2.3711,
+      "step": 1199
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.9189263104392104e-06,
+      "loss": 2.8711,
+      "step": 1200
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.91284910419681e-06,
+      "loss": 2.2109,
+      "step": 1201
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 5.906770502584497e-06,
+      "loss": 2.7656,
+      "step": 1202
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.900690514893951e-06,
+      "loss": 2.6973,
+      "step": 1203
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.894609150418964e-06,
+      "loss": 2.332,
+      "step": 1204
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.888526418455436e-06,
+      "loss": 2.1113,
+      "step": 1205
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.882442328301356e-06,
+      "loss": 2.3047,
+      "step": 1206
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 5.876356889256788e-06,
+      "loss": 2.5586,
+      "step": 1207
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.870270110623859e-06,
+      "loss": 2.707,
+      "step": 1208
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.864182001706747e-06,
+      "loss": 2.6699,
+      "step": 1209
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.858092571811661e-06,
+      "loss": 1.959,
+      "step": 1210
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.852001830246823e-06,
+      "loss": 2.3867,
+      "step": 1211
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.8459097863224705e-06,
+      "loss": 2.3164,
+      "step": 1212
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 5.839816449350824e-06,
+      "loss": 2.5625,
+      "step": 1213
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.833721828646086e-06,
+      "loss": 2.4258,
+      "step": 1214
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.827625933524418e-06,
+      "loss": 2.5898,
+      "step": 1215
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.821528773303929e-06,
+      "loss": 2.375,
+      "step": 1216
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.815430357304664e-06,
+      "loss": 2.5195,
+      "step": 1217
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 5.809330694848586e-06,
+      "loss": 2.8516,
+      "step": 1218
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.803229795259566e-06,
+      "loss": 2.3145,
+      "step": 1219
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.7971276678633625e-06,
+      "loss": 2.3594,
+      "step": 1220
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.791024321987611e-06,
+      "loss": 2.6953,
+      "step": 1221
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.784919766961813e-06,
+      "loss": 2.582,
+      "step": 1222
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 5.778814012117315e-06,
+      "loss": 2.3594,
+      "step": 1223
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.772707066787302e-06,
+      "loss": 2.0762,
+      "step": 1224
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.76659894030677e-06,
+      "loss": 2.3262,
+      "step": 1225
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.76048964201253e-06,
+      "loss": 2.7461,
+      "step": 1226
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.754379181243179e-06,
+      "loss": 2.5039,
+      "step": 1227
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.748267567339093e-06,
+      "loss": 2.7695,
+      "step": 1228
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.742154809642408e-06,
+      "loss": 2.4336,
+      "step": 1229
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.7360409174970126e-06,
+      "loss": 2.7344,
+      "step": 1230
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.729925900248524e-06,
+      "loss": 2.2637,
+      "step": 1231
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.723809767244284e-06,
+      "loss": 2.2793,
+      "step": 1232
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.717692527833341e-06,
+      "loss": 2.8047,
+      "step": 1233
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 2.5547,
+      "step": 1234
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.705454767195958e-06,
+      "loss": 2.4609,
+      "step": 1235
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.699334264676011e-06,
+      "loss": 2.5859,
+      "step": 1236
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.693212693162311e-06,
+      "loss": 2.5938,
+      "step": 1237
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.6870900620122175e-06,
+      "loss": 2.6953,
+      "step": 1238
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.680966380584706e-06,
+      "loss": 2.7578,
+      "step": 1239
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.674841658240365e-06,
+      "loss": 2.8379,
+      "step": 1240
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.668715904341365e-06,
+      "loss": 2.9023,
+      "step": 1241
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.66258912825146e-06,
+      "loss": 2.1055,
+      "step": 1242
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.656461339335968e-06,
+      "loss": 2.2148,
+      "step": 1243
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.6503325469617475e-06,
+      "loss": 2.5234,
+      "step": 1244
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.644202760497195e-06,
+      "loss": 2.2773,
+      "step": 1245
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.638071989312228e-06,
+      "loss": 2.5312,
+      "step": 1246
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.631940242778271e-06,
+      "loss": 2.3242,
+      "step": 1247
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.62580753026823e-06,
+      "loss": 2.2402,
+      "step": 1248
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.619673861156498e-06,
+      "loss": 2.5156,
+      "step": 1249
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.613539244818926e-06,
+      "loss": 2.3066,
+      "step": 1250
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.607403690632814e-06,
+      "loss": 2.4609,
+      "step": 1251
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.60126720797689e-06,
+      "loss": 2.9062,
+      "step": 1252
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.595129806231311e-06,
+      "loss": 2.7891,
+      "step": 1253
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.5889914947776305e-06,
+      "loss": 2.1699,
+      "step": 1254
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.5828522829987965e-06,
+      "loss": 2.3223,
+      "step": 1255
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.576712180279134e-06,
+      "loss": 2.2734,
+      "step": 1256
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.570571196004328e-06,
+      "loss": 2.1289,
+      "step": 1257
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.564429339561411e-06,
+      "loss": 2.4941,
+      "step": 1258
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.55828662033875e-06,
+      "loss": 2.1953,
+      "step": 1259
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.552143047726029e-06,
+      "loss": 2.2422,
+      "step": 1260
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.5459986311142365e-06,
+      "loss": 2.4766,
+      "step": 1261
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.539853379895656e-06,
+      "loss": 2.3281,
+      "step": 1262
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.533707303463839e-06,
+      "loss": 2.334,
+      "step": 1263
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.527560411213604e-06,
+      "loss": 2.1855,
+      "step": 1264
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.521412712541013e-06,
+      "loss": 2.4023,
+      "step": 1265
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.515264216843368e-06,
+      "loss": 2.7227,
+      "step": 1266
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.509114933519179e-06,
+      "loss": 1.998,
+      "step": 1267
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.502964871968163e-06,
+      "loss": 2.1406,
+      "step": 1268
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.496814041591234e-06,
+      "loss": 2.6836,
+      "step": 1269
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.490662451790472e-06,
+      "loss": 2.4727,
+      "step": 1270
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.484510111969123e-06,
+      "loss": 2.4531,
+      "step": 1271
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.478357031531578e-06,
+      "loss": 2.3027,
+      "step": 1272
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.4722032198833595e-06,
+      "loss": 2.5078,
+      "step": 1273
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.466048686431108e-06,
+      "loss": 2.4414,
+      "step": 1274
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.459893440582568e-06,
+      "loss": 2.5234,
+      "step": 1275
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.453737491746572e-06,
+      "loss": 2.1055,
+      "step": 1276
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.447580849333028e-06,
+      "loss": 2.3691,
+      "step": 1277
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.441423522752904e-06,
+      "loss": 2.6992,
+      "step": 1278
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.435265521418215e-06,
+      "loss": 2.543,
+      "step": 1279
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.429106854742006e-06,
+      "loss": 2.3574,
+      "step": 1280
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.422947532138335e-06,
+      "loss": 2.5156,
+      "step": 1281
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.41678756302227e-06,
+      "loss": 2.3438,
+      "step": 1282
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.410626956809864e-06,
+      "loss": 2.3301,
+      "step": 1283
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.404465722918144e-06,
+      "loss": 2.7461,
+      "step": 1284
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.3983038707650955e-06,
+      "loss": 2.4941,
+      "step": 1285
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.39214140976965e-06,
+      "loss": 2.8789,
+      "step": 1286
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.38597834935167e-06,
+      "loss": 2.0742,
+      "step": 1287
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.379814698931935e-06,
+      "loss": 2.4199,
+      "step": 1288
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.373650467932122e-06,
+      "loss": 2.6641,
+      "step": 1289
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.367485665774802e-06,
+      "loss": 2.2344,
+      "step": 1290
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.361320301883413e-06,
+      "loss": 2.5352,
+      "step": 1291
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.355154385682255e-06,
+      "loss": 2.5,
+      "step": 1292
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.348987926596474e-06,
+      "loss": 2.0625,
+      "step": 1293
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.342820934052042e-06,
+      "loss": 2.7383,
+      "step": 1294
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.336653417475745e-06,
+      "loss": 2.332,
+      "step": 1295
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.3304853862951785e-06,
+      "loss": 2.1797,
+      "step": 1296
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.324316849938715e-06,
+      "loss": 2.6797,
+      "step": 1297
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.318147817835506e-06,
+      "loss": 2.3633,
+      "step": 1298
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3119782994154556e-06,
+      "loss": 2.3086,
+      "step": 1299
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.3058083041092145e-06,
+      "loss": 2.6816,
+      "step": 1300
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.299637841348162e-06,
+      "loss": 2.6758,
+      "step": 1301
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.29346692056439e-06,
+      "loss": 2.4688,
+      "step": 1302
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.2872955511906974e-06,
+      "loss": 2.2852,
+      "step": 1303
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.281123742660558e-06,
+      "loss": 2.5117,
+      "step": 1304
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.274951504408122e-06,
+      "loss": 2.5195,
+      "step": 1305
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.268778845868201e-06,
+      "loss": 2.4297,
+      "step": 1306
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.262605776476243e-06,
+      "loss": 2.5312,
+      "step": 1307
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.256432305668324e-06,
+      "loss": 2.4844,
+      "step": 1308
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.250258442881139e-06,
+      "loss": 2.6328,
+      "step": 1309
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.244084197551976e-06,
+      "loss": 2.3047,
+      "step": 1310
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.237909579118713e-06,
+      "loss": 2.1641,
+      "step": 1311
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.231734597019792e-06,
+      "loss": 2.6016,
+      "step": 1312
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.2255592606942185e-06,
+      "loss": 2.3203,
+      "step": 1313
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.219383579581534e-06,
+      "loss": 2.4297,
+      "step": 1314
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.2132075631218075e-06,
+      "loss": 2.1719,
+      "step": 1315
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.2070312207556225e-06,
+      "loss": 2.4883,
+      "step": 1316
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.200854561924063e-06,
+      "loss": 2.1465,
+      "step": 1317
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.194677596068689e-06,
+      "loss": 2.3477,
+      "step": 1318
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.188500332631538e-06,
+      "loss": 2.1465,
+      "step": 1319
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.182322781055097e-06,
+      "loss": 2.1094,
+      "step": 1320
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.176144950782296e-06,
+      "loss": 2.2383,
+      "step": 1321
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.169966851256489e-06,
+      "loss": 2.2773,
+      "step": 1322
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.163788491921447e-06,
+      "loss": 2.3965,
+      "step": 1323
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.1576098822213285e-06,
+      "loss": 2.1152,
+      "step": 1324
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.1514310316006835e-06,
+      "loss": 2.5195,
+      "step": 1325
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.1452519495044265e-06,
+      "loss": 2.1641,
+      "step": 1326
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.139072645377829e-06,
+      "loss": 2.2559,
+      "step": 1327
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.132893128666493e-06,
+      "loss": 2.0664,
+      "step": 1328
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1267134088163576e-06,
+      "loss": 2.3125,
+      "step": 1329
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.120533495273664e-06,
+      "loss": 2.5391,
+      "step": 1330
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1143533974849526e-06,
+      "loss": 2.4141,
+      "step": 1331
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.1081731248970435e-06,
+      "loss": 2.2773,
+      "step": 1332
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.101992686957028e-06,
+      "loss": 2.2949,
+      "step": 1333
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.095812093112245e-06,
+      "loss": 2.1777,
+      "step": 1334
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.089631352810275e-06,
+      "loss": 2.082,
+      "step": 1335
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0834504754989225e-06,
+      "loss": 2.1797,
+      "step": 1336
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0772694706262e-06,
+      "loss": 2.6367,
+      "step": 1337
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.0710883476403135e-06,
+      "loss": 2.1875,
+      "step": 1338
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.064907115989655e-06,
+      "loss": 2.4121,
+      "step": 1339
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.058725785122777e-06,
+      "loss": 2.1133,
+      "step": 1340
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.052544364488386e-06,
+      "loss": 2.3867,
+      "step": 1341
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.046362863535327e-06,
+      "loss": 2.1016,
+      "step": 1342
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.040181291712564e-06,
+      "loss": 1.9805,
+      "step": 1343
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.033999658469174e-06,
+      "loss": 2.4707,
+      "step": 1344
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.0278179732543244e-06,
+      "loss": 2.2305,
+      "step": 1345
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.021636245517261e-06,
+      "loss": 2.2422,
+      "step": 1346
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.0154544847073015e-06,
+      "loss": 2.5898,
+      "step": 1347
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.009272700273804e-06,
+      "loss": 2.1055,
+      "step": 1348
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.003090901666173e-06,
+      "loss": 2.5273,
+      "step": 1349
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.9969090983338285e-06,
+      "loss": 2.4375,
+      "step": 1350
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.990727299726196e-06,
+      "loss": 2.2969,
+      "step": 1351
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.9845455152927e-06,
+      "loss": 2.4922,
+      "step": 1352
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.978363754482741e-06,
+      "loss": 2.5547,
+      "step": 1353
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 4.972182026745678e-06,
+      "loss": 2.5391,
+      "step": 1354
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.966000341530827e-06,
+      "loss": 2.5039,
+      "step": 1355
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.959818708287436e-06,
+      "loss": 2.1816,
+      "step": 1356
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.953637136464675e-06,
+      "loss": 2.2461,
+      "step": 1357
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.947455635511615e-06,
+      "loss": 2.2539,
+      "step": 1358
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 4.941274214877223e-06,
+      "loss": 2.3047,
+      "step": 1359
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.935092884010347e-06,
+      "loss": 2.5352,
+      "step": 1360
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.928911652359687e-06,
+      "loss": 2.1797,
+      "step": 1361
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.922730529373801e-06,
+      "loss": 2.2031,
+      "step": 1362
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.91654952450108e-06,
+      "loss": 2.1055,
+      "step": 1363
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.910368647189726e-06,
+      "loss": 2.334,
+      "step": 1364
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 4.904187906887756e-06,
+      "loss": 2.498,
+      "step": 1365
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.898007313042975e-06,
+      "loss": 2.668,
+      "step": 1366
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.891826875102958e-06,
+      "loss": 2.123,
+      "step": 1367
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.885646602515049e-06,
+      "loss": 2.3516,
+      "step": 1368
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.879466504726337e-06,
+      "loss": 2.4141,
+      "step": 1369
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 4.873286591183644e-06,
+      "loss": 2.2441,
+      "step": 1370
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.867106871333509e-06,
+      "loss": 2.2617,
+      "step": 1371
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.860927354622173e-06,
+      "loss": 2.4141,
+      "step": 1372
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.854748050495574e-06,
+      "loss": 2.2148,
+      "step": 1373
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.848568968399317e-06,
+      "loss": 2.7383,
+      "step": 1374
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 4.842390117778673e-06,
+      "loss": 2.2246,
+      "step": 1375
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.836211508078557e-06,
+      "loss": 2.1367,
+      "step": 1376
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.830033148743512e-06,
+      "loss": 2.2148,
+      "step": 1377
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.8238550492177065e-06,
+      "loss": 2.0254,
+      "step": 1378
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.817677218944904e-06,
+      "loss": 2.3477,
+      "step": 1379
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 4.811499667368464e-06,
+      "loss": 2.0488,
+      "step": 1380
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.805322403931312e-06,
+      "loss": 2.2969,
+      "step": 1381
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.7991454380759376e-06,
+      "loss": 2.1914,
+      "step": 1382
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.792968779244378e-06,
+      "loss": 2.0977,
+      "step": 1383
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.786792436878193e-06,
+      "loss": 2.0664,
+      "step": 1384
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 4.780616420418468e-06,
+      "loss": 2.3242,
+      "step": 1385
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.774440739305784e-06,
+      "loss": 2.0938,
+      "step": 1386
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.768265402980209e-06,
+      "loss": 2.293,
+      "step": 1387
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.762090420881289e-06,
+      "loss": 1.9121,
+      "step": 1388
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.755915802448024e-06,
+      "loss": 2.3105,
+      "step": 1389
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.749741557118863e-06,
+      "loss": 2.6758,
+      "step": 1390
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 4.743567694331677e-06,
+      "loss": 2.1738,
+      "step": 1391
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.737394223523758e-06,
+      "loss": 2.498,
+      "step": 1392
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7312211541318e-06,
+      "loss": 2.0195,
+      "step": 1393
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.7250484955918794e-06,
+      "loss": 2.3926,
+      "step": 1394
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.718876257339444e-06,
+      "loss": 2.1543,
+      "step": 1395
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 4.712704448809305e-06,
+      "loss": 2.5781,
+      "step": 1396
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7065330794356096e-06,
+      "loss": 2.2773,
+      "step": 1397
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.7003621586518385e-06,
+      "loss": 2.1016,
+      "step": 1398
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.694191695890788e-06,
+      "loss": 2.457,
+      "step": 1399
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.688021700584546e-06,
+      "loss": 2.0137,
+      "step": 1400
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 4.681852182164496e-06,
+      "loss": 2.3008,
+      "step": 1401
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6756831500612846e-06,
+      "loss": 2.3984,
+      "step": 1402
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.669514613704823e-06,
+      "loss": 2.1602,
+      "step": 1403
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.6633465825242555e-06,
+      "loss": 2.043,
+      "step": 1404
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.657179065947959e-06,
+      "loss": 2.166,
+      "step": 1405
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 4.651012073403528e-06,
+      "loss": 1.9375,
+      "step": 1406
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.644845614317745e-06,
+      "loss": 2.3809,
+      "step": 1407
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.638679698116588e-06,
+      "loss": 2.75,
+      "step": 1408
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.632514334225201e-06,
+      "loss": 2.123,
+      "step": 1409
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.626349532067879e-06,
+      "loss": 2.7109,
+      "step": 1410
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.620185301068067e-06,
+      "loss": 2.418,
+      "step": 1411
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 4.614021650648329e-06,
+      "loss": 2.082,
+      "step": 1412
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.607858590230351e-06,
+      "loss": 2.3125,
+      "step": 1413
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.601696129234907e-06,
+      "loss": 1.9863,
+      "step": 1414
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.5955342770818564e-06,
+      "loss": 2.1777,
+      "step": 1415
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.589373043190137e-06,
+      "loss": 2.1289,
+      "step": 1416
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.583212436977731e-06,
+      "loss": 2.4805,
+      "step": 1417
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5770524678616665e-06,
+      "loss": 2.0352,
+      "step": 1418
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.5708931452579975e-06,
+      "loss": 2.3789,
+      "step": 1419
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.564734478581785e-06,
+      "loss": 2.0098,
+      "step": 1420
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.558576477247097e-06,
+      "loss": 2.4648,
+      "step": 1421
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.552419150666973e-06,
+      "loss": 2.209,
+      "step": 1422
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.546262508253429e-06,
+      "loss": 2.3867,
+      "step": 1423
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5401065594174345e-06,
+      "loss": 2.4863,
+      "step": 1424
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.533951313568894e-06,
+      "loss": 2.2031,
+      "step": 1425
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.527796780116642e-06,
+      "loss": 2.0039,
+      "step": 1426
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5216429684684245e-06,
+      "loss": 1.9023,
+      "step": 1427
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.515489888030878e-06,
+      "loss": 1.8789,
+      "step": 1428
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.509337548209529e-06,
+      "loss": 2.25,
+      "step": 1429
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.503185958408767e-06,
+      "loss": 1.9883,
+      "step": 1430
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.497035128031838e-06,
+      "loss": 2.1484,
+      "step": 1431
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4908850664808245e-06,
+      "loss": 2.4297,
+      "step": 1432
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.484735783156634e-06,
+      "loss": 2.2188,
+      "step": 1433
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.478587287458988e-06,
+      "loss": 2.0469,
+      "step": 1434
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.472439588786398e-06,
+      "loss": 2.4297,
+      "step": 1435
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.466292696536162e-06,
+      "loss": 2.1895,
+      "step": 1436
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.460146620104347e-06,
+      "loss": 2.0996,
+      "step": 1437
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.454001368885764e-06,
+      "loss": 2.1367,
+      "step": 1438
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.4478569522739736e-06,
+      "loss": 1.9844,
+      "step": 1439
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.441713379661251e-06,
+      "loss": 2.1797,
+      "step": 1440
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.4355706604385905e-06,
+      "loss": 2.1797,
+      "step": 1441
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.4294288039956735e-06,
+      "loss": 1.877,
+      "step": 1442
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.423287819720866e-06,
+      "loss": 2.0957,
+      "step": 1443
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.417147717001205e-06,
+      "loss": 2.3086,
+      "step": 1444
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.411008505222371e-06,
+      "loss": 1.8418,
+      "step": 1445
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.4048701937686914e-06,
+      "loss": 2.3066,
+      "step": 1446
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.3987327920231124e-06,
+      "loss": 1.8594,
+      "step": 1447
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.392596309367189e-06,
+      "loss": 2.1445,
+      "step": 1448
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.386460755181075e-06,
+      "loss": 2.0,
+      "step": 1449
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.380326138843503e-06,
+      "loss": 2.1758,
+      "step": 1450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.374192469731771e-06,
+      "loss": 2.7188,
+      "step": 1451
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.368059757221731e-06,
+      "loss": 2.0918,
+      "step": 1452
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.3619280106877716e-06,
+      "loss": 2.2109,
+      "step": 1453
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 4.355797239502807e-06,
+      "loss": 2.1367,
+      "step": 1454
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.349667453038256e-06,
+      "loss": 2.3984,
+      "step": 1455
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.343538660664033e-06,
+      "loss": 1.9336,
+      "step": 1456
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.337410871748541e-06,
+      "loss": 1.9941,
+      "step": 1457
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.331284095658637e-06,
+      "loss": 2.0469,
+      "step": 1458
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 4.325158341759638e-06,
+      "loss": 2.2578,
+      "step": 1459
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.319033619415296e-06,
+      "loss": 1.8789,
+      "step": 1460
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.312909937987785e-06,
+      "loss": 2.7383,
+      "step": 1461
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.30678730683769e-06,
+      "loss": 2.2148,
+      "step": 1462
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.3006657353239885e-06,
+      "loss": 2.2754,
+      "step": 1463
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 4.294545232804043e-06,
+      "loss": 2.5059,
+      "step": 1464
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 2.1953,
+      "step": 1465
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.282307472166661e-06,
+      "loss": 1.9824,
+      "step": 1466
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.276190232755718e-06,
+      "loss": 1.7676,
+      "step": 1467
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.270074099751478e-06,
+      "loss": 2.2402,
+      "step": 1468
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 4.26395908250299e-06,
+      "loss": 1.8633,
+      "step": 1469
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.257845190357594e-06,
+      "loss": 2.0391,
+      "step": 1470
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.251732432660909e-06,
+      "loss": 2.4844,
+      "step": 1471
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.245620818756822e-06,
+      "loss": 1.9023,
+      "step": 1472
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.239510357987469e-06,
+      "loss": 2.2559,
+      "step": 1473
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.233401059693231e-06,
+      "loss": 2.2461,
+      "step": 1474
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 4.2272929332127e-06,
+      "loss": 2.293,
+      "step": 1475
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.221185987882684e-06,
+      "loss": 2.0625,
+      "step": 1476
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.215080233038189e-06,
+      "loss": 2.0566,
+      "step": 1477
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.208975678012391e-06,
+      "loss": 1.9648,
+      "step": 1478
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.202872332136639e-06,
+      "loss": 2.4238,
+      "step": 1479
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 4.1967702047404365e-06,
+      "loss": 1.5078,
+      "step": 1480
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.190669305151414e-06,
+      "loss": 2.2578,
+      "step": 1481
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.184569642695337e-06,
+      "loss": 2.3906,
+      "step": 1482
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.178471226696073e-06,
+      "loss": 2.4922,
+      "step": 1483
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.172374066475583e-06,
+      "loss": 2.2129,
+      "step": 1484
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 4.166278171353915e-06,
+      "loss": 1.8691,
+      "step": 1485
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.160183550649176e-06,
+      "loss": 2.0039,
+      "step": 1486
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.154090213677531e-06,
+      "loss": 2.082,
+      "step": 1487
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.147998169753179e-06,
+      "loss": 2.1367,
+      "step": 1488
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.141907428188341e-06,
+      "loss": 2.5645,
+      "step": 1489
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 4.1358179982932535e-06,
+      "loss": 2.5586,
+      "step": 1490
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.129729889376141e-06,
+      "loss": 1.8691,
+      "step": 1491
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.123643110743214e-06,
+      "loss": 1.8613,
+      "step": 1492
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.117557671698648e-06,
+      "loss": 2.3652,
+      "step": 1493
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.111473581544566e-06,
+      "loss": 2.668,
+      "step": 1494
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.105390849581037e-06,
+      "loss": 2.0977,
+      "step": 1495
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.09930948510605e-06,
+      "loss": 2.3047,
+      "step": 1496
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.093229497415504e-06,
+      "loss": 2.1328,
+      "step": 1497
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.087150895803192e-06,
+      "loss": 2.1875,
+      "step": 1498
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.08107368956079e-06,
+      "loss": 2.1523,
+      "step": 1499
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.074997887977843e-06,
+      "loss": 2.2227,
+      "step": 1500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.0689235003417425e-06,
+      "loss": 2.2109,
+      "step": 1501
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.062850535937726e-06,
+      "loss": 1.8398,
+      "step": 1502
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.056779004048852e-06,
+      "loss": 1.8418,
+      "step": 1503
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.050708913955987e-06,
+      "loss": 2.2266,
+      "step": 1504
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.044640274937798e-06,
+      "loss": 1.8711,
+      "step": 1505
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.0385730962707325e-06,
+      "loss": 2.1484,
+      "step": 1506
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.032507387229002e-06,
+      "loss": 1.959,
+      "step": 1507
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.026443157084579e-06,
+      "loss": 2.4805,
+      "step": 1508
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.020380415107167e-06,
+      "loss": 2.0273,
+      "step": 1509
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.014319170564202e-06,
+      "loss": 2.2754,
+      "step": 1510
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.008259432720827e-06,
+      "loss": 2.1641,
+      "step": 1511
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.002201210839881e-06,
+      "loss": 2.1484,
+      "step": 1512
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.996144514181891e-06,
+      "loss": 2.043,
+      "step": 1513
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9900893520050446e-06,
+      "loss": 1.9805,
+      "step": 1514
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.9840357335651914e-06,
+      "loss": 2.1738,
+      "step": 1515
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.977983668115818e-06,
+      "loss": 2.2656,
+      "step": 1516
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.971933164908034e-06,
+      "loss": 2.0703,
+      "step": 1517
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9658842331905675e-06,
+      "loss": 1.8242,
+      "step": 1518
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.959836882209739e-06,
+      "loss": 2.1406,
+      "step": 1519
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.953791121209458e-06,
+      "loss": 2.168,
+      "step": 1520
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.9477469594311975e-06,
+      "loss": 1.9297,
+      "step": 1521
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.941704406113989e-06,
+      "loss": 2.4629,
+      "step": 1522
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.9356634704944065e-06,
+      "loss": 2.1523,
+      "step": 1523
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.929624161806548e-06,
+      "loss": 2.0664,
+      "step": 1524
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.923586489282028e-06,
+      "loss": 2.1445,
+      "step": 1525
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.917550462149959e-06,
+      "loss": 2.0938,
+      "step": 1526
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.911516089636936e-06,
+      "loss": 2.7891,
+      "step": 1527
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.905483380967027e-06,
+      "loss": 2.3867,
+      "step": 1528
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.899452345361756e-06,
+      "loss": 2.3594,
+      "step": 1529
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.893422992040092e-06,
+      "loss": 1.9609,
+      "step": 1530
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.887395330218429e-06,
+      "loss": 2.1992,
+      "step": 1531
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.881369369110575e-06,
+      "loss": 2.1035,
+      "step": 1532
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.8753451179277425e-06,
+      "loss": 2.1973,
+      "step": 1533
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.869322585878527e-06,
+      "loss": 1.7754,
+      "step": 1534
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.863301782168896e-06,
+      "loss": 1.8594,
+      "step": 1535
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.857282716002178e-06,
+      "loss": 2.1445,
+      "step": 1536
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.8512653965790395e-06,
+      "loss": 1.9414,
+      "step": 1537
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.845249833097486e-06,
+      "loss": 2.293,
+      "step": 1538
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8392360347528315e-06,
+      "loss": 2.0273,
+      "step": 1539
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.8332240107376925e-06,
+      "loss": 1.8223,
+      "step": 1540
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.827213770241977e-06,
+      "loss": 2.2422,
+      "step": 1541
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.821205322452863e-06,
+      "loss": 2.1055,
+      "step": 1542
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.815198676554794e-06,
+      "loss": 2.3008,
+      "step": 1543
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.8091938417294515e-06,
+      "loss": 2.1719,
+      "step": 1544
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.8031908271557504e-06,
+      "loss": 1.8359,
+      "step": 1545
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7971896420098296e-06,
+      "loss": 1.9648,
+      "step": 1546
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7911902954650226e-06,
+      "loss": 1.9512,
+      "step": 1547
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.7851927966918602e-06,
+      "loss": 1.9727,
+      "step": 1548
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.779197154858044e-06,
+      "loss": 2.1348,
+      "step": 1549
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.7732033791284367e-06,
+      "loss": 2.2422,
+      "step": 1550
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.767211478665054e-06,
+      "loss": 2.1758,
+      "step": 1551
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.761221462627034e-06,
+      "loss": 2.082,
+      "step": 1552
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.7552333401706508e-06,
+      "loss": 1.5742,
+      "step": 1553
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.74924712044927e-06,
+      "loss": 1.9512,
+      "step": 1554
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7432628126133497e-06,
+      "loss": 2.0332,
+      "step": 1555
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7372804258104367e-06,
+      "loss": 2.2207,
+      "step": 1556
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.731299969185131e-06,
+      "loss": 2.0,
+      "step": 1557
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.7253214518790814e-06,
+      "loss": 1.6973,
+      "step": 1558
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7193448830309802e-06,
+      "loss": 2.2012,
+      "step": 1559
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7133702717765345e-06,
+      "loss": 2.0195,
+      "step": 1560
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7073976272484647e-06,
+      "loss": 1.7637,
+      "step": 1561
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.7014269585764796e-06,
+      "loss": 2.4062,
+      "step": 1562
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.695458274887268e-06,
+      "loss": 1.9609,
+      "step": 1563
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.689491585304491e-06,
+      "loss": 1.8301,
+      "step": 1564
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6835268989487518e-06,
+      "loss": 1.9824,
+      "step": 1565
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.6775642249376e-06,
+      "loss": 1.7227,
+      "step": 1566
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.671603572385506e-06,
+      "loss": 2.3203,
+      "step": 1567
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.665644950403847e-06,
+      "loss": 2.0039,
+      "step": 1568
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.659688368100902e-06,
+      "loss": 1.7148,
+      "step": 1569
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.6537338345818273e-06,
+      "loss": 2.2773,
+      "step": 1570
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.647781358948653e-06,
+      "loss": 2.2031,
+      "step": 1571
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.641830950300257e-06,
+      "loss": 1.9727,
+      "step": 1572
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.63588261773236e-06,
+      "loss": 1.9629,
+      "step": 1573
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.629936370337514e-06,
+      "loss": 1.9453,
+      "step": 1574
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.623992217205075e-06,
+      "loss": 2.2227,
+      "step": 1575
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.6180501674212052e-06,
+      "loss": 2.0176,
+      "step": 1576
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.6121102300688504e-06,
+      "loss": 1.7949,
+      "step": 1577
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.606172414227721e-06,
+      "loss": 2.082,
+      "step": 1578
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.600236728974292e-06,
+      "loss": 1.8633,
+      "step": 1579
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5943031833817788e-06,
+      "loss": 1.8027,
+      "step": 1580
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.588371786520127e-06,
+      "loss": 1.7207,
+      "step": 1581
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5824425474559956e-06,
+      "loss": 1.8418,
+      "step": 1582
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.576515475252745e-06,
+      "loss": 1.916,
+      "step": 1583
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5705905789704296e-06,
+      "loss": 2.3203,
+      "step": 1584
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.5646678676657687e-06,
+      "loss": 2.0137,
+      "step": 1585
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.558747350392146e-06,
+      "loss": 1.7422,
+      "step": 1586
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.552829036199595e-06,
+      "loss": 2.1367,
+      "step": 1587
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.546912934134773e-06,
+      "loss": 2.2422,
+      "step": 1588
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.540999053240965e-06,
+      "loss": 1.8398,
+      "step": 1589
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.5350874025580555e-06,
+      "loss": 2.2266,
+      "step": 1590
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.529177991122519e-06,
+      "loss": 1.7422,
+      "step": 1591
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.523270827967411e-06,
+      "loss": 2.1855,
+      "step": 1592
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.517365922122348e-06,
+      "loss": 2.1504,
+      "step": 1593
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.5114632826134977e-06,
+      "loss": 2.1152,
+      "step": 1594
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.50556291846356e-06,
+      "loss": 2.209,
+      "step": 1595
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4996648386917585e-06,
+      "loss": 1.9492,
+      "step": 1596
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4937690523138302e-06,
+      "loss": 1.75,
+      "step": 1597
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.487875568341995e-06,
+      "loss": 1.8457,
+      "step": 1598
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4819843957849655e-06,
+      "loss": 1.8887,
+      "step": 1599
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.4760955436479134e-06,
+      "loss": 1.7871,
+      "step": 1600
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4702090209324634e-06,
+      "loss": 1.8438,
+      "step": 1601
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4643248366366853e-06,
+      "loss": 1.8047,
+      "step": 1602
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4584429997550685e-06,
+      "loss": 2.1133,
+      "step": 1603
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.45256351927852e-06,
+      "loss": 1.998,
+      "step": 1604
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.446686404194337e-06,
+      "loss": 2.0508,
+      "step": 1605
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.440811663486206e-06,
+      "loss": 1.8574,
+      "step": 1606
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.434939306134186e-06,
+      "loss": 1.8809,
+      "step": 1607
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4290693411146882e-06,
+      "loss": 2.3828,
+      "step": 1608
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.423201777400469e-06,
+      "loss": 2.2812,
+      "step": 1609
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4173366239606153e-06,
+      "loss": 1.8086,
+      "step": 1610
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.4114738897605267e-06,
+      "loss": 1.9492,
+      "step": 1611
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.4056135837619077e-06,
+      "loss": 2.0117,
+      "step": 1612
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.399755714922751e-06,
+      "loss": 2.0801,
+      "step": 1613
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3939002921973197e-06,
+      "loss": 2.0508,
+      "step": 1614
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.388047324536143e-06,
+      "loss": 1.9785,
+      "step": 1615
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.3821968208859955e-06,
+      "loss": 2.0195,
+      "step": 1616
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.376348790189886e-06,
+      "loss": 1.8926,
+      "step": 1617
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.3705032413870402e-06,
+      "loss": 2.0156,
+      "step": 1618
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.3646601834128924e-06,
+      "loss": 2.3789,
+      "step": 1619
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.358819625199071e-06,
+      "loss": 1.9609,
+      "step": 1620
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 3.3529815756733773e-06,
+      "loss": 1.7461,
+      "step": 1621
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.347146043759787e-06,
+      "loss": 1.7891,
+      "step": 1622
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.34131303837842e-06,
+      "loss": 1.9512,
+      "step": 1623
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.335482568445536e-06,
+      "loss": 1.7422,
+      "step": 1624
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.3296546428735206e-06,
+      "loss": 2.0137,
+      "step": 1625
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.3238292705708675e-06,
+      "loss": 1.7012,
+      "step": 1626
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.318006460442173e-06,
+      "loss": 1.7793,
+      "step": 1627
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.3121862213881093e-06,
+      "loss": 1.9805,
+      "step": 1628
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.306368562305422e-06,
+      "loss": 1.9141,
+      "step": 1629
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.3005534920869175e-06,
+      "loss": 1.9004,
+      "step": 1630
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.2947410196214353e-06,
+      "loss": 1.8418,
+      "step": 1631
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.288931153793853e-06,
+      "loss": 1.916,
+      "step": 1632
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2831239034850593e-06,
+      "loss": 1.7715,
+      "step": 1633
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2773192775719432e-06,
+      "loss": 2.6055,
+      "step": 1634
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.2715172849273867e-06,
+      "loss": 2.0254,
+      "step": 1635
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.265717934420244e-06,
+      "loss": 1.957,
+      "step": 1636
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.259921234915332e-06,
+      "loss": 2.3477,
+      "step": 1637
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.254127195273412e-06,
+      "loss": 2.4023,
+      "step": 1638
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.248335824351183e-06,
+      "loss": 2.4297,
+      "step": 1639
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2425471310012645e-06,
+      "loss": 2.2598,
+      "step": 1640
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.2367611240721796e-06,
+      "loss": 1.8008,
+      "step": 1641
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.230977812408348e-06,
+      "loss": 1.7539,
+      "step": 1642
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.22519720485007e-06,
+      "loss": 1.9453,
+      "step": 1643
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.219419310233509e-06,
+      "loss": 2.1543,
+      "step": 1644
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.213644137390686e-06,
+      "loss": 1.9141,
+      "step": 1645
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.207871695149459e-06,
+      "loss": 1.918,
+      "step": 1646
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.2021019923335093e-06,
+      "loss": 2.0293,
+      "step": 1647
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.196335037762337e-06,
+      "loss": 1.8535,
+      "step": 1648
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1905708402512347e-06,
+      "loss": 1.6699,
+      "step": 1649
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1848094086112887e-06,
+      "loss": 1.6055,
+      "step": 1650
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.1790507516493473e-06,
+      "loss": 1.5332,
+      "step": 1651
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.173294878168025e-06,
+      "loss": 1.7734,
+      "step": 1652
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.167541796965681e-06,
+      "loss": 1.8184,
+      "step": 1653
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1617915168363994e-06,
+      "loss": 1.7637,
+      "step": 1654
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1560440465699924e-06,
+      "loss": 1.9121,
+      "step": 1655
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.150299394951972e-06,
+      "loss": 1.9883,
+      "step": 1656
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1445575707635383e-06,
+      "loss": 2.0039,
+      "step": 1657
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.1388185827815766e-06,
+      "loss": 1.8848,
+      "step": 1658
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.1330824397786317e-06,
+      "loss": 2.3242,
+      "step": 1659
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.127349150522904e-06,
+      "loss": 1.8379,
+      "step": 1660
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.121618723778225e-06,
+      "loss": 2.043,
+      "step": 1661
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.115891168304055e-06,
+      "loss": 2.1191,
+      "step": 1662
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.110166492855468e-06,
+      "loss": 1.8203,
+      "step": 1663
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.104444706183129e-06,
+      "loss": 1.7715,
+      "step": 1664
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.0987258170332935e-06,
+      "loss": 1.7676,
+      "step": 1665
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.093009834147784e-06,
+      "loss": 1.7773,
+      "step": 1666
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.0872967662639795e-06,
+      "loss": 1.8105,
+      "step": 1667
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.081586622114809e-06,
+      "loss": 2.2109,
+      "step": 1668
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0758794104287275e-06,
+      "loss": 1.75,
+      "step": 1669
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0701751399297063e-06,
+      "loss": 1.8945,
+      "step": 1670
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0644738193372246e-06,
+      "loss": 1.7285,
+      "step": 1671
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0587754573662508e-06,
+      "loss": 1.8477,
+      "step": 1672
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.053080062727233e-06,
+      "loss": 2.0898,
+      "step": 1673
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.0473876441260786e-06,
+      "loss": 1.791,
+      "step": 1674
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.041698210264149e-06,
+      "loss": 1.7559,
+      "step": 1675
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0360117698382446e-06,
+      "loss": 2.0508,
+      "step": 1676
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0303283315405853e-06,
+      "loss": 1.6387,
+      "step": 1677
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0246479040588077e-06,
+      "loss": 1.9336,
+      "step": 1678
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.0189704960759436e-06,
+      "loss": 1.7461,
+      "step": 1679
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.0132961162704054e-06,
+      "loss": 1.9805,
+      "step": 1680
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.0076247733159846e-06,
+      "loss": 2.1934,
+      "step": 1681
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.001956475881822e-06,
+      "loss": 1.8789,
+      "step": 1682
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.996291232632412e-06,
+      "loss": 2.1172,
+      "step": 1683
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 2.9906290522275738e-06,
+      "loss": 1.7324,
+      "step": 1684
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9849699433224423e-06,
+      "loss": 1.7383,
+      "step": 1685
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.97931391456747e-06,
+      "loss": 2.0312,
+      "step": 1686
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.9736609746083853e-06,
+      "loss": 1.9336,
+      "step": 1687
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.968011132086209e-06,
+      "loss": 1.6602,
+      "step": 1688
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 2.962364395637216e-06,
+      "loss": 1.6191,
+      "step": 1689
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.9567207738929395e-06,
+      "loss": 2.0664,
+      "step": 1690
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.9510802754801526e-06,
+      "loss": 1.9492,
+      "step": 1691
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.945442909020847e-06,
+      "loss": 1.9824,
+      "step": 1692
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.939808683132238e-06,
+      "loss": 2.1641,
+      "step": 1693
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.93417760642673e-06,
+      "loss": 1.5586,
+      "step": 1694
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 2.928549687511917e-06,
+      "loss": 1.9844,
+      "step": 1695
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 1.6543,
+      "step": 1696
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9173033574606112e-06,
+      "loss": 2.0371,
+      "step": 1697
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.911684963515119e-06,
+      "loss": 2.0781,
+      "step": 1698
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.906069761742302e-06,
+      "loss": 1.6445,
+      "step": 1699
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 2.9004577607254847e-06,
+      "loss": 1.6621,
+      "step": 1700
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.8948489690431062e-06,
+      "loss": 1.9336,
+      "step": 1701
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.889243395268696e-06,
+      "loss": 1.8574,
+      "step": 1702
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.8836410479708625e-06,
+      "loss": 2.0684,
+      "step": 1703
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.8780419357132864e-06,
+      "loss": 2.1426,
+      "step": 1704
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 2.872446067054703e-06,
+      "loss": 1.6641,
+      "step": 1705
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.8668534505488887e-06,
+      "loss": 2.3457,
+      "step": 1706
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.861264094744647e-06,
+      "loss": 1.9395,
+      "step": 1707
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.8556780081857966e-06,
+      "loss": 1.7344,
+      "step": 1708
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.8500951994111647e-06,
+      "loss": 1.8379,
+      "step": 1709
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 2.84451567695456e-06,
+      "loss": 1.9551,
+      "step": 1710
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8389394493447732e-06,
+      "loss": 1.9043,
+      "step": 1711
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.833366525105561e-06,
+      "loss": 1.6875,
+      "step": 1712
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8277969127556204e-06,
+      "loss": 2.0605,
+      "step": 1713
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8222306208085988e-06,
+      "loss": 2.0801,
+      "step": 1714
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.8166676577730556e-06,
+      "loss": 1.6875,
+      "step": 1715
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 2.811108032152473e-06,
+      "loss": 1.8828,
+      "step": 1716
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.805551752445222e-06,
+      "loss": 1.8145,
+      "step": 1717
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.7999988271445643e-06,
+      "loss": 1.8516,
+      "step": 1718
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.794449264738637e-06,
+      "loss": 2.043,
+      "step": 1719
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.7889030737104283e-06,
+      "loss": 1.8359,
+      "step": 1720
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 2.7833602625377803e-06,
+      "loss": 1.9395,
+      "step": 1721
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.7778208396933657e-06,
+      "loss": 1.7188,
+      "step": 1722
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.772284813644675e-06,
+      "loss": 1.9023,
+      "step": 1723
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.766752192854012e-06,
+      "loss": 1.8965,
+      "step": 1724
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.7612229857784743e-06,
+      "loss": 1.6602,
+      "step": 1725
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 2.755697200869937e-06,
+      "loss": 2.2227,
+      "step": 1726
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.7501748465750498e-06,
+      "loss": 1.5137,
+      "step": 1727
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.7446559313352117e-06,
+      "loss": 1.9082,
+      "step": 1728
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.7391404635865725e-06,
+      "loss": 1.6367,
+      "step": 1729
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.7336284517600053e-06,
+      "loss": 1.627,
+      "step": 1730
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 2.728119904281105e-06,
+      "loss": 2.1797,
+      "step": 1731
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.7226148295701704e-06,
+      "loss": 1.8145,
+      "step": 1732
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.7171132360421894e-06,
+      "loss": 1.9727,
+      "step": 1733
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.711615132106832e-06,
+      "loss": 1.6035,
+      "step": 1734
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.7061205261684307e-06,
+      "loss": 1.7656,
+      "step": 1735
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.7006294266259716e-06,
+      "loss": 1.9453,
+      "step": 1736
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 2.6951418418730825e-06,
+      "loss": 2.1582,
+      "step": 1737
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.6896577802980184e-06,
+      "loss": 1.9199,
+      "step": 1738
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.6841772502836518e-06,
+      "loss": 1.6895,
+      "step": 1739
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.678700260207449e-06,
+      "loss": 1.6836,
+      "step": 1740
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.6732268184414695e-06,
+      "loss": 1.5625,
+      "step": 1741
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 2.667756933352352e-06,
+      "loss": 1.8555,
+      "step": 1742
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.6622906133012926e-06,
+      "loss": 1.9785,
+      "step": 1743
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.6568278666440417e-06,
+      "loss": 1.9102,
+      "step": 1744
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.651368701730889e-06,
+      "loss": 1.709,
+      "step": 1745
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.645913126906644e-06,
+      "loss": 1.7227,
+      "step": 1746
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 2.6404611505106337e-06,
+      "loss": 1.6504,
+      "step": 1747
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.6350127808766786e-06,
+      "loss": 1.5859,
+      "step": 1748
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.6295680263330936e-06,
+      "loss": 1.8262,
+      "step": 1749
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.6241268952026592e-06,
+      "loss": 2.168,
+      "step": 1750
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.6186893958026245e-06,
+      "loss": 1.8457,
+      "step": 1751
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 2.6132555364446856e-06,
+      "loss": 1.9629,
+      "step": 1752
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.6078253254349706e-06,
+      "loss": 1.6523,
+      "step": 1753
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.6023987710740317e-06,
+      "loss": 1.9082,
+      "step": 1754
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.5969758816568367e-06,
+      "loss": 1.625,
+      "step": 1755
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.591556665472744e-06,
+      "loss": 1.8828,
+      "step": 1756
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.5861411308055038e-06,
+      "loss": 1.9785,
+      "step": 1757
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 2.5807292859332354e-06,
+      "loss": 1.6348,
+      "step": 1758
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.5753211391284172e-06,
+      "loss": 1.7832,
+      "step": 1759
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.5699166986578773e-06,
+      "loss": 1.7305,
+      "step": 1760
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.564515972782774e-06,
+      "loss": 2.1309,
+      "step": 1761
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.559118969758595e-06,
+      "loss": 1.7227,
+      "step": 1762
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 2.5537256978351276e-06,
+      "loss": 1.8027,
+      "step": 1763
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.548336165256463e-06,
+      "loss": 1.8398,
+      "step": 1764
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.5429503802609777e-06,
+      "loss": 1.7578,
+      "step": 1765
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.537568351081311e-06,
+      "loss": 1.8926,
+      "step": 1766
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.5321900859443714e-06,
+      "loss": 1.7754,
+      "step": 1767
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 2.526815593071306e-06,
+      "loss": 1.7227,
+      "step": 1768
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.521444880677497e-06,
+      "loss": 1.6641,
+      "step": 1769
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5160779569725514e-06,
+      "loss": 1.7227,
+      "step": 1770
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5107148301602825e-06,
+      "loss": 1.9141,
+      "step": 1771
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5053555084387022e-06,
+      "loss": 2.4453,
+      "step": 1772
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 2.5000000000000015e-06,
+      "loss": 1.5547,
+      "step": 1773
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.494648313030542e-06,
+      "loss": 2.0273,
+      "step": 1774
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.4893004557108526e-06,
+      "loss": 1.9199,
+      "step": 1775
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.483956436215596e-06,
+      "loss": 1.4609,
+      "step": 1776
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.478616262713578e-06,
+      "loss": 2.1016,
+      "step": 1777
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 2.4732799433677236e-06,
+      "loss": 2.1172,
+      "step": 1778
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.467947486335061e-06,
+      "loss": 1.748,
+      "step": 1779
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4626188997667224e-06,
+      "loss": 1.8828,
+      "step": 1780
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4572941918079178e-06,
+      "loss": 1.959,
+      "step": 1781
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4519733705979283e-06,
+      "loss": 1.7246,
+      "step": 1782
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4466564442700974e-06,
+      "loss": 1.8086,
+      "step": 1783
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 2.4413434209518137e-06,
+      "loss": 2.0527,
+      "step": 1784
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.436034308764501e-06,
+      "loss": 1.8027,
+      "step": 1785
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.430729115823601e-06,
+      "loss": 1.791,
+      "step": 1786
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.425427850238565e-06,
+      "loss": 1.7812,
+      "step": 1787
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.4201305201128444e-06,
+      "loss": 1.7129,
+      "step": 1788
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 2.414837133543871e-06,
+      "loss": 1.75,
+      "step": 1789
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.4095476986230516e-06,
+      "loss": 2.1016,
+      "step": 1790
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.4042622234357526e-06,
+      "loss": 1.6621,
+      "step": 1791
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.398980716061284e-06,
+      "loss": 1.9062,
+      "step": 1792
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.393703184572895e-06,
+      "loss": 1.5879,
+      "step": 1793
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 2.388429637037753e-06,
+      "loss": 1.8223,
+      "step": 1794
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.383160081516941e-06,
+      "loss": 1.5742,
+      "step": 1795
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.377894526065433e-06,
+      "loss": 1.6035,
+      "step": 1796
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.3726329787320943e-06,
+      "loss": 1.7188,
+      "step": 1797
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.3673754475596634e-06,
+      "loss": 1.9492,
+      "step": 1798
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 2.3621219405847333e-06,
+      "loss": 1.5352,
+      "step": 1799
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.356872465837755e-06,
+      "loss": 1.8887,
+      "step": 1800
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.3516270313430085e-06,
+      "loss": 2.375,
+      "step": 1801
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.3463856451185986e-06,
+      "loss": 1.6836,
+      "step": 1802
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.341148315176446e-06,
+      "loss": 1.9219,
+      "step": 1803
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.335915049522268e-06,
+      "loss": 1.873,
+      "step": 1804
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 2.3306858561555728e-06,
+      "loss": 1.6094,
+      "step": 1805
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.3254607430696393e-06,
+      "loss": 1.8359,
+      "step": 1806
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.320239718251508e-06,
+      "loss": 1.5273,
+      "step": 1807
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.3150227896819782e-06,
+      "loss": 1.9531,
+      "step": 1808
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.3098099653355804e-06,
+      "loss": 2.3438,
+      "step": 1809
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 2.30460125318057e-06,
+      "loss": 1.6367,
+      "step": 1810
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.2993966611789277e-06,
+      "loss": 1.7207,
+      "step": 1811
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.294196197286323e-06,
+      "loss": 1.8047,
+      "step": 1812
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.2889998694521257e-06,
+      "loss": 1.8164,
+      "step": 1813
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.2838076856193758e-06,
+      "loss": 1.9668,
+      "step": 1814
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 2.278619653724781e-06,
+      "loss": 2.3105,
+      "step": 1815
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2734357816987046e-06,
+      "loss": 1.7031,
+      "step": 1816
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2682560774651458e-06,
+      "loss": 1.9805,
+      "step": 1817
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2630805489417437e-06,
+      "loss": 1.668,
+      "step": 1818
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.2579092040397436e-06,
+      "loss": 1.957,
+      "step": 1819
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 2.252742050663998e-06,
+      "loss": 1.6973,
+      "step": 1820
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.247579096712958e-06,
+      "loss": 1.8828,
+      "step": 1821
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.2424203500786473e-06,
+      "loss": 1.5762,
+      "step": 1822
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.2372658186466665e-06,
+      "loss": 1.7617,
+      "step": 1823
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.232115510296165e-06,
+      "loss": 1.3457,
+      "step": 1824
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.226969432899843e-06,
+      "loss": 1.7812,
+      "step": 1825
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 2.2218275943239333e-06,
+      "loss": 1.75,
+      "step": 1826
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.2166900024281818e-06,
+      "loss": 1.582,
+      "step": 1827
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.211556665065854e-06,
+      "loss": 1.8906,
+      "step": 1828
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.206427590083703e-06,
+      "loss": 1.5781,
+      "step": 1829
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.201302785321968e-06,
+      "loss": 1.7559,
+      "step": 1830
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 2.196182258614365e-06,
+      "loss": 1.6855,
+      "step": 1831
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.1910660177880675e-06,
+      "loss": 1.668,
+      "step": 1832
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.1859540706637007e-06,
+      "loss": 2.1133,
+      "step": 1833
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.180846425055322e-06,
+      "loss": 1.998,
+      "step": 1834
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.1757430887704135e-06,
+      "loss": 1.7363,
+      "step": 1835
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 2.170644069609876e-06,
+      "loss": 1.4414,
+      "step": 1836
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.165549375368004e-06,
+      "loss": 1.7773,
+      "step": 1837
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.1604590138324866e-06,
+      "loss": 1.4805,
+      "step": 1838
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.1553729927843894e-06,
+      "loss": 1.8867,
+      "step": 1839
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.150291319998137e-06,
+      "loss": 1.7051,
+      "step": 1840
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 2.1452140032415174e-06,
+      "loss": 1.8594,
+      "step": 1841
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1401410502756513e-06,
+      "loss": 1.7109,
+      "step": 1842
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1350724688549906e-06,
+      "loss": 1.748,
+      "step": 1843
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1300082667273087e-06,
+      "loss": 1.5469,
+      "step": 1844
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.1249484516336826e-06,
+      "loss": 1.6914,
+      "step": 1845
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.119893031308485e-06,
+      "loss": 1.8438,
+      "step": 1846
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 2.114842013479367e-06,
+      "loss": 1.8203,
+      "step": 1847
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1097954058672515e-06,
+      "loss": 1.8008,
+      "step": 1848
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.1047532161863242e-06,
+      "loss": 2.1484,
+      "step": 1849
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.09971545214401e-06,
+      "loss": 2.1914,
+      "step": 1850
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.094682121440975e-06,
+      "loss": 1.6562,
+      "step": 1851
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 2.0896532317711087e-06,
+      "loss": 2.0391,
+      "step": 1852
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0846287908215063e-06,
+      "loss": 1.3555,
+      "step": 1853
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0796088062724697e-06,
+      "loss": 1.8125,
+      "step": 1854
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0745932857974825e-06,
+      "loss": 1.8379,
+      "step": 1855
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0695822370632106e-06,
+      "loss": 1.8398,
+      "step": 1856
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 2.0645756677294788e-06,
+      "loss": 1.748,
+      "step": 1857
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.0595735854492675e-06,
+      "loss": 1.5469,
+      "step": 1858
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.054575997868702e-06,
+      "loss": 1.6797,
+      "step": 1859
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.0495829126270277e-06,
+      "loss": 1.9297,
+      "step": 1860
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.0445943373566178e-06,
+      "loss": 1.9473,
+      "step": 1861
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 2.0396102796829458e-06,
+      "loss": 1.8535,
+      "step": 1862
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.034630747224578e-06,
+      "loss": 1.8945,
+      "step": 1863
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.029655747593169e-06,
+      "loss": 1.5664,
+      "step": 1864
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.0246852883934437e-06,
+      "loss": 1.9004,
+      "step": 1865
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.019719377223181e-06,
+      "loss": 2.0977,
+      "step": 1866
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.014758021673216e-06,
+      "loss": 1.6836,
+      "step": 1867
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 2.009801229327411e-06,
+      "loss": 1.9492,
+      "step": 1868
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 2.0048490077626624e-06,
+      "loss": 1.5137,
+      "step": 1869
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9999013645488714e-06,
+      "loss": 1.5586,
+      "step": 1870
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9949583072489455e-06,
+      "loss": 1.8789,
+      "step": 1871
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9900198434187838e-06,
+      "loss": 1.5723,
+      "step": 1872
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.9850859806072576e-06,
+      "loss": 1.5469,
+      "step": 1873
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9801567263562126e-06,
+      "loss": 1.8535,
+      "step": 1874
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9752320882004445e-06,
+      "loss": 1.5352,
+      "step": 1875
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.970312073667693e-06,
+      "loss": 1.4219,
+      "step": 1876
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9653966902786327e-06,
+      "loss": 1.6973,
+      "step": 1877
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.9604859455468587e-06,
+      "loss": 1.9082,
+      "step": 1878
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9555798469788773e-06,
+      "loss": 1.707,
+      "step": 1879
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9506784020740875e-06,
+      "loss": 1.5312,
+      "step": 1880
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9457816183247767e-06,
+      "loss": 1.9805,
+      "step": 1881
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9408895032161116e-06,
+      "loss": 1.7148,
+      "step": 1882
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.9360020642261163e-06,
+      "loss": 1.5566,
+      "step": 1883
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.93111930882567e-06,
+      "loss": 1.5195,
+      "step": 1884
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.926241244478496e-06,
+      "loss": 1.8887,
+      "step": 1885
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9213678786411392e-06,
+      "loss": 1.9609,
+      "step": 1886
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9164992187629705e-06,
+      "loss": 1.8848,
+      "step": 1887
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.9116352722861596e-06,
+      "loss": 1.4902,
+      "step": 1888
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.906776046645679e-06,
+      "loss": 1.584,
+      "step": 1889
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.9019215492692778e-06,
+      "loss": 1.8027,
+      "step": 1890
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.897071787577482e-06,
+      "loss": 1.5547,
+      "step": 1891
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.8922267689835806e-06,
+      "loss": 1.748,
+      "step": 1892
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.8873865008936065e-06,
+      "loss": 1.8105,
+      "step": 1893
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.8825509907063328e-06,
+      "loss": 1.6797,
+      "step": 1894
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8777202458132644e-06,
+      "loss": 1.873,
+      "step": 1895
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8728942735986149e-06,
+      "loss": 1.7754,
+      "step": 1896
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8680730814393084e-06,
+      "loss": 1.5918,
+      "step": 1897
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8632566767049603e-06,
+      "loss": 1.5664,
+      "step": 1898
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.8584450667578656e-06,
+      "loss": 1.8047,
+      "step": 1899
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8536382589529944e-06,
+      "loss": 1.7871,
+      "step": 1900
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8488362606379713e-06,
+      "loss": 1.793,
+      "step": 1901
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.844039079153075e-06,
+      "loss": 1.875,
+      "step": 1902
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.839246721831215e-06,
+      "loss": 1.6758,
+      "step": 1903
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.8344591959979302e-06,
+      "loss": 1.666,
+      "step": 1904
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.829676508971377e-06,
+      "loss": 1.4531,
+      "step": 1905
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8248986680623077e-06,
+      "loss": 1.6836,
+      "step": 1906
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8201256805740752e-06,
+      "loss": 2.0957,
+      "step": 1907
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8153575538026064e-06,
+      "loss": 1.4141,
+      "step": 1908
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.8105942950364008e-06,
+      "loss": 2.1035,
+      "step": 1909
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.805835911556519e-06,
+      "loss": 1.4785,
+      "step": 1910
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.801082410636566e-06,
+      "loss": 1.5762,
+      "step": 1911
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7963337995426876e-06,
+      "loss": 1.709,
+      "step": 1912
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7915900855335506e-06,
+      "loss": 1.8398,
+      "step": 1913
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7868512758603362e-06,
+      "loss": 1.502,
+      "step": 1914
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.7821173777667333e-06,
+      "loss": 1.4922,
+      "step": 1915
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7773883984889178e-06,
+      "loss": 1.7559,
+      "step": 1916
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7726643452555504e-06,
+      "loss": 1.7441,
+      "step": 1917
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7679452252877622e-06,
+      "loss": 1.5488,
+      "step": 1918
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7632310457991387e-06,
+      "loss": 1.9102,
+      "step": 1919
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.7585218139957205e-06,
+      "loss": 1.5156,
+      "step": 1920
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7538175370759797e-06,
+      "loss": 1.5254,
+      "step": 1921
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7491182222308139e-06,
+      "loss": 1.7617,
+      "step": 1922
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.74442387664354e-06,
+      "loss": 2.1562,
+      "step": 1923
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.739734507489877e-06,
+      "loss": 1.4785,
+      "step": 1924
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.7350501219379384e-06,
+      "loss": 1.8398,
+      "step": 1925
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7303707271482161e-06,
+      "loss": 1.5859,
+      "step": 1926
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 1.7734,
+      "step": 1927
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7210269384592433e-06,
+      "loss": 1.5098,
+      "step": 1928
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7163625588427918e-06,
+      "loss": 1.6699,
+      "step": 1929
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.7117031985541366e-06,
+      "loss": 1.7754,
+      "step": 1930
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.7070488647155204e-06,
+      "loss": 1.6484,
+      "step": 1931
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.702399564441496e-06,
+      "loss": 1.5918,
+      "step": 1932
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6977553048389306e-06,
+      "loss": 1.7129,
+      "step": 1933
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6931160930069789e-06,
+      "loss": 1.5332,
+      "step": 1934
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6884819360370868e-06,
+      "loss": 1.6562,
+      "step": 1935
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.6838528410129662e-06,
+      "loss": 1.6426,
+      "step": 1936
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6792288150105957e-06,
+      "loss": 1.4512,
+      "step": 1937
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6746098650982072e-06,
+      "loss": 1.4473,
+      "step": 1938
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6699959983362668e-06,
+      "loss": 1.8594,
+      "step": 1939
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.665387221777478e-06,
+      "loss": 1.5254,
+      "step": 1940
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.6607835424667578e-06,
+      "loss": 1.4707,
+      "step": 1941
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.6561849674412327e-06,
+      "loss": 1.625,
+      "step": 1942
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.6515915037302298e-06,
+      "loss": 2.0332,
+      "step": 1943
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.647003158355262e-06,
+      "loss": 1.6406,
+      "step": 1944
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.6424199383300149e-06,
+      "loss": 1.6309,
+      "step": 1945
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.6378418506603456e-06,
+      "loss": 1.7227,
+      "step": 1946
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.633268902344259e-06,
+      "loss": 1.625,
+      "step": 1947
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.6287011003719105e-06,
+      "loss": 1.6289,
+      "step": 1948
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.6241384517255854e-06,
+      "loss": 1.8984,
+      "step": 1949
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.6195809633796889e-06,
+      "loss": 1.6465,
+      "step": 1950
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.6150286423007472e-06,
+      "loss": 1.6328,
+      "step": 1951
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.610481495447378e-06,
+      "loss": 1.7461,
+      "step": 1952
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.6059395297702967e-06,
+      "loss": 1.4473,
+      "step": 1953
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.601402752212296e-06,
+      "loss": 1.6797,
+      "step": 1954
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.596871169708235e-06,
+      "loss": 1.4941,
+      "step": 1955
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5923447891850396e-06,
+      "loss": 1.5625,
+      "step": 1956
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.5878236175616736e-06,
+      "loss": 1.4141,
+      "step": 1957
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.583307661749151e-06,
+      "loss": 1.8984,
+      "step": 1958
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5787969286505028e-06,
+      "loss": 1.3965,
+      "step": 1959
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5742914251607794e-06,
+      "loss": 1.5469,
+      "step": 1960
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.56979115816704e-06,
+      "loss": 1.4531,
+      "step": 1961
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.5652961345483353e-06,
+      "loss": 1.6445,
+      "step": 1962
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5608063611757058e-06,
+      "loss": 1.5996,
+      "step": 1963
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5563218449121603e-06,
+      "loss": 1.4668,
+      "step": 1964
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5518425926126774e-06,
+      "loss": 1.6641,
+      "step": 1965
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5473686111241881e-06,
+      "loss": 1.6602,
+      "step": 1966
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.5428999072855626e-06,
+      "loss": 1.3535,
+      "step": 1967
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5384364879276104e-06,
+      "loss": 1.4043,
+      "step": 1968
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5339783598730568e-06,
+      "loss": 1.5859,
+      "step": 1969
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5295255299365397e-06,
+      "loss": 1.6914,
+      "step": 1970
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5250780049246028e-06,
+      "loss": 1.3711,
+      "step": 1971
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.5206357916356796e-06,
+      "loss": 1.6309,
+      "step": 1972
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5161988968600787e-06,
+      "loss": 1.5664,
+      "step": 1973
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5117673273799866e-06,
+      "loss": 1.4785,
+      "step": 1974
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.507341089969443e-06,
+      "loss": 1.9219,
+      "step": 1975
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.5029201913943425e-06,
+      "loss": 1.7969,
+      "step": 1976
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.498504638412413e-06,
+      "loss": 1.6641,
+      "step": 1977
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4940944377732168e-06,
+      "loss": 1.9219,
+      "step": 1978
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4896895962181335e-06,
+      "loss": 1.5625,
+      "step": 1979
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.485290120480346e-06,
+      "loss": 1.5098,
+      "step": 1980
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.480896017284843e-06,
+      "loss": 1.4746,
+      "step": 1981
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4765072933483949e-06,
+      "loss": 1.6172,
+      "step": 1982
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4721239553795485e-06,
+      "loss": 1.5117,
+      "step": 1983
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4677460100786228e-06,
+      "loss": 1.5234,
+      "step": 1984
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4633734641376913e-06,
+      "loss": 1.5234,
+      "step": 1985
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4590063242405745e-06,
+      "loss": 1.8848,
+      "step": 1986
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4546445970628275e-06,
+      "loss": 1.377,
+      "step": 1987
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4502882892717319e-06,
+      "loss": 1.748,
+      "step": 1988
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.445937407526289e-06,
+      "loss": 1.6562,
+      "step": 1989
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4415919584771999e-06,
+      "loss": 1.248,
+      "step": 1990
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4372519487668657e-06,
+      "loss": 1.4629,
+      "step": 1991
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4329173850293738e-06,
+      "loss": 1.7383,
+      "step": 1992
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4285882738904822e-06,
+      "loss": 1.5898,
+      "step": 1993
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4242646219676193e-06,
+      "loss": 1.6641,
+      "step": 1994
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4199464358698638e-06,
+      "loss": 1.3359,
+      "step": 1995
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.415633722197945e-06,
+      "loss": 1.502,
+      "step": 1996
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4113264875442201e-06,
+      "loss": 1.4434,
+      "step": 1997
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4070247384926771e-06,
+      "loss": 1.7266,
+      "step": 1998
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.4027284816189179e-06,
+      "loss": 1.832,
+      "step": 1999
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.398437723490147e-06,
+      "loss": 1.4688,
+      "step": 2000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3941524706651628e-06,
+      "loss": 1.7207,
+      "step": 2001
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3898727296943526e-06,
+      "loss": 1.5312,
+      "step": 2002
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3855985071196725e-06,
+      "loss": 1.5078,
+      "step": 2003
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.3813298094746491e-06,
+      "loss": 1.3633,
+      "step": 2004
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3770666432843622e-06,
+      "loss": 1.5742,
+      "step": 2005
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3728090150654316e-06,
+      "loss": 2.0078,
+      "step": 2006
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3685569313260194e-06,
+      "loss": 1.5527,
+      "step": 2007
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3643103985658047e-06,
+      "loss": 1.6816,
+      "step": 2008
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.3600694232759886e-06,
+      "loss": 1.6641,
+      "step": 2009
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3558340119392704e-06,
+      "loss": 1.9434,
+      "step": 2010
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.35160417102985e-06,
+      "loss": 1.4688,
+      "step": 2011
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3473799070134107e-06,
+      "loss": 1.793,
+      "step": 2012
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3431612263471082e-06,
+      "loss": 1.502,
+      "step": 2013
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.3389481354795691e-06,
+      "loss": 1.5137,
+      "step": 2014
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3347406408508695e-06,
+      "loss": 1.9961,
+      "step": 2015
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3305387488925342e-06,
+      "loss": 1.8086,
+      "step": 2016
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3263424660275237e-06,
+      "loss": 1.541,
+      "step": 2017
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3221517986702249e-06,
+      "loss": 1.6445,
+      "step": 2018
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3179667532264417e-06,
+      "loss": 1.5117,
+      "step": 2019
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.3137873360933812e-06,
+      "loss": 1.627,
+      "step": 2020
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3096135536596488e-06,
+      "loss": 1.416,
+      "step": 2021
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.3054454123052385e-06,
+      "loss": 1.582,
+      "step": 2022
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.301282918401518e-06,
+      "loss": 1.7832,
+      "step": 2023
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.2971260783112249e-06,
+      "loss": 1.6602,
+      "step": 2024
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.292974898388456e-06,
+      "loss": 1.498,
+      "step": 2025
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2888293849786503e-06,
+      "loss": 1.4258,
+      "step": 2026
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.284689544418592e-06,
+      "loss": 1.7441,
+      "step": 2027
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2805553830363886e-06,
+      "loss": 1.4805,
+      "step": 2028
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.276426907151468e-06,
+      "loss": 1.4297,
+      "step": 2029
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.2723041230745676e-06,
+      "loss": 1.6895,
+      "step": 2030
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2681870371077249e-06,
+      "loss": 1.6016,
+      "step": 2031
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2640756555442684e-06,
+      "loss": 1.3945,
+      "step": 2032
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.259969984668804e-06,
+      "loss": 1.6074,
+      "step": 2033
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2558700307572085e-06,
+      "loss": 1.9453,
+      "step": 2034
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.2517758000766234e-06,
+      "loss": 1.4688,
+      "step": 2035
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2476872988854366e-06,
+      "loss": 1.5762,
+      "step": 2036
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2436045334332824e-06,
+      "loss": 1.3691,
+      "step": 2037
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2395275099610272e-06,
+      "loss": 1.7246,
+      "step": 2038
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.235456234700756e-06,
+      "loss": 1.5293,
+      "step": 2039
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.2313907138757724e-06,
+      "loss": 1.6328,
+      "step": 2040
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2273309537005801e-06,
+      "loss": 1.2578,
+      "step": 2041
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2232769603808797e-06,
+      "loss": 1.5781,
+      "step": 2042
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2192287401135538e-06,
+      "loss": 1.6758,
+      "step": 2043
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.215186299086663e-06,
+      "loss": 1.5312,
+      "step": 2044
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.2111496434794345e-06,
+      "loss": 1.4082,
+      "step": 2045
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.207118779462248e-06,
+      "loss": 2.0332,
+      "step": 2046
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.2030937131966346e-06,
+      "loss": 1.8477,
+      "step": 2047
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1990744508352604e-06,
+      "loss": 1.9395,
+      "step": 2048
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.195060998521918e-06,
+      "loss": 1.5527,
+      "step": 2049
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1910533623915244e-06,
+      "loss": 1.498,
+      "step": 2050
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.1870515485701013e-06,
+      "loss": 1.7168,
+      "step": 2051
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1830555631747736e-06,
+      "loss": 1.6504,
+      "step": 2052
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1790654123137552e-06,
+      "loss": 1.8828,
+      "step": 2053
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.175081102086339e-06,
+      "loss": 1.6992,
+      "step": 2054
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.171102638582896e-06,
+      "loss": 1.4238,
+      "step": 2055
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.1671300278848536e-06,
+      "loss": 1.3613,
+      "step": 2056
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1631632760646965e-06,
+      "loss": 1.668,
+      "step": 2057
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1592023891859554e-06,
+      "loss": 1.666,
+      "step": 2058
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1552473733031893e-06,
+      "loss": 1.6152,
+      "step": 2059
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1512982344619904e-06,
+      "loss": 1.3066,
+      "step": 2060
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.1473549786989618e-06,
+      "loss": 1.3535,
+      "step": 2061
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1434176120417152e-06,
+      "loss": 1.418,
+      "step": 2062
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.139486140508861e-06,
+      "loss": 1.6289,
+      "step": 2063
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1355605701099985e-06,
+      "loss": 1.4492,
+      "step": 2064
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1316409068457084e-06,
+      "loss": 1.293,
+      "step": 2065
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1277271567075377e-06,
+      "loss": 1.3477,
+      "step": 2066
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.1238193256779955e-06,
+      "loss": 1.6191,
+      "step": 2067
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1199174197305473e-06,
+      "loss": 1.3516,
+      "step": 2068
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1160214448295958e-06,
+      "loss": 1.7051,
+      "step": 2069
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.112131406930481e-06,
+      "loss": 1.4473,
+      "step": 2070
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1082473119794695e-06,
+      "loss": 1.5352,
+      "step": 2071
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.1043691659137372e-06,
+      "loss": 1.6094,
+      "step": 2072
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.1004969746613736e-06,
+      "loss": 1.4121,
+      "step": 2073
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0966307441413598e-06,
+      "loss": 1.459,
+      "step": 2074
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0927704802635708e-06,
+      "loss": 1.6504,
+      "step": 2075
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0889161889287563e-06,
+      "loss": 1.3379,
+      "step": 2076
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.0850678760285399e-06,
+      "loss": 1.4297,
+      "step": 2077
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0812255474454059e-06,
+      "loss": 1.3633,
+      "step": 2078
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0773892090526888e-06,
+      "loss": 1.6582,
+      "step": 2079
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0735588667145702e-06,
+      "loss": 1.5371,
+      "step": 2080
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0697345262860638e-06,
+      "loss": 1.4512,
+      "step": 2081
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.0659161936130069e-06,
+      "loss": 1.5117,
+      "step": 2082
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0621038745320579e-06,
+      "loss": 1.6934,
+      "step": 2083
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0582975748706819e-06,
+      "loss": 1.4453,
+      "step": 2084
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0544973004471398e-06,
+      "loss": 1.332,
+      "step": 2085
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0507030570704858e-06,
+      "loss": 1.6211,
+      "step": 2086
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0469148505405518e-06,
+      "loss": 1.7598,
+      "step": 2087
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.0431326866479457e-06,
+      "loss": 1.584,
+      "step": 2088
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0393565711740361e-06,
+      "loss": 1.4805,
+      "step": 2089
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0355865098909452e-06,
+      "loss": 1.3594,
+      "step": 2090
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0318225085615436e-06,
+      "loss": 1.375,
+      "step": 2091
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0280645729394368e-06,
+      "loss": 1.541,
+      "step": 2092
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.0243127087689609e-06,
+      "loss": 1.4551,
+      "step": 2093
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0205669217851678e-06,
+      "loss": 1.5352,
+      "step": 2094
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.01682721771382e-06,
+      "loss": 1.5684,
+      "step": 2095
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0130936022713861e-06,
+      "loss": 1.5312,
+      "step": 2096
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0093660811650218e-06,
+      "loss": 1.416,
+      "step": 2097
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.0056446600925718e-06,
+      "loss": 1.7148,
+      "step": 2098
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.0019293447425555e-06,
+      "loss": 1.4043,
+      "step": 2099
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.98220140794155e-07,
+      "loss": 1.4766,
+      "step": 2100
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.945170539172171e-07,
+      "loss": 1.5078,
+      "step": 2101
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.908200897722332e-07,
+      "loss": 1.3496,
+      "step": 2102
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.871292540103377e-07,
+      "loss": 1.3125,
+      "step": 2103
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.834445522732954e-07,
+      "loss": 1.7285,
+      "step": 2104
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.797659901934981e-07,
+      "loss": 1.6152,
+      "step": 2105
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.760935733939497e-07,
+      "loss": 1.7266,
+      "step": 2106
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.724273074882606e-07,
+      "loss": 1.6816,
+      "step": 2107
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.687671980806417e-07,
+      "loss": 1.4824,
+      "step": 2108
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.6511325076589e-07,
+      "loss": 1.834,
+      "step": 2109
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.61465471129383e-07,
+      "loss": 1.4551,
+      "step": 2110
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.578238647470732e-07,
+      "loss": 1.4688,
+      "step": 2111
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.541884371854765e-07,
+      "loss": 1.4883,
+      "step": 2112
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.505591940016601e-07,
+      "loss": 1.4355,
+      "step": 2113
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.469361407432431e-07,
+      "loss": 1.25,
+      "step": 2114
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.433192829483778e-07,
+      "loss": 1.4707,
+      "step": 2115
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.397086261457511e-07,
+      "loss": 1.4668,
+      "step": 2116
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.361041758545659e-07,
+      "loss": 1.4199,
+      "step": 2117
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.325059375845414e-07,
+      "loss": 1.5059,
+      "step": 2118
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.289139168359019e-07,
+      "loss": 2.0938,
+      "step": 2119
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.25328119099364e-07,
+      "loss": 1.3262,
+      "step": 2120
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.217485498561363e-07,
+      "loss": 1.2949,
+      "step": 2121
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.181752145779027e-07,
+      "loss": 1.6641,
+      "step": 2122
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.146081187268185e-07,
+      "loss": 1.5156,
+      "step": 2123
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.110472677555043e-07,
+      "loss": 1.6133,
+      "step": 2124
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.074926671070322e-07,
+      "loss": 1.541,
+      "step": 2125
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.039443222149231e-07,
+      "loss": 1.5684,
+      "step": 2126
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.004022385031314e-07,
+      "loss": 1.582,
+      "step": 2127
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.968664213860417e-07,
+      "loss": 1.4766,
+      "step": 2128
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.933368762684635e-07,
+      "loss": 1.5078,
+      "step": 2129
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 8.898136085456127e-07,
+      "loss": 1.4844,
+      "step": 2130
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.862966236031151e-07,
+      "loss": 1.4785,
+      "step": 2131
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.827859268169908e-07,
+      "loss": 1.3926,
+      "step": 2132
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.792815235536467e-07,
+      "loss": 1.2539,
+      "step": 2133
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.757834191698716e-07,
+      "loss": 1.2227,
+      "step": 2134
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 8.722916190128233e-07,
+      "loss": 1.3965,
+      "step": 2135
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.688061284200266e-07,
+      "loss": 1.5605,
+      "step": 2136
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.65326952719357e-07,
+      "loss": 1.4805,
+      "step": 2137
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.618540972290407e-07,
+      "loss": 1.291,
+      "step": 2138
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.583875672576441e-07,
+      "loss": 1.5078,
+      "step": 2139
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 8.549273681040598e-07,
+      "loss": 1.5059,
+      "step": 2140
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.514735050575058e-07,
+      "loss": 1.3555,
+      "step": 2141
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.480259833975152e-07,
+      "loss": 1.4258,
+      "step": 2142
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.445848083939267e-07,
+      "loss": 1.3594,
+      "step": 2143
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.411499853068783e-07,
+      "loss": 1.4492,
+      "step": 2144
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 8.377215193867999e-07,
+      "loss": 1.3926,
+      "step": 2145
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.342994158743994e-07,
+      "loss": 1.5508,
+      "step": 2146
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.308836800006648e-07,
+      "loss": 1.334,
+      "step": 2147
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.27474316986846e-07,
+      "loss": 1.5664,
+      "step": 2148
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.240713320444549e-07,
+      "loss": 1.6484,
+      "step": 2149
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.206747303752505e-07,
+      "loss": 1.6191,
+      "step": 2150
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 8.172845171712379e-07,
+      "loss": 1.3066,
+      "step": 2151
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.139006976146557e-07,
+      "loss": 1.2959,
+      "step": 2152
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.10523276877967e-07,
+      "loss": 1.6113,
+      "step": 2153
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.071522601238574e-07,
+      "loss": 1.293,
+      "step": 2154
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.037876525052213e-07,
+      "loss": 1.1045,
+      "step": 2155
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 8.004294591651546e-07,
+      "loss": 1.5586,
+      "step": 2156
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.970776852369522e-07,
+      "loss": 1.7246,
+      "step": 2157
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 1.5742,
+      "step": 2158
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.903934161002408e-07,
+      "loss": 1.4316,
+      "step": 2159
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.870609311092236e-07,
+      "loss": 1.498,
+      "step": 2160
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 7.837348859650367e-07,
+      "loss": 1.4062,
+      "step": 2161
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.804152857518338e-07,
+      "loss": 1.6875,
+      "step": 2162
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.771021355439135e-07,
+      "loss": 1.3906,
+      "step": 2163
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.737954404057163e-07,
+      "loss": 1.3359,
+      "step": 2164
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.70495205391818e-07,
+      "loss": 1.3066,
+      "step": 2165
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 7.672014355469143e-07,
+      "loss": 1.7871,
+      "step": 2166
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.639141359058238e-07,
+      "loss": 1.4434,
+      "step": 2167
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.606333114934711e-07,
+      "loss": 1.4902,
+      "step": 2168
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.573589673248833e-07,
+      "loss": 1.3594,
+      "step": 2169
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.540911084051827e-07,
+      "loss": 1.5352,
+      "step": 2170
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 7.508297397295794e-07,
+      "loss": 1.6523,
+      "step": 2171
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.475748662833615e-07,
+      "loss": 1.5195,
+      "step": 2172
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.443264930418886e-07,
+      "loss": 1.5801,
+      "step": 2173
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.410846249705816e-07,
+      "loss": 1.5195,
+      "step": 2174
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.378492670249238e-07,
+      "loss": 1.4766,
+      "step": 2175
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.346204241504401e-07,
+      "loss": 1.3555,
+      "step": 2176
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 7.313981012827015e-07,
+      "loss": 1.457,
+      "step": 2177
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.281823033473129e-07,
+      "loss": 1.4648,
+      "step": 2178
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.249730352599e-07,
+      "loss": 1.6641,
+      "step": 2179
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.217703019261135e-07,
+      "loss": 1.377,
+      "step": 2180
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.185741082416103e-07,
+      "loss": 1.293,
+      "step": 2181
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 7.153844590920539e-07,
+      "loss": 1.5918,
+      "step": 2182
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.122013593531019e-07,
+      "loss": 1.5625,
+      "step": 2183
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.090248138904011e-07,
+      "loss": 1.4902,
+      "step": 2184
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.058548275595816e-07,
+      "loss": 1.4258,
+      "step": 2185
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 7.026914052062433e-07,
+      "loss": 1.3711,
+      "step": 2186
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 6.995345516659574e-07,
+      "loss": 1.5,
+      "step": 2187
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.963842717642488e-07,
+      "loss": 1.4082,
+      "step": 2188
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.932405703165973e-07,
+      "loss": 1.4609,
+      "step": 2189
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.901034521284261e-07,
+      "loss": 1.4941,
+      "step": 2190
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.86972921995096e-07,
+      "loss": 1.4395,
+      "step": 2191
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 6.838489847018975e-07,
+      "loss": 1.2129,
+      "step": 2192
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.807316450240425e-07,
+      "loss": 1.5176,
+      "step": 2193
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.77620907726656e-07,
+      "loss": 1.666,
+      "step": 2194
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.745167775647754e-07,
+      "loss": 1.416,
+      "step": 2195
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.714192592833335e-07,
+      "loss": 1.4648,
+      "step": 2196
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.6832835761716e-07,
+      "loss": 1.5605,
+      "step": 2197
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 6.652440772909708e-07,
+      "loss": 1.3223,
+      "step": 2198
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.621664230193553e-07,
+      "loss": 1.5508,
+      "step": 2199
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.590953995067812e-07,
+      "loss": 1.7617,
+      "step": 2200
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.560310114475754e-07,
+      "loss": 1.4004,
+      "step": 2201
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.529732635259234e-07,
+      "loss": 1.7891,
+      "step": 2202
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 6.499221604158623e-07,
+      "loss": 1.666,
+      "step": 2203
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.468777067812699e-07,
+      "loss": 1.4902,
+      "step": 2204
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.438399072758622e-07,
+      "loss": 1.4199,
+      "step": 2205
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.408087665431806e-07,
+      "loss": 1.3164,
+      "step": 2206
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.377842892165892e-07,
+      "loss": 1.2852,
+      "step": 2207
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 6.347664799192688e-07,
+      "loss": 1.3262,
+      "step": 2208
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.317553432642032e-07,
+      "loss": 1.3848,
+      "step": 2209
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.287508838541795e-07,
+      "loss": 1.2715,
+      "step": 2210
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.257531062817795e-07,
+      "loss": 1.3184,
+      "step": 2211
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.227620151293662e-07,
+      "loss": 1.4219,
+      "step": 2212
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 6.197776149690871e-07,
+      "loss": 1.4023,
+      "step": 2213
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.167999103628569e-07,
+      "loss": 1.7285,
+      "step": 2214
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.138289058623615e-07,
+      "loss": 1.2734,
+      "step": 2215
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.1086460600904e-07,
+      "loss": 1.5898,
+      "step": 2216
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.079070153340827e-07,
+      "loss": 1.5391,
+      "step": 2217
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.049561383584301e-07,
+      "loss": 1.3652,
+      "step": 2218
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 6.020119795927542e-07,
+      "loss": 1.457,
+      "step": 2219
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.990745435374612e-07,
+      "loss": 1.4375,
+      "step": 2220
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.961438346826792e-07,
+      "loss": 1.3164,
+      "step": 2221
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.932198575082526e-07,
+      "loss": 1.6406,
+      "step": 2222
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.903026164837389e-07,
+      "loss": 1.3711,
+      "step": 2223
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 5.873921160683943e-07,
+      "loss": 1.4453,
+      "step": 2224
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.844883607111751e-07,
+      "loss": 1.2734,
+      "step": 2225
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.81591354850728e-07,
+      "loss": 1.4805,
+      "step": 2226
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.787011029153777e-07,
+      "loss": 1.6426,
+      "step": 2227
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.758176093231294e-07,
+      "loss": 1.334,
+      "step": 2228
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 5.729408784816559e-07,
+      "loss": 1.5508,
+      "step": 2229
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.70070914788291e-07,
+      "loss": 1.9062,
+      "step": 2230
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.672077226300265e-07,
+      "loss": 1.3945,
+      "step": 2231
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.643513063835038e-07,
+      "loss": 1.4062,
+      "step": 2232
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.615016704150056e-07,
+      "loss": 1.1973,
+      "step": 2233
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 5.586588190804499e-07,
+      "loss": 1.6367,
+      "step": 2234
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.558227567253832e-07,
+      "loss": 1.2266,
+      "step": 2235
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.529934876849774e-07,
+      "loss": 1.3203,
+      "step": 2236
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.501710162840162e-07,
+      "loss": 1.3691,
+      "step": 2237
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.473553468368963e-07,
+      "loss": 1.4668,
+      "step": 2238
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.445464836476155e-07,
+      "loss": 1.4121,
+      "step": 2239
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 5.417444310097664e-07,
+      "loss": 1.4316,
+      "step": 2240
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.389491932065338e-07,
+      "loss": 1.3223,
+      "step": 2241
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.361607745106817e-07,
+      "loss": 1.2852,
+      "step": 2242
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.333791791845549e-07,
+      "loss": 1.4023,
+      "step": 2243
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.306044114800635e-07,
+      "loss": 1.6973,
+      "step": 2244
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 5.278364756386839e-07,
+      "loss": 1.4512,
+      "step": 2245
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.250753758914506e-07,
+      "loss": 1.4551,
+      "step": 2246
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.223211164589448e-07,
+      "loss": 1.5352,
+      "step": 2247
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.195737015512947e-07,
+      "loss": 1.4629,
+      "step": 2248
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.168331353681643e-07,
+      "loss": 1.4707,
+      "step": 2249
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 5.140994220987489e-07,
+      "loss": 1.3828,
+      "step": 2250
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.113725659217689e-07,
+      "loss": 1.3613,
+      "step": 2251
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.086525710054652e-07,
+      "loss": 1.4766,
+      "step": 2252
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.059394415075858e-07,
+      "loss": 1.4375,
+      "step": 2253
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.032331815753894e-07,
+      "loss": 1.4844,
+      "step": 2254
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 5.005337953456297e-07,
+      "loss": 1.3223,
+      "step": 2255
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.97841286944557e-07,
+      "loss": 1.248,
+      "step": 2256
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.951556604879049e-07,
+      "loss": 1.332,
+      "step": 2257
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.924769200808893e-07,
+      "loss": 1.4883,
+      "step": 2258
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.898050698182017e-07,
+      "loss": 1.3105,
+      "step": 2259
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.871401137839971e-07,
+      "loss": 1.418,
+      "step": 2260
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 4.844820560518959e-07,
+      "loss": 1.3809,
+      "step": 2261
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.81830900684973e-07,
+      "loss": 1.2773,
+      "step": 2262
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.791866517357491e-07,
+      "loss": 1.3848,
+      "step": 2263
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.765493132461929e-07,
+      "loss": 1.2148,
+      "step": 2264
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.739188892477076e-07,
+      "loss": 1.4434,
+      "step": 2265
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 4.7129538376112673e-07,
+      "loss": 1.1318,
+      "step": 2266
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.686788007967086e-07,
+      "loss": 1.4043,
+      "step": 2267
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.660691443541282e-07,
+      "loss": 1.5156,
+      "step": 2268
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.634664184224769e-07,
+      "loss": 1.4883,
+      "step": 2269
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.608706269802471e-07,
+      "loss": 1.5293,
+      "step": 2270
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 4.5828177399533436e-07,
+      "loss": 1.332,
+      "step": 2271
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.556998634250287e-07,
+      "loss": 1.5078,
+      "step": 2272
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.531248992160053e-07,
+      "loss": 1.3691,
+      "step": 2273
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.5055688530432407e-07,
+      "loss": 1.3203,
+      "step": 2274
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.4799582561541934e-07,
+      "loss": 1.375,
+      "step": 2275
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 4.4544172406409345e-07,
+      "loss": 1.8125,
+      "step": 2276
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.428945845545168e-07,
+      "loss": 1.5,
+      "step": 2277
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.403544109802144e-07,
+      "loss": 1.5059,
+      "step": 2278
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.3782120722406565e-07,
+      "loss": 1.2148,
+      "step": 2279
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.352949771582937e-07,
+      "loss": 1.9043,
+      "step": 2280
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.3277572464446247e-07,
+      "loss": 1.8984,
+      "step": 2281
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 4.30263453533471e-07,
+      "loss": 1.4551,
+      "step": 2282
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.277581676655451e-07,
+      "loss": 1.5918,
+      "step": 2283
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2525987087023433e-07,
+      "loss": 1.5234,
+      "step": 2284
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.2276856696640557e-07,
+      "loss": 1.5234,
+      "step": 2285
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.202842597622336e-07,
+      "loss": 1.5742,
+      "step": 2286
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 4.1780695305520147e-07,
+      "loss": 1.4355,
+      "step": 2287
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.1533665063208773e-07,
+      "loss": 1.4492,
+      "step": 2288
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.128733562689685e-07,
+      "loss": 1.2051,
+      "step": 2289
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.1041707373120354e-07,
+      "loss": 1.2969,
+      "step": 2290
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.0796780677343606e-07,
+      "loss": 1.4023,
+      "step": 2291
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 4.055255591395879e-07,
+      "loss": 1.1973,
+      "step": 2292
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.0309033456284565e-07,
+      "loss": 1.2051,
+      "step": 2293
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 4.0066213676566544e-07,
+      "loss": 1.4609,
+      "step": 2294
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.9824096945976056e-07,
+      "loss": 1.4131,
+      "step": 2295
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.95826836346096e-07,
+      "loss": 1.2861,
+      "step": 2296
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 3.934197411148871e-07,
+      "loss": 1.7988,
+      "step": 2297
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.910196874455896e-07,
+      "loss": 1.5254,
+      "step": 2298
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.886266790068971e-07,
+      "loss": 1.3379,
+      "step": 2299
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.8624071945673246e-07,
+      "loss": 1.2754,
+      "step": 2300
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.8386181244224274e-07,
+      "loss": 1.2773,
+      "step": 2301
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 3.8148996159979803e-07,
+      "loss": 1.4629,
+      "step": 2302
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.7912517055497756e-07,
+      "loss": 1.3809,
+      "step": 2303
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.767674429225743e-07,
+      "loss": 1.5195,
+      "step": 2304
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.744167823065814e-07,
+      "loss": 1.4121,
+      "step": 2305
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.720731923001886e-07,
+      "loss": 1.459,
+      "step": 2306
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.6973667648578016e-07,
+      "loss": 1.4316,
+      "step": 2307
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 3.674072384349242e-07,
+      "loss": 1.4727,
+      "step": 2308
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.6508488170837133e-07,
+      "loss": 1.6035,
+      "step": 2309
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.6276960985604736e-07,
+      "loss": 1.8496,
+      "step": 2310
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.604614264170486e-07,
+      "loss": 1.377,
+      "step": 2311
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.581603349196372e-07,
+      "loss": 1.5664,
+      "step": 2312
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 3.5586633888123166e-07,
+      "loss": 1.667,
+      "step": 2313
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.5357944180840544e-07,
+      "loss": 1.5469,
+      "step": 2314
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.5129964719688383e-07,
+      "loss": 1.5586,
+      "step": 2315
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.4902695853152977e-07,
+      "loss": 1.2988,
+      "step": 2316
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.467613792863489e-07,
+      "loss": 1.3906,
+      "step": 2317
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 3.445029129244787e-07,
+      "loss": 1.3574,
+      "step": 2318
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.4225156289818096e-07,
+      "loss": 1.7383,
+      "step": 2319
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.4000733264884343e-07,
+      "loss": 1.3418,
+      "step": 2320
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.377702256069676e-07,
+      "loss": 1.3164,
+      "step": 2321
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.355402451921685e-07,
+      "loss": 1.1641,
+      "step": 2322
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 3.333173948131663e-07,
+      "loss": 1.2441,
+      "step": 2323
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.311016778677828e-07,
+      "loss": 1.3047,
+      "step": 2324
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.2889309774293666e-07,
+      "loss": 1.416,
+      "step": 2325
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.26691657814634e-07,
+      "loss": 1.334,
+      "step": 2326
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.2449736144797183e-07,
+      "loss": 1.3311,
+      "step": 2327
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.2231021199712274e-07,
+      "loss": 1.4375,
+      "step": 2328
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 3.2013021280533576e-07,
+      "loss": 1.4395,
+      "step": 2329
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.1795736720493184e-07,
+      "loss": 1.3242,
+      "step": 2330
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.157916785172971e-07,
+      "loss": 1.4062,
+      "step": 2331
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.13633150052875e-07,
+      "loss": 1.5391,
+      "step": 2332
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.1148178511116624e-07,
+      "loss": 1.416,
+      "step": 2333
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 3.0933758698072023e-07,
+      "loss": 1.1885,
+      "step": 2334
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.07200558939133e-07,
+      "loss": 1.3086,
+      "step": 2335
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.050707042530382e-07,
+      "loss": 1.4551,
+      "step": 2336
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.029480261781054e-07,
+      "loss": 1.4219,
+      "step": 2337
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 3.008325279590357e-07,
+      "loss": 1.4395,
+      "step": 2338
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.98724212829552e-07,
+      "loss": 1.4531,
+      "step": 2339
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.966230840124007e-07,
+      "loss": 1.2383,
+      "step": 2340
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.945291447193399e-07,
+      "loss": 1.3438,
+      "step": 2341
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.9244239815113874e-07,
+      "loss": 1.4219,
+      "step": 2342
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.903628474975739e-07,
+      "loss": 1.4004,
+      "step": 2343
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.882904959374205e-07,
+      "loss": 1.2207,
+      "step": 2344
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.862253466384507e-07,
+      "loss": 1.4297,
+      "step": 2345
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.8416740275742484e-07,
+      "loss": 1.332,
+      "step": 2346
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.821166674400905e-07,
+      "loss": 1.543,
+      "step": 2347
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.8007314382117733e-07,
+      "loss": 1.4336,
+      "step": 2348
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.780368350243895e-07,
+      "loss": 1.3828,
+      "step": 2349
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.760077441624043e-07,
+      "loss": 1.5117,
+      "step": 2350
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7398587433686563e-07,
+      "loss": 1.3184,
+      "step": 2351
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.719712286383769e-07,
+      "loss": 1.4668,
+      "step": 2352
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.6996381014650353e-07,
+      "loss": 1.1855,
+      "step": 2353
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.6796362192975766e-07,
+      "loss": 1.2148,
+      "step": 2354
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.659706670456047e-07,
+      "loss": 1.6191,
+      "step": 2355
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.6398494854045055e-07,
+      "loss": 1.8555,
+      "step": 2356
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.620064694496377e-07,
+      "loss": 1.2695,
+      "step": 2357
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.600352327974487e-07,
+      "loss": 1.5117,
+      "step": 2358
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.5807124159708987e-07,
+      "loss": 1.6797,
+      "step": 2359
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.561144988506931e-07,
+      "loss": 1.4062,
+      "step": 2360
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.5416500754931294e-07,
+      "loss": 1.1455,
+      "step": 2361
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.5222277067291633e-07,
+      "loss": 1.3789,
+      "step": 2362
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.502877911903839e-07,
+      "loss": 1.3926,
+      "step": 2363
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.4836007205950087e-07,
+      "loss": 1.2285,
+      "step": 2364
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.4643961622695444e-07,
+      "loss": 1.4922,
+      "step": 2365
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.4452642662833146e-07,
+      "loss": 1.752,
+      "step": 2366
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.426205061881082e-07,
+      "loss": 1.2275,
+      "step": 2367
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.407218578196524e-07,
+      "loss": 1.4395,
+      "step": 2368
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.388304844252143e-07,
+      "loss": 1.4531,
+      "step": 2369
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.3694638889592336e-07,
+      "loss": 1.2617,
+      "step": 2370
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.3506957411178478e-07,
+      "loss": 1.2852,
+      "step": 2371
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.3320004294167587e-07,
+      "loss": 1.6641,
+      "step": 2372
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.3133779824333847e-07,
+      "loss": 1.4766,
+      "step": 2373
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.2948284286337663e-07,
+      "loss": 1.5977,
+      "step": 2374
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.2763517963725169e-07,
+      "loss": 1.4492,
+      "step": 2375
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.257948113892805e-07,
+      "loss": 1.5469,
+      "step": 2376
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.23961740932625e-07,
+      "loss": 1.623,
+      "step": 2377
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.2213597106929608e-07,
+      "loss": 1.2402,
+      "step": 2378
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.2031750459014345e-07,
+      "loss": 1.4668,
+      "step": 2379
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.1850634427485085e-07,
+      "loss": 1.2695,
+      "step": 2380
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.16702492891937e-07,
+      "loss": 1.1738,
+      "step": 2381
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.1490595319874574e-07,
+      "loss": 1.4961,
+      "step": 2382
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.13116727941447e-07,
+      "loss": 1.3281,
+      "step": 2383
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.1133481985502747e-07,
+      "loss": 1.4277,
+      "step": 2384
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.0956023166328942e-07,
+      "loss": 1.2246,
+      "step": 2385
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.0779296607884857e-07,
+      "loss": 1.4941,
+      "step": 2386
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0603302580312402e-07,
+      "loss": 1.2695,
+      "step": 2387
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0428041352633766e-07,
+      "loss": 1.1914,
+      "step": 2388
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": 1.5684,
+      "step": 2389
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.0079718367446544e-07,
+      "loss": 1.2246,
+      "step": 2390
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.9906657142380105e-07,
+      "loss": 1.3203,
+      "step": 2391
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.9734329782091277e-07,
+      "loss": 1.373,
+      "step": 2392
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.9562736549997463e-07,
+      "loss": 1.334,
+      "step": 2393
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.9391877708394125e-07,
+      "loss": 1.707,
+      "step": 2394
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.922175351845379e-07,
+      "loss": 1.252,
+      "step": 2395
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.905236424022633e-07,
+      "loss": 1.4941,
+      "step": 2396
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.8883710132638068e-07,
+      "loss": 1.3164,
+      "step": 2397
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8715791453491562e-07,
+      "loss": 1.3047,
+      "step": 2398
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8548608459465377e-07,
+      "loss": 1.4199,
+      "step": 2399
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8382161406113208e-07,
+      "loss": 1.3398,
+      "step": 2400
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.821645054786414e-07,
+      "loss": 1.6016,
+      "step": 2401
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.8051476138021663e-07,
+      "loss": 1.459,
+      "step": 2402
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7887238428763553e-07,
+      "loss": 1.165,
+      "step": 2403
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7723737671141662e-07,
+      "loss": 1.6172,
+      "step": 2404
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7560974115081176e-07,
+      "loss": 1.2539,
+      "step": 2405
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7398948009380578e-07,
+      "loss": 1.3799,
+      "step": 2406
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.7237659601710866e-07,
+      "loss": 1.7031,
+      "step": 2407
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.7077109138615544e-07,
+      "loss": 1.4414,
+      "step": 2408
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.691729686551008e-07,
+      "loss": 1.5098,
+      "step": 2409
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6758223026681507e-07,
+      "loss": 1.3652,
+      "step": 2410
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.659988786528821e-07,
+      "loss": 1.3154,
+      "step": 2411
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6442291623359364e-07,
+      "loss": 1.6406,
+      "step": 2412
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6285434541794598e-07,
+      "loss": 1.3184,
+      "step": 2413
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.612931686036373e-07,
+      "loss": 1.6172,
+      "step": 2414
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.597393881770637e-07,
+      "loss": 1.2881,
+      "step": 2415
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5819300651331416e-07,
+      "loss": 1.1953,
+      "step": 2416
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5665402597616842e-07,
+      "loss": 1.2227,
+      "step": 2417
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5512244891809413e-07,
+      "loss": 1.6348,
+      "step": 2418
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5359827768024138e-07,
+      "loss": 1.4248,
+      "step": 2419
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.520815145924387e-07,
+      "loss": 1.2207,
+      "step": 2420
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5057216197319147e-07,
+      "loss": 1.2256,
+      "step": 2421
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.4907022212967803e-07,
+      "loss": 1.3184,
+      "step": 2422
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.475756973577447e-07,
+      "loss": 1.3184,
+      "step": 2423
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4608858994190344e-07,
+      "loss": 1.5312,
+      "step": 2424
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4460890215532986e-07,
+      "loss": 1.2695,
+      "step": 2425
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4313663625985464e-07,
+      "loss": 1.4297,
+      "step": 2426
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4167179450596647e-07,
+      "loss": 1.3477,
+      "step": 2427
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4021437913280366e-07,
+      "loss": 1.2695,
+      "step": 2428
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.3876439236815475e-07,
+      "loss": 1.5332,
+      "step": 2429
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.373218364284501e-07,
+      "loss": 1.2441,
+      "step": 2430
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.358867135187636e-07,
+      "loss": 1.3125,
+      "step": 2431
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.3445902583280656e-07,
+      "loss": 1.5605,
+      "step": 2432
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.3303877555292443e-07,
+      "loss": 1.1465,
+      "step": 2433
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.316259648500956e-07,
+      "loss": 1.4746,
+      "step": 2434
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3022059588392367e-07,
+      "loss": 1.5371,
+      "step": 2435
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.288226708026391e-07,
+      "loss": 1.3145,
+      "step": 2436
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2743219174309208e-07,
+      "loss": 1.4375,
+      "step": 2437
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2604916083075236e-07,
+      "loss": 1.4355,
+      "step": 2438
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.2467358017970498e-07,
+      "loss": 1.1211,
+      "step": 2439
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2330545189264464e-07,
+      "loss": 1.3125,
+      "step": 2440
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2194477806087458e-07,
+      "loss": 1.3047,
+      "step": 2441
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.2059156076430545e-07,
+      "loss": 1.4434,
+      "step": 2442
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.192458020714482e-07,
+      "loss": 1.5156,
+      "step": 2443
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1790750403941231e-07,
+      "loss": 1.4062,
+      "step": 2444
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1657666871390471e-07,
+      "loss": 1.4531,
+      "step": 2445
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1525329812922315e-07,
+      "loss": 1.1855,
+      "step": 2446
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1393739430825667e-07,
+      "loss": 1.3496,
+      "step": 2447
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1262895926247852e-07,
+      "loss": 1.668,
+      "step": 2448
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1132799499194658e-07,
+      "loss": 1.5342,
+      "step": 2449
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.1003450348529898e-07,
+      "loss": 1.3926,
+      "step": 2450
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.087484867197508e-07,
+      "loss": 1.2637,
+      "step": 2451
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0746994666109234e-07,
+      "loss": 1.4277,
+      "step": 2452
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0619888526368361e-07,
+      "loss": 1.498,
+      "step": 2453
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.0493530447045263e-07,
+      "loss": 1.3945,
+      "step": 2454
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0367920621289496e-07,
+      "loss": 1.4727,
+      "step": 2455
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0243059241106524e-07,
+      "loss": 1.2832,
+      "step": 2456
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.0118946497358006e-07,
+      "loss": 1.4414,
+      "step": 2457
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.995582579761243e-08,
+      "loss": 1.7422,
+      "step": 2458
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.872967676888611e-08,
+      "loss": 1.416,
+      "step": 2459
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.751101976167964e-08,
+      "loss": 1.3047,
+      "step": 2460
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.629985663881514e-08,
+      "loss": 1.3633,
+      "step": 2461
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.509618925166276e-08,
+      "loss": 1.4043,
+      "step": 2462
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.390001944013293e-08,
+      "loss": 1.3633,
+      "step": 2463
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.271134903267642e-08,
+      "loss": 1.502,
+      "step": 2464
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.153017984628143e-08,
+      "loss": 1.2617,
+      "step": 2465
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.035651368646647e-08,
+      "loss": 1.1904,
+      "step": 2466
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.919035234728479e-08,
+      "loss": 1.25,
+      "step": 2467
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.803169761131547e-08,
+      "loss": 1.5996,
+      "step": 2468
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.688055124966454e-08,
+      "loss": 1.459,
+      "step": 2469
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 8.573691502195891e-08,
+      "loss": 1.541,
+      "step": 2470
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.460079067634853e-08,
+      "loss": 1.5117,
+      "step": 2471
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.347217994949808e-08,
+      "loss": 1.4043,
+      "step": 2472
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.235108456658814e-08,
+      "loss": 1.2539,
+      "step": 2473
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.123750624131121e-08,
+      "loss": 1.3887,
+      "step": 2474
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 8.013144667587014e-08,
+      "loss": 1.2695,
+      "step": 2475
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.903290756097304e-08,
+      "loss": 1.3535,
+      "step": 2476
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.794189057583335e-08,
+      "loss": 1.4863,
+      "step": 2477
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.685839738816703e-08,
+      "loss": 1.2324,
+      "step": 2478
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.578242965418814e-08,
+      "loss": 1.4785,
+      "step": 2479
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.471398901860772e-08,
+      "loss": 1.6191,
+      "step": 2480
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 7.365307711463043e-08,
+      "loss": 1.2832,
+      "step": 2481
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.259969556395352e-08,
+      "loss": 1.377,
+      "step": 2482
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.155384597676285e-08,
+      "loss": 1.2227,
+      "step": 2483
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 7.051552995173128e-08,
+      "loss": 1.6211,
+      "step": 2484
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.948474907601699e-08,
+      "loss": 1.3828,
+      "step": 2485
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 6.846150492525739e-08,
+      "loss": 1.2539,
+      "step": 2486
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.744579906357185e-08,
+      "loss": 1.3379,
+      "step": 2487
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.643763304355566e-08,
+      "loss": 1.6895,
+      "step": 2488
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.543700840627832e-08,
+      "loss": 1.6621,
+      "step": 2489
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.444392668128352e-08,
+      "loss": 1.3867,
+      "step": 2490
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 6.34583893865831e-08,
+      "loss": 1.3652,
+      "step": 2491
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.248039802865702e-08,
+      "loss": 1.3008,
+      "step": 2492
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.150995410245164e-08,
+      "loss": 1.623,
+      "step": 2493
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 6.054705909137426e-08,
+      "loss": 1.416,
+      "step": 2494
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.959171446729584e-08,
+      "loss": 1.6191,
+      "step": 2495
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 5.864392169054378e-08,
+      "loss": 1.5488,
+      "step": 2496
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.770368220990252e-08,
+      "loss": 1.3887,
+      "step": 2497
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.677099746261017e-08,
+      "loss": 1.2598,
+      "step": 2498
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.584586887435739e-08,
+      "loss": 1.502,
+      "step": 2499
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.492829785928466e-08,
+      "loss": 1.1445,
+      "step": 2500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.401828581997948e-08,
+      "loss": 1.4492,
+      "step": 2501
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 5.311583414747523e-08,
+      "loss": 1.1758,
+      "step": 2502
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.222094422124846e-08,
+      "loss": 1.5742,
+      "step": 2503
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.133361740921605e-08,
+      "loss": 1.25,
+      "step": 2504
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 5.0453855067735815e-08,
+      "loss": 1.2734,
+      "step": 2505
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.9581658541602015e-08,
+      "loss": 1.2402,
+      "step": 2506
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 4.871702916404208e-08,
+      "loss": 1.6484,
+      "step": 2507
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.7859968256719344e-08,
+      "loss": 1.2461,
+      "step": 2508
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.701047712972584e-08,
+      "loss": 1.6875,
+      "step": 2509
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.6168557081582854e-08,
+      "loss": 1.2969,
+      "step": 2510
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.533420939923927e-08,
+      "loss": 1.2285,
+      "step": 2511
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 4.450743535806934e-08,
+      "loss": 1.3828,
+      "step": 2512
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.368823622186935e-08,
+      "loss": 1.3574,
+      "step": 2513
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.287661324285652e-08,
+      "loss": 1.2617,
+      "step": 2514
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.207256766166845e-08,
+      "loss": 1.4688,
+      "step": 2515
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.127610070735921e-08,
+      "loss": 1.3828,
+      "step": 2516
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 4.0487213597397715e-08,
+      "loss": 1.3164,
+      "step": 2517
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.970590753766712e-08,
+      "loss": 1.4766,
+      "step": 2518
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.8932183722463765e-08,
+      "loss": 1.2793,
+      "step": 2519
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.8166043334490475e-08,
+      "loss": 1.4238,
+      "step": 2520
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.7407487544861565e-08,
+      "loss": 1.3672,
+      "step": 2521
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.665651751309451e-08,
+      "loss": 1.3438,
+      "step": 2522
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 3.5913134387114414e-08,
+      "loss": 1.1367,
+      "step": 2523
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.517733930324674e-08,
+      "loss": 1.3184,
+      "step": 2524
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.444913338621847e-08,
+      "loss": 1.4473,
+      "step": 2525
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.372851774915697e-08,
+      "loss": 1.4023,
+      "step": 2526
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.301549349358668e-08,
+      "loss": 1.5059,
+      "step": 2527
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 3.2310061709426296e-08,
+      "loss": 1.2246,
+      "step": 2528
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.16122234749916e-08,
+      "loss": 1.2832,
+      "step": 2529
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.092197985698875e-08,
+      "loss": 1.4785,
+      "step": 2530
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 3.023933191051598e-08,
+      "loss": 1.5566,
+      "step": 2531
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.9564280679060255e-08,
+      "loss": 1.3516,
+      "step": 2532
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.8896827194496713e-08,
+      "loss": 1.3965,
+      "step": 2533
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.823697247708701e-08,
+      "loss": 1.5098,
+      "step": 2534
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.758471753547598e-08,
+      "loss": 1.2676,
+      "step": 2535
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6940063366693303e-08,
+      "loss": 1.3086,
+      "step": 2536
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.6303010956149065e-08,
+      "loss": 1.5527,
+      "step": 2537
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5673561277633764e-08,
+      "loss": 1.4434,
+      "step": 2538
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.5051715293316624e-08,
+      "loss": 1.3789,
+      "step": 2539
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.4437473953743962e-08,
+      "loss": 1.502,
+      "step": 2540
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.3830838197836937e-08,
+      "loss": 1.2695,
+      "step": 2541
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.3231808952891566e-08,
+      "loss": 1.4961,
+      "step": 2542
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.264038713457706e-08,
+      "loss": 1.3613,
+      "step": 2543
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.2056573646933033e-08,
+      "loss": 1.4922,
+      "step": 2544
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.148036938237008e-08,
+      "loss": 1.3711,
+      "step": 2545
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.091177522166643e-08,
+      "loss": 1.3477,
+      "step": 2546
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.0350792033969057e-08,
+      "loss": 1.6992,
+      "step": 2547
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.9797420676788692e-08,
+      "loss": 1.3652,
+      "step": 2548
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.925166199600259e-08,
+      "loss": 1.5098,
+      "step": 2549
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.8713516825851207e-08,
+      "loss": 1.0664,
+      "step": 2550
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.818298598893542e-08,
+      "loss": 1.418,
+      "step": 2551
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7660070296219856e-08,
+      "loss": 1.3027,
+      "step": 2552
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.7144770547025125e-08,
+      "loss": 1.6816,
+      "step": 2553
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.6637087529033925e-08,
+      "loss": 1.3281,
+      "step": 2554
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.613702201828271e-08,
+      "loss": 1.2812,
+      "step": 2555
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.5644574779166698e-08,
+      "loss": 1.3867,
+      "step": 2556
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.51597465644332e-08,
+      "loss": 1.3252,
+      "step": 2557
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.4682538115186052e-08,
+      "loss": 1.2656,
+      "step": 2558
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.421295016088009e-08,
+      "loss": 1.4199,
+      "step": 2559
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.3750983419321129e-08,
+      "loss": 1.332,
+      "step": 2560
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.3296638596665968e-08,
+      "loss": 1.4219,
+      "step": 2561
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.2849916387420724e-08,
+      "loss": 1.2812,
+      "step": 2562
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.241081747443862e-08,
+      "loss": 1.334,
+      "step": 2563
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.1979342528922189e-08,
+      "loss": 1.3438,
+      "step": 2564
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.1555492210418295e-08,
+      "loss": 1.2305,
+      "step": 2565
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.113926716681868e-08,
+      "loss": 1.4121,
+      "step": 2566
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0730668034360514e-08,
+      "loss": 1.2168,
+      "step": 2567
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0329695437623077e-08,
+      "loss": 1.4121,
+      "step": 2568
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.936349989528305e-09,
+      "loss": 1.1963,
+      "step": 2569
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.550632291339124e-09,
+      "loss": 1.3105,
+      "step": 2570
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.1725429326589e-09,
+      "loss": 1.5527,
+      "step": 2571
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.802082491430331e-09,
+      "loss": 1.2129,
+      "step": 2572
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.439251533934323e-09,
+      "loss": 1.2871,
+      "step": 2573
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 8.084050614791117e-09,
+      "loss": 1.4707,
+      "step": 2574
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 7.736480276954728e-09,
+      "loss": 1.4727,
+      "step": 2575
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.3965410517179426e-09,
+      "loss": 1.2422,
+      "step": 2576
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 7.064233458707326e-09,
+      "loss": 1.1914,
+      "step": 2577
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.739558005884883e-09,
+      "loss": 1.3145,
+      "step": 2578
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.4225151895447315e-09,
+      "loss": 1.4746,
+      "step": 2579
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 6.113105494314209e-09,
+      "loss": 1.584,
+      "step": 2580
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.81132939315443e-09,
+      "loss": 1.2617,
+      "step": 2581
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.517187347355846e-09,
+      "loss": 1.4355,
+      "step": 2582
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 5.230679806540462e-09,
+      "loss": 1.1172,
+      "step": 2583
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.951807208660176e-09,
+      "loss": 1.3506,
+      "step": 2584
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 4.6805699799967744e-09,
+      "loss": 1.2969,
+      "step": 2585
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.41696853515916e-09,
+      "loss": 1.3262,
+      "step": 2586
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 4.161003277085574e-09,
+      "loss": 1.4805,
+      "step": 2587
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.912674597041921e-09,
+      "loss": 1.3096,
+      "step": 2588
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.671982874619562e-09,
+      "loss": 1.1328,
+      "step": 2589
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.4389284777375242e-09,
+      "loss": 1.2324,
+      "step": 2590
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 3.213511762639732e-09,
+      "loss": 1.208,
+      "step": 2591
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.995733073895557e-09,
+      "loss": 1.3203,
+      "step": 2592
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.785592744398713e-09,
+      "loss": 1.4121,
+      "step": 2593
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.5830910953666966e-09,
+      "loss": 1.0938,
+      "step": 2594
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.3882284363413444e-09,
+      "loss": 1.2969,
+      "step": 2595
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.2010050651871673e-09,
+      "loss": 1.3672,
+      "step": 2596
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.0214212680919056e-09,
+      "loss": 1.5742,
+      "step": 2597
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.849477319564863e-09,
+      "loss": 1.3555,
+      "step": 2598
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6851734824380184e-09,
+      "loss": 1.3652,
+      "step": 2599
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5285100078632487e-09,
+      "loss": 1.4082,
+      "step": 2600
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.37948713531455e-09,
+      "loss": 1.4238,
+      "step": 2601
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2381050925869276e-09,
+      "loss": 1.4277,
+      "step": 2602
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.104364095795285e-09,
+      "loss": 1.1172,
+      "step": 2603
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 9.782643493744248e-10,
+      "loss": 1.5742,
+      "step": 2604
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 8.598060460779378e-10,
+      "loss": 1.582,
+      "step": 2605
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 7.48989366980979e-10,
+      "loss": 1.4277,
+      "step": 2606
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.45814481475826e-10,
+      "loss": 1.4727,
+      "step": 2607
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.50281547275211e-10,
+      "loss": 1.0605,
+      "step": 2608
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.623907104084335e-10,
+      "loss": 1.4629,
+      "step": 2609
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.821421052246921e-10,
+      "loss": 1.3721,
+      "step": 2610
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 3.095358543914184e-10,
+      "loss": 1.4688,
+      "step": 2611
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.4457206889316744e-10,
+      "loss": 1.207,
+      "step": 2612
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.872508480332824e-10,
+      "loss": 1.3789,
+      "step": 2613
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.3757227943167472e-10,
+      "loss": 1.4941,
+      "step": 2614
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.553643902704412e-11,
+      "loss": 1.4629,
+      "step": 2615
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 6.114339107465838e-11,
+      "loss": 1.7266,
+      "step": 2616
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.43931881469084e-11,
+      "loss": 1.1475,
+      "step": 2617
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.5285871134418372e-11,
+      "loss": 1.4629,
+      "step": 2618
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.8214692438254266e-12,
+      "loss": 1.6582,
+      "step": 2619
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 1.6289,
+      "step": 2620
+    },
+    {
+      "epoch": 5.0,
+      "step": 2620,
+      "total_flos": 96312508600320.0,
+      "train_loss": 2.4847503429150763,
+      "train_runtime": 2249.5196,
+      "train_samples_per_second": 74.44,
+      "train_steps_per_second": 1.165
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2620,
+  "num_train_epochs": 5,
+  "save_steps": 10000,
+  "total_flos": 96312508600320.0,
+  "trial_name": null,
+  "trial_params": null
+}