diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,19339 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.9986027014438754,
+  "global_step": 3219,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.061855670103093e-07,
+      "loss": 1.8911,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.123711340206186e-07,
+      "loss": 1.9268,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.185567010309279e-07,
+      "loss": 1.9531,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.247422680412372e-07,
+      "loss": 1.9146,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0309278350515464e-06,
+      "loss": 1.9058,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2371134020618557e-06,
+      "loss": 1.875,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4432989690721649e-06,
+      "loss": 1.9326,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6494845360824744e-06,
+      "loss": 1.9048,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8556701030927837e-06,
+      "loss": 1.8472,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.061855670103093e-06,
+      "loss": 1.8496,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.268041237113402e-06,
+      "loss": 1.7266,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4742268041237115e-06,
+      "loss": 1.7432,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.680412371134021e-06,
+      "loss": 1.6343,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8865979381443297e-06,
+      "loss": 1.4761,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0927835051546395e-06,
+      "loss": 1.4399,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.298969072164949e-06,
+      "loss": 1.397,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.5051546391752577e-06,
+      "loss": 1.4282,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7113402061855674e-06,
+      "loss": 1.3462,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.917525773195877e-06,
+      "loss": 1.2822,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.123711340206186e-06,
+      "loss": 1.1787,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.329896907216495e-06,
+      "loss": 1.1567,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.536082474226804e-06,
+      "loss": 1.1489,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.742268041237113e-06,
+      "loss": 1.1079,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.948453608247423e-06,
+      "loss": 1.0691,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.154639175257732e-06,
+      "loss": 1.04,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.360824742268042e-06,
+      "loss": 1.0254,
+      "step": 26
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.567010309278351e-06,
+      "loss": 0.9856,
+      "step": 27
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.7731958762886594e-06,
+      "loss": 0.9565,
+      "step": 28
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.979381443298969e-06,
+      "loss": 0.9863,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.185567010309279e-06,
+      "loss": 0.9231,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.391752577319588e-06,
+      "loss": 0.8923,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.597938144329898e-06,
+      "loss": 0.8979,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.804123711340207e-06,
+      "loss": 0.866,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.010309278350515e-06,
+      "loss": 0.8804,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.216494845360825e-06,
+      "loss": 0.8406,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.422680412371135e-06,
+      "loss": 0.8555,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.628865979381444e-06,
+      "loss": 0.8376,
+      "step": 37
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.835051546391754e-06,
+      "loss": 0.8469,
+      "step": 38
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.041237113402063e-06,
+      "loss": 0.7988,
+      "step": 39
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.247422680412371e-06,
+      "loss": 0.8198,
+      "step": 40
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.453608247422681e-06,
+      "loss": 0.7981,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.65979381443299e-06,
+      "loss": 0.7961,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.865979381443299e-06,
+      "loss": 0.7839,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.072164948453609e-06,
+      "loss": 0.8083,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.278350515463918e-06,
+      "loss": 0.782,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.484536082474226e-06,
+      "loss": 0.7925,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.690721649484536e-06,
+      "loss": 0.7332,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.896907216494846e-06,
+      "loss": 0.7576,
+      "step": 48
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0103092783505156e-05,
+      "loss": 0.7644,
+      "step": 49
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0309278350515464e-05,
+      "loss": 0.7537,
+      "step": 50
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0515463917525775e-05,
+      "loss": 0.7468,
+      "step": 51
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0721649484536083e-05,
+      "loss": 0.7415,
+      "step": 52
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0927835051546391e-05,
+      "loss": 0.7148,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1134020618556703e-05,
+      "loss": 0.719,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.134020618556701e-05,
+      "loss": 0.7112,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1546391752577319e-05,
+      "loss": 0.7083,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.175257731958763e-05,
+      "loss": 0.7295,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1958762886597938e-05,
+      "loss": 0.7034,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2164948453608248e-05,
+      "loss": 0.71,
+      "step": 59
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2371134020618558e-05,
+      "loss": 0.7161,
+      "step": 60
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2577319587628866e-05,
+      "loss": 0.6812,
+      "step": 61
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2783505154639176e-05,
+      "loss": 0.6853,
+      "step": 62
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2989690721649485e-05,
+      "loss": 0.6804,
+      "step": 63
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3195876288659795e-05,
+      "loss": 0.6851,
+      "step": 64
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3402061855670103e-05,
+      "loss": 0.6833,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3608247422680415e-05,
+      "loss": 0.6628,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3814432989690723e-05,
+      "loss": 0.6501,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.402061855670103e-05,
+      "loss": 0.6729,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4226804123711342e-05,
+      "loss": 0.6616,
+      "step": 69
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.443298969072165e-05,
+      "loss": 0.6707,
+      "step": 70
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4639175257731958e-05,
+      "loss": 0.6716,
+      "step": 71
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.484536082474227e-05,
+      "loss": 0.6423,
+      "step": 72
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5051546391752578e-05,
+      "loss": 0.656,
+      "step": 73
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5257731958762888e-05,
+      "loss": 0.6377,
+      "step": 74
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5463917525773197e-05,
+      "loss": 0.6558,
+      "step": 75
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5670103092783507e-05,
+      "loss": 0.655,
+      "step": 76
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5876288659793813e-05,
+      "loss": 0.649,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6082474226804127e-05,
+      "loss": 0.676,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6288659793814433e-05,
+      "loss": 0.6672,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6494845360824743e-05,
+      "loss": 0.647,
+      "step": 80
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6701030927835052e-05,
+      "loss": 0.6455,
+      "step": 81
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6907216494845362e-05,
+      "loss": 0.6819,
+      "step": 82
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7113402061855672e-05,
+      "loss": 0.6404,
+      "step": 83
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.731958762886598e-05,
+      "loss": 0.6053,
+      "step": 84
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.752577319587629e-05,
+      "loss": 0.6367,
+      "step": 85
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7731958762886598e-05,
+      "loss": 0.6309,
+      "step": 86
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.793814432989691e-05,
+      "loss": 0.6545,
+      "step": 87
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8144329896907217e-05,
+      "loss": 0.6169,
+      "step": 88
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8350515463917527e-05,
+      "loss": 0.6396,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8556701030927837e-05,
+      "loss": 0.6565,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8762886597938147e-05,
+      "loss": 0.6521,
+      "step": 91
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.8969072164948453e-05,
+      "loss": 0.6292,
+      "step": 92
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9175257731958766e-05,
+      "loss": 0.6221,
+      "step": 93
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9381443298969072e-05,
+      "loss": 0.6213,
+      "step": 94
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9587628865979382e-05,
+      "loss": 0.6387,
+      "step": 95
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9793814432989692e-05,
+      "loss": 0.6711,
+      "step": 96
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.6094,
+      "step": 97
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999999493704677e-05,
+      "loss": 0.5996,
+      "step": 98
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999979748192203e-05,
+      "loss": 0.584,
+      "step": 99
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999995443345168e-05,
+      "loss": 0.6058,
+      "step": 100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999991899285084e-05,
+      "loss": 0.6067,
+      "step": 101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999987342642556e-05,
+      "loss": 0.6301,
+      "step": 102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999817734221983e-05,
+      "loss": 0.6372,
+      "step": 103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999751916296506e-05,
+      "loss": 0.6249,
+      "step": 104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999675972715777e-05,
+      "loss": 0.657,
+      "step": 105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999958990355669e-05,
+      "loss": 0.647,
+      "step": 106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99994937089064e-05,
+      "loss": 0.6117,
+      "step": 107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999387388862318e-05,
+      "loss": 0.6088,
+      "step": 108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999270943532092e-05,
+      "loss": 0.5997,
+      "step": 109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999144373033642e-05,
+      "loss": 0.6069,
+      "step": 110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999007677495127e-05,
+      "loss": 0.6304,
+      "step": 111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999886085705497e-05,
+      "loss": 0.6492,
+      "step": 112
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998703911861837e-05,
+      "loss": 0.6226,
+      "step": 113
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998536842074644e-05,
+      "loss": 0.6301,
+      "step": 114
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999835964786257e-05,
+      "loss": 0.6013,
+      "step": 115
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998172329405043e-05,
+      "loss": 0.627,
+      "step": 116
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997974886891732e-05,
+      "loss": 0.6316,
+      "step": 117
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999776732052257e-05,
+      "loss": 0.627,
+      "step": 118
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997549630507738e-05,
+      "loss": 0.6255,
+      "step": 119
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997321817067662e-05,
+      "loss": 0.6256,
+      "step": 120
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999708388043303e-05,
+      "loss": 0.6289,
+      "step": 121
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999683582084477e-05,
+      "loss": 0.599,
+      "step": 122
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999657763855406e-05,
+      "loss": 0.5879,
+      "step": 123
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996309333822347e-05,
+      "loss": 0.6145,
+      "step": 124
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996030906921302e-05,
+      "loss": 0.6042,
+      "step": 125
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995742358132863e-05,
+      "loss": 0.6328,
+      "step": 126
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995443687749207e-05,
+      "loss": 0.5934,
+      "step": 127
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999513489607277e-05,
+      "loss": 0.6255,
+      "step": 128
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999481598341623e-05,
+      "loss": 0.6053,
+      "step": 129
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994486950102513e-05,
+      "loss": 0.6116,
+      "step": 130
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994147796464796e-05,
+      "loss": 0.5989,
+      "step": 131
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999379852284651e-05,
+      "loss": 0.6281,
+      "step": 132
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993439129601314e-05,
+      "loss": 0.5724,
+      "step": 133
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999306961709313e-05,
+      "loss": 0.6344,
+      "step": 134
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992689985696126e-05,
+      "loss": 0.6118,
+      "step": 135
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992300235794713e-05,
+      "loss": 0.5969,
+      "step": 136
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991900367783546e-05,
+      "loss": 0.6135,
+      "step": 137
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999149038206753e-05,
+      "loss": 0.6254,
+      "step": 138
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991070279061808e-05,
+      "loss": 0.6064,
+      "step": 139
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990640059191778e-05,
+      "loss": 0.625,
+      "step": 140
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990199722893077e-05,
+      "loss": 0.5938,
+      "step": 141
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989749270611577e-05,
+      "loss": 0.6138,
+      "step": 142
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989288702803408e-05,
+      "loss": 0.6189,
+      "step": 143
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998881801993494e-05,
+      "loss": 0.6157,
+      "step": 144
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9988337222482776e-05,
+      "loss": 0.6083,
+      "step": 145
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987846310933768e-05,
+      "loss": 0.5974,
+      "step": 146
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998734528578501e-05,
+      "loss": 0.6245,
+      "step": 147
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986834147543835e-05,
+      "loss": 0.6041,
+      "step": 148
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998631289672782e-05,
+      "loss": 0.6053,
+      "step": 149
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998578153386477e-05,
+      "loss": 0.6006,
+      "step": 150
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998524005949275e-05,
+      "loss": 0.5951,
+      "step": 151
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998468847416004e-05,
+      "loss": 0.6143,
+      "step": 152
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984126778425178e-05,
+      "loss": 0.6006,
+      "step": 153
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983554972856927e-05,
+      "loss": 0.5977,
+      "step": 154
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982973058034297e-05,
+      "loss": 0.5881,
+      "step": 155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998238103454653e-05,
+      "loss": 0.604,
+      "step": 156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981778902993096e-05,
+      "loss": 0.623,
+      "step": 157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981166663983715e-05,
+      "loss": 0.6041,
+      "step": 158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9980544318138328e-05,
+      "loss": 0.585,
+      "step": 159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997991186608712e-05,
+      "loss": 0.6034,
+      "step": 160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997926930847051e-05,
+      "loss": 0.5951,
+      "step": 161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978616645939136e-05,
+      "loss": 0.5986,
+      "step": 162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997795387915389e-05,
+      "loss": 0.6152,
+      "step": 163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977281008785874e-05,
+      "loss": 0.5977,
+      "step": 164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976598035516436e-05,
+      "loss": 0.6047,
+      "step": 165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9975904960037145e-05,
+      "loss": 0.6058,
+      "step": 166
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9975201783049804e-05,
+      "loss": 0.5726,
+      "step": 167
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974488505266443e-05,
+      "loss": 0.6058,
+      "step": 168
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973765127409323e-05,
+      "loss": 0.6108,
+      "step": 169
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973031650210925e-05,
+      "loss": 0.5962,
+      "step": 170
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997228807441396e-05,
+      "loss": 0.6226,
+      "step": 171
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971534400771376e-05,
+      "loss": 0.6088,
+      "step": 172
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970770630046325e-05,
+      "loss": 0.5883,
+      "step": 173
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969996763012198e-05,
+      "loss": 0.5864,
+      "step": 174
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9969212800452608e-05,
+      "loss": 0.6118,
+      "step": 175
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968418743161388e-05,
+      "loss": 0.6074,
+      "step": 176
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9967614591942585e-05,
+      "loss": 0.6069,
+      "step": 177
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9966800347610483e-05,
+      "loss": 0.5697,
+      "step": 178
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996597601098958e-05,
+      "loss": 0.5913,
+      "step": 179
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965141582914588e-05,
+      "loss": 0.5912,
+      "step": 180
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964297064230437e-05,
+      "loss": 0.6432,
+      "step": 181
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963442455792282e-05,
+      "loss": 0.6206,
+      "step": 182
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962577758465493e-05,
+      "loss": 0.5896,
+      "step": 183
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961702973125658e-05,
+      "loss": 0.5815,
+      "step": 184
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960818100658566e-05,
+      "loss": 0.5818,
+      "step": 185
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959923141960236e-05,
+      "loss": 0.6077,
+      "step": 186
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.99590180979369e-05,
+      "loss": 0.6157,
+      "step": 187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995810296950499e-05,
+      "loss": 0.6047,
+      "step": 188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9957177757591157e-05,
+      "loss": 0.5898,
+      "step": 189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956242463132268e-05,
+      "loss": 0.6055,
+      "step": 190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955297087075387e-05,
+      "loss": 0.5531,
+      "step": 191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9954341630377795e-05,
+      "loss": 0.5999,
+      "step": 192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995337609400698e-05,
+      "loss": 0.5878,
+      "step": 193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952400478940633e-05,
+      "loss": 0.605,
+      "step": 194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9951414786166656e-05,
+      "loss": 0.6401,
+      "step": 195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950419016683144e-05,
+      "loss": 0.5625,
+      "step": 196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949413171498417e-05,
+      "loss": 0.6052,
+      "step": 197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994839725163097e-05,
+      "loss": 0.5963,
+      "step": 198
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9947371258109522e-05,
+      "loss": 0.614,
+      "step": 199
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994633519197299e-05,
+      "loss": 0.5961,
+      "step": 200
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945289054270472e-05,
+      "loss": 0.5549,
+      "step": 201
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9944232846061284e-05,
+      "loss": 0.6011,
+      "step": 202
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943166568414933e-05,
+      "loss": 0.615,
+      "step": 203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942090222411122e-05,
+      "loss": 0.6115,
+      "step": 204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994100380913975e-05,
+      "loss": 0.5896,
+      "step": 205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99399073297009e-05,
+      "loss": 0.6154,
+      "step": 206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993880078520487e-05,
+      "loss": 0.5792,
+      "step": 207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993768417677213e-05,
+      "loss": 0.603,
+      "step": 208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936557505533346e-05,
+      "loss": 0.6184,
+      "step": 209
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935420772629374e-05,
+      "loss": 0.5775,
+      "step": 210
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9934273979211262e-05,
+      "loss": 0.5771,
+      "step": 211
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9933117126440244e-05,
+      "loss": 0.6213,
+      "step": 212
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931950215487735e-05,
+      "loss": 0.5774,
+      "step": 213
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930773247535338e-05,
+      "loss": 0.6014,
+      "step": 214
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992958622377484e-05,
+      "loss": 0.5974,
+      "step": 215
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9928389145408212e-05,
+      "loss": 0.6213,
+      "step": 216
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9927182013647604e-05,
+      "loss": 0.5649,
+      "step": 217
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9925964829715345e-05,
+      "loss": 0.5994,
+      "step": 218
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924737594843944e-05,
+      "loss": 0.5658,
+      "step": 219
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992350031027609e-05,
+      "loss": 0.6047,
+      "step": 220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922252977264642e-05,
+      "loss": 0.5845,
+      "step": 221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920995597072638e-05,
+      "loss": 0.5616,
+      "step": 222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9919728170973297e-05,
+      "loss": 0.5974,
+      "step": 223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918450700249994e-05,
+      "loss": 0.5603,
+      "step": 224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917163186196285e-05,
+      "loss": 0.5753,
+      "step": 225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915865630115898e-05,
+      "loss": 0.5938,
+      "step": 226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9914558033322723e-05,
+      "loss": 0.5862,
+      "step": 227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913240397140824e-05,
+      "loss": 0.5775,
+      "step": 228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911912722904424e-05,
+      "loss": 0.5972,
+      "step": 229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9910575011957917e-05,
+      "loss": 0.5654,
+      "step": 230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990922726565585e-05,
+      "loss": 0.5729,
+      "step": 231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907869485362945e-05,
+      "loss": 0.5835,
+      "step": 232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906501672454076e-05,
+      "loss": 0.5923,
+      "step": 233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905123828314274e-05,
+      "loss": 0.576,
+      "step": 234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990373595433874e-05,
+      "loss": 0.6031,
+      "step": 235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902338051932808e-05,
+      "loss": 0.5652,
+      "step": 236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9900930122511993e-05,
+      "loss": 0.6172,
+      "step": 237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899512167501945e-05,
+      "loss": 0.5596,
+      "step": 238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9898084188338472e-05,
+      "loss": 0.5469,
+      "step": 239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989664618646754e-05,
+      "loss": 0.5955,
+      "step": 240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989519816334525e-05,
+      "loss": 0.5951,
+      "step": 241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893740120437855e-05,
+      "loss": 0.5698,
+      "step": 242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892272059221755e-05,
+      "loss": 0.5505,
+      "step": 243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.98907939811835e-05,
+      "loss": 0.5886,
+      "step": 244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988930588781978e-05,
+      "loss": 0.5613,
+      "step": 245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887807780637415e-05,
+      "loss": 0.5607,
+      "step": 246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988629966115338e-05,
+      "loss": 0.5697,
+      "step": 247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9884781530894784e-05,
+      "loss": 0.5743,
+      "step": 248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883253391398873e-05,
+      "loss": 0.564,
+      "step": 249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988171524421302e-05,
+      "loss": 0.5549,
+      "step": 250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988016709089474e-05,
+      "loss": 0.5629,
+      "step": 251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878608933011682e-05,
+      "loss": 0.5518,
+      "step": 252
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877040772141624e-05,
+      "loss": 0.593,
+      "step": 253
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875462609872466e-05,
+      "loss": 0.5795,
+      "step": 254
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9873874447802238e-05,
+      "loss": 0.5897,
+      "step": 255
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872276287539107e-05,
+      "loss": 0.5896,
+      "step": 256
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987066813070135e-05,
+      "loss": 0.5984,
+      "step": 257
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986904997891737e-05,
+      "loss": 0.592,
+      "step": 258
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9867421833825694e-05,
+      "loss": 0.6085,
+      "step": 259
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986578369707497e-05,
+      "loss": 0.5878,
+      "step": 260
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9864135570323954e-05,
+      "loss": 0.5746,
+      "step": 261
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862477455241526e-05,
+      "loss": 0.5723,
+      "step": 262
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9860809353506676e-05,
+      "loss": 0.5707,
+      "step": 263
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9859131266808513e-05,
+      "loss": 0.5499,
+      "step": 264
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985744319684625e-05,
+      "loss": 0.5815,
+      "step": 265
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985574514532921e-05,
+      "loss": 0.5751,
+      "step": 266
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854037113976826e-05,
+      "loss": 0.5743,
+      "step": 267
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985231910451863e-05,
+      "loss": 0.6272,
+      "step": 268
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9850591118694264e-05,
+      "loss": 0.6086,
+      "step": 269
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848853158253473e-05,
+      "loss": 0.603,
+      "step": 270
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847105224956096e-05,
+      "loss": 0.5494,
+      "step": 271
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984534732057208e-05,
+      "loss": 0.594,
+      "step": 272
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843579446881457e-05,
+      "loss": 0.5906,
+      "step": 273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9841801605674358e-05,
+      "loss": 0.5581,
+      "step": 274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840013798751014e-05,
+      "loss": 0.5852,
+      "step": 275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9838216027921736e-05,
+      "loss": 0.5876,
+      "step": 276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836408295006936e-05,
+      "loss": 0.6301,
+      "step": 277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.98345906018371e-05,
+      "loss": 0.5565,
+      "step": 278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832762950252813e-05,
+      "loss": 0.5677,
+      "step": 279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9830925342104737e-05,
+      "loss": 0.5525,
+      "step": 280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829077779253615e-05,
+      "loss": 0.576,
+      "step": 281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827220263570273e-05,
+      "loss": 0.5851,
+      "step": 282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9825352796935614e-05,
+      "loss": 0.6072,
+      "step": 283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9823475381240615e-05,
+      "loss": 0.5806,
+      "step": 284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982158801838633e-05,
+      "loss": 0.5714,
+      "step": 285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981969071028389e-05,
+      "loss": 0.5549,
+      "step": 286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981778345885449e-05,
+      "loss": 0.5671,
+      "step": 287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981586626602939e-05,
+      "loss": 0.5918,
+      "step": 288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9813939133749925e-05,
+      "loss": 0.5791,
+      "step": 289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.981200206396749e-05,
+      "loss": 0.5928,
+      "step": 290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9810055058643542e-05,
+      "loss": 0.5844,
+      "step": 291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808098119749607e-05,
+      "loss": 0.5704,
+      "step": 292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806131249267256e-05,
+      "loss": 0.588,
+      "step": 293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9804154449188128e-05,
+      "loss": 0.561,
+      "step": 294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9802167721513908e-05,
+      "loss": 0.6191,
+      "step": 295
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9800171068256342e-05,
+      "loss": 0.5714,
+      "step": 296
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979816449143722e-05,
+      "loss": 0.5944,
+      "step": 297
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9796147993088387e-05,
+      "loss": 0.5638,
+      "step": 298
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.979412157525172e-05,
+      "loss": 0.6134,
+      "step": 299
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792085239979164e-05,
+      "loss": 0.582,
+      "step": 300
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790038989332686e-05,
+      "loss": 0.6173,
+      "step": 301
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97879828253843e-05,
+      "loss": 0.5697,
+      "step": 302
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9785916750216062e-05,
+      "loss": 0.5485,
+      "step": 303
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9783840765920054e-05,
+      "loss": 0.5692,
+      "step": 304
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9781754874598403e-05,
+      "loss": 0.5817,
+      "step": 305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779659078363264e-05,
+      "loss": 0.5726,
+      "step": 306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977755337933682e-05,
+      "loss": 0.5562,
+      "step": 307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775437779651278e-05,
+      "loss": 0.5546,
+      "step": 308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977331228144888e-05,
+      "loss": 0.6105,
+      "step": 309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9771176886881886e-05,
+      "loss": 0.5371,
+      "step": 310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769031598112568e-05,
+      "loss": 0.5825,
+      "step": 311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9766876417313234e-05,
+      "loss": 0.5782,
+      "step": 312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9764711346666195e-05,
+      "loss": 0.574,
+      "step": 313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762536388363784e-05,
+      "loss": 0.603,
+      "step": 314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9760351544608345e-05,
+      "loss": 0.5604,
+      "step": 315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758156817612226e-05,
+      "loss": 0.5764,
+      "step": 316
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975595220959779e-05,
+      "loss": 0.5675,
+      "step": 317
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753737722797397e-05,
+      "loss": 0.6019,
+      "step": 318
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751513359453428e-05,
+      "loss": 0.5651,
+      "step": 319
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9749279121818235e-05,
+      "loss": 0.5587,
+      "step": 320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9747035012154203e-05,
+      "loss": 0.5378,
+      "step": 321
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974478103273369e-05,
+      "loss": 0.5657,
+      "step": 322
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742517185839048e-05,
+      "loss": 0.5746,
+      "step": 323
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974024347376264e-05,
+      "loss": 0.5736,
+      "step": 324
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9737959898806794e-05,
+      "loss": 0.583,
+      "step": 325
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973566646328384e-05,
+      "loss": 0.5894,
+      "step": 326
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733363169516092e-05,
+      "loss": 0.5549,
+      "step": 327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9731050019835843e-05,
+      "loss": 0.584,
+      "step": 328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9728727016585366e-05,
+      "loss": 0.5691,
+      "step": 329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972639416211691e-05,
+      "loss": 0.5864,
+      "step": 330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9724051458792705e-05,
+      "loss": 0.5796,
+      "step": 331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721698908984953e-05,
+      "loss": 0.5763,
+      "step": 332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9719336515075817e-05,
+      "loss": 0.6029,
+      "step": 333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9716964279457436e-05,
+      "loss": 0.5631,
+      "step": 334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971458220453192e-05,
+      "loss": 0.5684,
+      "step": 335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9712190292711327e-05,
+      "loss": 0.5442,
+      "step": 336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709788546417698e-05,
+      "loss": 0.5474,
+      "step": 337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9707376968083e-05,
+      "loss": 0.5493,
+      "step": 338
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9704955560149192e-05,
+      "loss": 0.6012,
+      "step": 339
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9702524325068157e-05,
+      "loss": 0.5927,
+      "step": 340
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970008326530175e-05,
+      "loss": 0.5483,
+      "step": 341
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9697632383321755e-05,
+      "loss": 0.5359,
+      "step": 342
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969517168160992e-05,
+      "loss": 0.5596,
+      "step": 343
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969270116265793e-05,
+      "loss": 0.5521,
+      "step": 344
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690220828967406e-05,
+      "loss": 0.5686,
+      "step": 345
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968773068304991e-05,
+      "loss": 0.5583,
+      "step": 346
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968523072742694e-05,
+      "loss": 0.5854,
+      "step": 347
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682720964629928e-05,
+      "loss": 0.5686,
+      "step": 348
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.968020139720024e-05,
+      "loss": 0.6111,
+      "step": 349
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9677672027689158e-05,
+      "loss": 0.5311,
+      "step": 350
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9675132858657907e-05,
+      "loss": 0.5458,
+      "step": 351
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9672583892677624e-05,
+      "loss": 0.5676,
+      "step": 352
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670025132329364e-05,
+      "loss": 0.588,
+      "step": 353
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966745658020411e-05,
+      "loss": 0.5552,
+      "step": 354
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9664878238902744e-05,
+      "loss": 0.5602,
+      "step": 355
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966229011103608e-05,
+      "loss": 0.5774,
+      "step": 356
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9659692199224828e-05,
+      "loss": 0.5758,
+      "step": 357
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965708450609961e-05,
+      "loss": 0.5552,
+      "step": 358
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.965446703430095e-05,
+      "loss": 0.5659,
+      "step": 359
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9651839786479277e-05,
+      "loss": 0.5409,
+      "step": 360
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964920276529492e-05,
+      "loss": 0.5747,
+      "step": 361
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9646555973418096e-05,
+      "loss": 0.5681,
+      "step": 362
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643899413528926e-05,
+      "loss": 0.5875,
+      "step": 363
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641233088317416e-05,
+      "loss": 0.5562,
+      "step": 364
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9638557000483464e-05,
+      "loss": 0.5679,
+      "step": 365
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963587115273685e-05,
+      "loss": 0.5895,
+      "step": 366
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963317554779724e-05,
+      "loss": 0.5726,
+      "step": 367
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630470188394174e-05,
+      "loss": 0.554,
+      "step": 368
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9627755077267076e-05,
+      "loss": 0.5713,
+      "step": 369
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9625030217165246e-05,
+      "loss": 0.541,
+      "step": 370
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962229561084785e-05,
+      "loss": 0.5613,
+      "step": 371
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9619551261083917e-05,
+      "loss": 0.5669,
+      "step": 372
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9616797170652358e-05,
+      "loss": 0.5938,
+      "step": 373
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9614033342341938e-05,
+      "loss": 0.5823,
+      "step": 374
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611259778951283e-05,
+      "loss": 0.6033,
+      "step": 375
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960847648328888e-05,
+      "loss": 0.5376,
+      "step": 376
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960568345817306e-05,
+      "loss": 0.5676,
+      "step": 377
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9602880706432024e-05,
+      "loss": 0.5488,
+      "step": 378
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960006823090381e-05,
+      "loss": 0.5935,
+      "step": 379
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9597246034436296e-05,
+      "loss": 0.5883,
+      "step": 380
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959441411988722e-05,
+      "loss": 0.5627,
+      "step": 381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.959157249012415e-05,
+      "loss": 0.5787,
+      "step": 382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9588721148024498e-05,
+      "loss": 0.5575,
+      "step": 383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.95858600964755e-05,
+      "loss": 0.5739,
+      "step": 384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958298933837423e-05,
+      "loss": 0.5576,
+      "step": 385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958010887662759e-05,
+      "loss": 0.5829,
+      "step": 386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577218714152318e-05,
+      "loss": 0.59,
+      "step": 387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957431885387496e-05,
+      "loss": 0.5391,
+      "step": 388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9571409298731883e-05,
+      "loss": 0.5693,
+      "step": 389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.956849005166928e-05,
+      "loss": 0.5675,
+      "step": 390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9565561115643153e-05,
+      "loss": 0.5889,
+      "step": 391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.956262249361931e-05,
+      "loss": 0.5654,
+      "step": 392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.955967418857338e-05,
+      "loss": 0.5728,
+      "step": 393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9556716203490788e-05,
+      "loss": 0.5596,
+      "step": 394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9553748541366756e-05,
+      "loss": 0.5333,
+      "step": 395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550771205206315e-05,
+      "loss": 0.5686,
+      "step": 396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954778419802429e-05,
+      "loss": 0.5496,
+      "step": 397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954478752284529e-05,
+      "loss": 0.5793,
+      "step": 398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954178118270373e-05,
+      "loss": 0.5487,
+      "step": 399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953876518064379e-05,
+      "loss": 0.5403,
+      "step": 400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535739519719454e-05,
+      "loss": 0.5729,
+      "step": 401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9532704202994476e-05,
+      "loss": 0.574,
+      "step": 402
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9529659233542388e-05,
+      "loss": 0.5565,
+      "step": 403
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526604614446497e-05,
+      "loss": 0.5349,
+      "step": 404
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523540348799887e-05,
+      "loss": 0.5537,
+      "step": 405
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9520466439705395e-05,
+      "loss": 0.59,
+      "step": 406
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9517382890275644e-05,
+      "loss": 0.5651,
+      "step": 407
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9514289703633e-05,
+      "loss": 0.5494,
+      "step": 408
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511186882909596e-05,
+      "loss": 0.5658,
+      "step": 409
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950807443124732e-05,
+      "loss": 0.5665,
+      "step": 410
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950495235179781e-05,
+      "loss": 0.6187,
+      "step": 411
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501820647722458e-05,
+      "loss": 0.5387,
+      "step": 412
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9498679322192396e-05,
+      "loss": 0.5568,
+      "step": 413
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9495528378388496e-05,
+      "loss": 0.5536,
+      "step": 414
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9492367819501384e-05,
+      "loss": 0.5508,
+      "step": 415
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489197648731406e-05,
+      "loss": 0.5463,
+      "step": 416
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948601786928865e-05,
+      "loss": 0.5679,
+      "step": 417
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9482828484392927e-05,
+      "loss": 0.5536,
+      "step": 418
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9479629497273783e-05,
+      "loss": 0.5443,
+      "step": 419
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947642091117048e-05,
+      "loss": 0.5656,
+      "step": 420
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9473202729332e-05,
+      "loss": 0.5604,
+      "step": 421
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946997495501705e-05,
+      "loss": 0.5275,
+      "step": 422
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946673759149404e-05,
+      "loss": 0.5686,
+      "step": 423
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9463490642041094e-05,
+      "loss": 0.584,
+      "step": 424
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9460234109946046e-05,
+      "loss": 0.5471,
+      "step": 425
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9456967998506423e-05,
+      "loss": 0.5568,
+      "step": 426
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9453692311029463e-05,
+      "loss": 0.5647,
+      "step": 427
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.94504070508321e-05,
+      "loss": 0.5564,
+      "step": 428
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9447112221240952e-05,
+      "loss": 0.5677,
+      "step": 429
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443807825592332e-05,
+      "loss": 0.5834,
+      "step": 430
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944049386723225e-05,
+      "loss": 0.5289,
+      "step": 431
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437170349516377e-05,
+      "loss": 0.5787,
+      "step": 432
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9433837275810084e-05,
+      "loss": 0.5586,
+      "step": 433
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9430494649488404e-05,
+      "loss": 0.595,
+      "step": 434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427142473936053e-05,
+      "loss": 0.5542,
+      "step": 435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9423780752547415e-05,
+      "loss": 0.5704,
+      "step": 436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9420409488726534e-05,
+      "loss": 0.5516,
+      "step": 437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9417028685887122e-05,
+      "loss": 0.5562,
+      "step": 438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9413638347452546e-05,
+      "loss": 0.5629,
+      "step": 439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9410238476855832e-05,
+      "loss": 0.5922,
+      "step": 440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940682907753966e-05,
+      "loss": 0.5298,
+      "step": 441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940341015295635e-05,
+      "loss": 0.574,
+      "step": 442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939998170656788e-05,
+      "loss": 0.6244,
+      "step": 443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9396543741845855e-05,
+      "loss": 0.5568,
+      "step": 444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9393096262271533e-05,
+      "loss": 0.5579,
+      "step": 445
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389639271335795e-05,
+      "loss": 0.5717,
+      "step": 446
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386172772539162e-05,
+      "loss": 0.5712,
+      "step": 447
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9382696769391775e-05,
+      "loss": 0.5575,
+      "step": 448
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9379211265413402e-05,
+      "loss": 0.5648,
+      "step": 449
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375716264133433e-05,
+      "loss": 0.5718,
+      "step": 450
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.937221176909087e-05,
+      "loss": 0.5844,
+      "step": 451
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936869778383434e-05,
+      "loss": 0.5879,
+      "step": 452
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936517431192207e-05,
+      "loss": 0.5507,
+      "step": 453
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9361641356921883e-05,
+      "loss": 0.5546,
+      "step": 454
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9358098922411227e-05,
+      "loss": 0.5393,
+      "step": 455
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935454701197714e-05,
+      "loss": 0.5789,
+      "step": 456
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350985629216243e-05,
+      "loss": 0.5601,
+      "step": 457
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347414777734768e-05,
+      "loss": 0.5651,
+      "step": 458
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9343834461148524e-05,
+      "loss": 0.577,
+      "step": 459
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9340244683082902e-05,
+      "loss": 0.5596,
+      "step": 460
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933664544717288e-05,
+      "loss": 0.5409,
+      "step": 461
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9333036757063013e-05,
+      "loss": 0.5302,
+      "step": 462
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9329418616407425e-05,
+      "loss": 0.5568,
+      "step": 463
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325791028869813e-05,
+      "loss": 0.5336,
+      "step": 464
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322153998123433e-05,
+      "loss": 0.5397,
+      "step": 465
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9318507527851113e-05,
+      "loss": 0.578,
+      "step": 466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9314851621745232e-05,
+      "loss": 0.6005,
+      "step": 467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.931118628350773e-05,
+      "loss": 0.5634,
+      "step": 468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930751151685009e-05,
+      "loss": 0.5726,
+      "step": 469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930382732549335e-05,
+      "loss": 0.5472,
+      "step": 470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930013371316809e-05,
+      "loss": 0.5704,
+      "step": 471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9296430683614416e-05,
+      "loss": 0.5845,
+      "step": 472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929271824058199e-05,
+      "loss": 0.5674,
+      "step": 473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928899638783e-05,
+      "loss": 0.5558,
+      "step": 474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928526512912715e-05,
+      "loss": 0.5897,
+      "step": 475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281524468251686e-05,
+      "loss": 0.5986,
+      "step": 476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9277774408991362e-05,
+      "loss": 0.5262,
+      "step": 477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9274014955143452e-05,
+      "loss": 0.5439,
+      "step": 478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9270246110514747e-05,
+      "loss": 0.5427,
+      "step": 479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.926646787892154e-05,
+      "loss": 0.5444,
+      "step": 480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9262680264189637e-05,
+      "loss": 0.5422,
+      "step": 481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925888327015434e-05,
+      "loss": 0.5273,
+      "step": 482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.925507690066045e-05,
+      "loss": 0.5736,
+      "step": 483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251261159562257e-05,
+      "loss": 0.5673,
+      "step": 484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9247436050723548e-05,
+      "loss": 0.5562,
+      "step": 485
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9243601578017593e-05,
+      "loss": 0.5653,
+      "step": 486
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9239757745327144e-05,
+      "loss": 0.5715,
+      "step": 487
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9235904556544432e-05,
+      "loss": 0.5717,
+      "step": 488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9232042015571152e-05,
+      "loss": 0.5847,
+      "step": 489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9228170126318485e-05,
+      "loss": 0.5416,
+      "step": 490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9224288892707063e-05,
+      "loss": 0.5625,
+      "step": 491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922039831866699e-05,
+      "loss": 0.5665,
+      "step": 492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9216498408137827e-05,
+      "loss": 0.5488,
+      "step": 493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921258916506858e-05,
+      "loss": 0.5541,
+      "step": 494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208670593417723e-05,
+      "loss": 0.5425,
+      "step": 495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9204742697153155e-05,
+      "loss": 0.5524,
+      "step": 496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9200805480252236e-05,
+      "loss": 0.5366,
+      "step": 497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196858946701743e-05,
+      "loss": 0.5719,
+      "step": 498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919290310049791e-05,
+      "loss": 0.5352,
+      "step": 499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918893794564639e-05,
+      "loss": 0.5626,
+      "step": 500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918496348616225e-05,
+      "loss": 0.5779,
+      "step": 501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918097972607e-05,
+      "loss": 0.5469,
+      "step": 502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9176986669403556e-05,
+      "loss": 0.5312,
+      "step": 503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9172984320206254e-05,
+      "loss": 0.584,
+      "step": 504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9168972682530827e-05,
+      "loss": 0.5649,
+      "step": 505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164951760439427e-05,
+      "loss": 0.5428,
+      "step": 506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9160921558003604e-05,
+      "loss": 0.5974,
+      "step": 507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9156882079304297e-05,
+      "loss": 0.5662,
+      "step": 508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915283332843185e-05,
+      "loss": 0.564,
+      "step": 509
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148775309485985e-05,
+      "loss": 0.5413,
+      "step": 510
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.914470802657582e-05,
+      "loss": 0.5774,
+      "step": 511
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9140631483819845e-05,
+      "loss": 0.5466,
+      "step": 512
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136545685345927e-05,
+      "loss": 0.5447,
+      "step": 513
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132450635291312e-05,
+      "loss": 0.5272,
+      "step": 514
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9128346337802605e-05,
+      "loss": 0.5476,
+      "step": 515
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9124232797035777e-05,
+      "loss": 0.5577,
+      "step": 516
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9120110017156172e-05,
+      "loss": 0.5974,
+      "step": 517
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9115978002338466e-05,
+      "loss": 0.5289,
+      "step": 518
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9111836756766707e-05,
+      "loss": 0.5547,
+      "step": 519
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9107686284634273e-05,
+      "loss": 0.5703,
+      "step": 520
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9103526590143903e-05,
+      "loss": 0.5387,
+      "step": 521
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9099357677507656e-05,
+      "loss": 0.5487,
+      "step": 522
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909517955094694e-05,
+      "loss": 0.5542,
+      "step": 523
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909099221469249e-05,
+      "loss": 0.5626,
+      "step": 524
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9086795672984354e-05,
+      "loss": 0.5752,
+      "step": 525
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9082589930071916e-05,
+      "loss": 0.5533,
+      "step": 526
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907837499021387e-05,
+      "loss": 0.5801,
+      "step": 527
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9074150857678228e-05,
+      "loss": 0.5685,
+      "step": 528
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9069917536742303e-05,
+      "loss": 0.5417,
+      "step": 529
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.906567503169272e-05,
+      "loss": 0.5354,
+      "step": 530
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9061423346825395e-05,
+      "loss": 0.5717,
+      "step": 531
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905716248644555e-05,
+      "loss": 0.5626,
+      "step": 532
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905289245486769e-05,
+      "loss": 0.5472,
+      "step": 533
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9048613256415608e-05,
+      "loss": 0.5487,
+      "step": 534
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9044324895422383e-05,
+      "loss": 0.5734,
+      "step": 535
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.904002737623037e-05,
+      "loss": 0.5667,
+      "step": 536
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9035720703191187e-05,
+      "loss": 0.5674,
+      "step": 537
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9031404880665742e-05,
+      "loss": 0.5742,
+      "step": 538
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9027079913024192e-05,
+      "loss": 0.5786,
+      "step": 539
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022745804645958e-05,
+      "loss": 0.5771,
+      "step": 540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.901840255991972e-05,
+      "loss": 0.5884,
+      "step": 541
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9014050183243406e-05,
+      "loss": 0.5886,
+      "step": 542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.558,
+      "step": 543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005318051678495e-05,
+      "loss": 0.5503,
+      "step": 544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9000938305631975e-05,
+      "loss": 0.5404,
+      "step": 545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899654944531952e-05,
+      "loss": 0.5396,
+      "step": 546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8992151475185246e-05,
+      "loss": 0.5189,
+      "step": 547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89877443996825e-05,
+      "loss": 0.5646,
+      "step": 548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.898332822327384e-05,
+      "loss": 0.5385,
+      "step": 549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8978902950431056e-05,
+      "loss": 0.5371,
+      "step": 550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974468585635126e-05,
+      "loss": 0.5835,
+      "step": 551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970025133376252e-05,
+      "loss": 0.5638,
+      "step": 552
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8965572598153834e-05,
+      "loss": 0.5569,
+      "step": 553
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896111098447646e-05,
+      "loss": 0.5674,
+      "step": 554
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895664029686193e-05,
+      "loss": 0.5736,
+      "step": 555
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8952160539837213e-05,
+      "loss": 0.5431,
+      "step": 556
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8947671717938464e-05,
+      "loss": 0.5725,
+      "step": 557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8943173835711032e-05,
+      "loss": 0.5782,
+      "step": 558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8938666897709427e-05,
+      "loss": 0.5667,
+      "step": 559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893415090849733e-05,
+      "loss": 0.5659,
+      "step": 560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892962587264759e-05,
+      "loss": 0.543,
+      "step": 561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892509179474222e-05,
+      "loss": 0.5602,
+      "step": 562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8920548679372378e-05,
+      "loss": 0.5618,
+      "step": 563
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8915996531138388e-05,
+      "loss": 0.5376,
+      "step": 564
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891143535464971e-05,
+      "loss": 0.5562,
+      "step": 565
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906865154524943e-05,
+      "loss": 0.54,
+      "step": 566
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8902285935391832e-05,
+      "loss": 0.5425,
+      "step": 567
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8897697701887257e-05,
+      "loss": 0.5754,
+      "step": 568
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8893100458657215e-05,
+      "loss": 0.5667,
+      "step": 569
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888494210356832e-05,
+      "loss": 0.5653,
+      "step": 570
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8883878961650353e-05,
+      "loss": 0.5339,
+      "step": 571
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8879254717211136e-05,
+      "loss": 0.5681,
+      "step": 572
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874621481721645e-05,
+      "loss": 0.5513,
+      "step": 573
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8869979259873452e-05,
+      "loss": 0.5807,
+      "step": 574
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.886532805636723e-05,
+      "loss": 0.5636,
+      "step": 575
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8860667875912743e-05,
+      "loss": 0.5618,
+      "step": 576
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855998723228845e-05,
+      "loss": 0.5454,
+      "step": 577
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885132060304348e-05,
+      "loss": 0.5562,
+      "step": 578
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884663352009366e-05,
+      "loss": 0.5686,
+      "step": 579
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884193747912549e-05,
+      "loss": 0.5563,
+      "step": 580
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837232484894132e-05,
+      "loss": 0.5544,
+      "step": 581
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8832518542163823e-05,
+      "loss": 0.5767,
+      "step": 582
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8827795655707858e-05,
+      "loss": 0.5563,
+      "step": 583
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882306383030858e-05,
+      "loss": 0.5325,
+      "step": 584
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8818323070757397e-05,
+      "loss": 0.5328,
+      "step": 585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881357338185476e-05,
+      "loss": 0.568,
+      "step": 586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8808814768410157e-05,
+      "loss": 0.5778,
+      "step": 587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8804047235242114e-05,
+      "loss": 0.5448,
+      "step": 588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879927078717819e-05,
+      "loss": 0.5114,
+      "step": 589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8794485429054975e-05,
+      "loss": 0.5676,
+      "step": 590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8789691165718075e-05,
+      "loss": 0.5621,
+      "step": 591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.878488800202212e-05,
+      "loss": 0.5632,
+      "step": 592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780075942830742e-05,
+      "loss": 0.605,
+      "step": 593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8775254993016596e-05,
+      "loss": 0.5931,
+      "step": 594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877042515746132e-05,
+      "loss": 0.5555,
+      "step": 595
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765586441055572e-05,
+      "loss": 0.5367,
+      "step": 596
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8760738848698982e-05,
+      "loss": 0.5664,
+      "step": 597
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8755882385300182e-05,
+      "loss": 0.5433,
+      "step": 598
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8751017055776778e-05,
+      "loss": 0.5182,
+      "step": 599
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8746142865055354e-05,
+      "loss": 0.5471,
+      "step": 600
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874125981807148e-05,
+      "loss": 0.5641,
+      "step": 601
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8736367919769674e-05,
+      "loss": 0.5481,
+      "step": 602
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8731467175103433e-05,
+      "loss": 0.5299,
+      "step": 603
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726557589035202e-05,
+      "loss": 0.5356,
+      "step": 604
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872163916653638e-05,
+      "loss": 0.5312,
+      "step": 605
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.871671191258732e-05,
+      "loss": 0.5568,
+      "step": 606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8711775832177315e-05,
+      "loss": 0.5297,
+      "step": 607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8706830930304587e-05,
+      "loss": 0.5551,
+      "step": 608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8701877211976302e-05,
+      "loss": 0.557,
+      "step": 609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8696914682208547e-05,
+      "loss": 0.5581,
+      "step": 610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8691943346026333e-05,
+      "loss": 0.538,
+      "step": 611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8686963208463588e-05,
+      "loss": 0.5354,
+      "step": 612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868197427456316e-05,
+      "loss": 0.564,
+      "step": 613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8676976549376786e-05,
+      "loss": 0.5392,
+      "step": 614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867197003796512e-05,
+      "loss": 0.5344,
+      "step": 615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8666954745397708e-05,
+      "loss": 0.5664,
+      "step": 616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8661930676752993e-05,
+      "loss": 0.59,
+      "step": 617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8656897837118294e-05,
+      "loss": 0.563,
+      "step": 618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8651856231589822e-05,
+      "loss": 0.5721,
+      "step": 619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8646805865272656e-05,
+      "loss": 0.5734,
+      "step": 620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8641746743280753e-05,
+      "loss": 0.5769,
+      "step": 621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636678870736928e-05,
+      "loss": 0.5371,
+      "step": 622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8631602252772865e-05,
+      "loss": 0.5693,
+      "step": 623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8626516894529102e-05,
+      "loss": 0.6096,
+      "step": 624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8621422801155015e-05,
+      "loss": 0.5652,
+      "step": 625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616319977808846e-05,
+      "loss": 0.5291,
+      "step": 626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8611208429657667e-05,
+      "loss": 0.5453,
+      "step": 627
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8606088161877376e-05,
+      "loss": 0.5526,
+      "step": 628
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8600959179652708e-05,
+      "loss": 0.5658,
+      "step": 629
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.859582148817723e-05,
+      "loss": 0.5623,
+      "step": 630
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8590675092653315e-05,
+      "loss": 0.5861,
+      "step": 631
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858551999829216e-05,
+      "loss": 0.5653,
+      "step": 632
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8580356210313756e-05,
+      "loss": 0.5518,
+      "step": 633
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575183733946915e-05,
+      "loss": 0.5466,
+      "step": 634
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857000257442924e-05,
+      "loss": 0.5546,
+      "step": 635
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8564812737007113e-05,
+      "loss": 0.5889,
+      "step": 636
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8559614226935725e-05,
+      "loss": 0.5609,
+      "step": 637
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8554407049479032e-05,
+      "loss": 0.5704,
+      "step": 638
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.854919120990978e-05,
+      "loss": 0.522,
+      "step": 639
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8543966713509473e-05,
+      "loss": 0.5306,
+      "step": 640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538733565568394e-05,
+      "loss": 0.556,
+      "step": 641
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8533491771385573e-05,
+      "loss": 0.5306,
+      "step": 642
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852824133626881e-05,
+      "loss": 0.53,
+      "step": 643
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522982265534634e-05,
+      "loss": 0.5505,
+      "step": 644
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517714564508342e-05,
+      "loss": 0.5139,
+      "step": 645
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8512438238523956e-05,
+      "loss": 0.5455,
+      "step": 646
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850715329292423e-05,
+      "loss": 0.5638,
+      "step": 647
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8501859733060654e-05,
+      "loss": 0.5265,
+      "step": 648
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8496557564293435e-05,
+      "loss": 0.569,
+      "step": 649
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8491246791991506e-05,
+      "loss": 0.5353,
+      "step": 650
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8485927421532498e-05,
+      "loss": 0.569,
+      "step": 651
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480599458302756e-05,
+      "loss": 0.5367,
+      "step": 652
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847526290769733e-05,
+      "loss": 0.5737,
+      "step": 653
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8469917775119962e-05,
+      "loss": 0.5267,
+      "step": 654
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8464564065983078e-05,
+      "loss": 0.5641,
+      "step": 655
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8459201785707796e-05,
+      "loss": 0.5458,
+      "step": 656
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8453830939723913e-05,
+      "loss": 0.556,
+      "step": 657
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8448451533469896e-05,
+      "loss": 0.5586,
+      "step": 658
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844306357239288e-05,
+      "loss": 0.5483,
+      "step": 659
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8437667061948664e-05,
+      "loss": 0.5693,
+      "step": 660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8432262007601706e-05,
+      "loss": 0.5206,
+      "step": 661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8426848414825115e-05,
+      "loss": 0.5441,
+      "step": 662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8421426289100636e-05,
+      "loss": 0.5377,
+      "step": 663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8415995635918677e-05,
+      "loss": 0.5461,
+      "step": 664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8410556460778252e-05,
+      "loss": 0.5652,
+      "step": 665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8405108769187024e-05,
+      "loss": 0.5463,
+      "step": 666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399652566661282e-05,
+      "loss": 0.5514,
+      "step": 667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8394187858725914e-05,
+      "loss": 0.5764,
+      "step": 668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8388714650914443e-05,
+      "loss": 0.5637,
+      "step": 669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8383232948768978e-05,
+      "loss": 0.5543,
+      "step": 670
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377742757840246e-05,
+      "loss": 0.5658,
+      "step": 671
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8372244083687557e-05,
+      "loss": 0.5461,
+      "step": 672
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8366736931878825e-05,
+      "loss": 0.5366,
+      "step": 673
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8361221307990535e-05,
+      "loss": 0.5347,
+      "step": 674
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835569721760776e-05,
+      "loss": 0.5475,
+      "step": 675
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8350164666324137e-05,
+      "loss": 0.5768,
+      "step": 676
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8344623659741877e-05,
+      "loss": 0.5348,
+      "step": 677
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833907420347176e-05,
+      "loss": 0.5293,
+      "step": 678
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.83335163031331e-05,
+      "loss": 0.5552,
+      "step": 679
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8327949964353782e-05,
+      "loss": 0.5273,
+      "step": 680
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832237519277023e-05,
+      "loss": 0.553,
+      "step": 681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.83167919940274e-05,
+      "loss": 0.5486,
+      "step": 682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311200373778792e-05,
+      "loss": 0.5237,
+      "step": 683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8305600337686428e-05,
+      "loss": 0.5461,
+      "step": 684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8299991891420848e-05,
+      "loss": 0.5237,
+      "step": 685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294375040661116e-05,
+      "loss": 0.554,
+      "step": 686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.82887497910948e-05,
+      "loss": 0.5693,
+      "step": 687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8283116148417982e-05,
+      "loss": 0.5498,
+      "step": 688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8277474118335227e-05,
+      "loss": 0.5634,
+      "step": 689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8271823706559604e-05,
+      "loss": 0.5638,
+      "step": 690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826616491881267e-05,
+      "loss": 0.5339,
+      "step": 691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.826049776082446e-05,
+      "loss": 0.5621,
+      "step": 692
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254822238333482e-05,
+      "loss": 0.5869,
+      "step": 693
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824913835708672e-05,
+      "loss": 0.5797,
+      "step": 694
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824344612283962e-05,
+      "loss": 0.5558,
+      "step": 695
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823774554135608e-05,
+      "loss": 0.5372,
+      "step": 696
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.823203661840846e-05,
+      "loss": 0.5721,
+      "step": 697
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8226319359777563e-05,
+      "loss": 0.573,
+      "step": 698
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822059377125263e-05,
+      "loss": 0.5277,
+      "step": 699
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8214859858631334e-05,
+      "loss": 0.5559,
+      "step": 700
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8209117627719788e-05,
+      "loss": 0.556,
+      "step": 701
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8203367084332513e-05,
+      "loss": 0.5298,
+      "step": 702
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8197608234292466e-05,
+      "loss": 0.5355,
+      "step": 703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8191841083431e-05,
+      "loss": 0.549,
+      "step": 704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818606563758788e-05,
+      "loss": 0.5397,
+      "step": 705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818028190261126e-05,
+      "loss": 0.5281,
+      "step": 706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8174489884357705e-05,
+      "loss": 0.5151,
+      "step": 707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816868958869215e-05,
+      "loss": 0.5647,
+      "step": 708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.816288102148793e-05,
+      "loss": 0.5323,
+      "step": 709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8157064188626744e-05,
+      "loss": 0.576,
+      "step": 710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8151239095998655e-05,
+      "loss": 0.5139,
+      "step": 711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8145405749502105e-05,
+      "loss": 0.5786,
+      "step": 712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8139564155043885e-05,
+      "loss": 0.5568,
+      "step": 713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8133714318539134e-05,
+      "loss": 0.5608,
+      "step": 714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8127856245911347e-05,
+      "loss": 0.5408,
+      "step": 715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812198994309235e-05,
+      "loss": 0.5388,
+      "step": 716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8116115416022308e-05,
+      "loss": 0.5421,
+      "step": 717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8110232670649712e-05,
+      "loss": 0.5421,
+      "step": 718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8104341712931376e-05,
+      "loss": 0.5641,
+      "step": 719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8098442548832428e-05,
+      "loss": 0.5491,
+      "step": 720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80925351843263e-05,
+      "loss": 0.5083,
+      "step": 721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8086619625394748e-05,
+      "loss": 0.5521,
+      "step": 722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.80806958780278e-05,
+      "loss": 0.5641,
+      "step": 723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8074763948223793e-05,
+      "loss": 0.519,
+      "step": 724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.806882384198934e-05,
+      "loss": 0.5421,
+      "step": 725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8062875565339338e-05,
+      "loss": 0.5288,
+      "step": 726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056919124296957e-05,
+      "loss": 0.5685,
+      "step": 727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8050954524893637e-05,
+      "loss": 0.5765,
+      "step": 728
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8044981773169064e-05,
+      "loss": 0.5408,
+      "step": 729
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8039000875171203e-05,
+      "loss": 0.574,
+      "step": 730
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8033011836956252e-05,
+      "loss": 0.5565,
+      "step": 731
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8027014664588654e-05,
+      "loss": 0.5504,
+      "step": 732
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8021009364141087e-05,
+      "loss": 0.5638,
+      "step": 733
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.801499594169447e-05,
+      "loss": 0.5571,
+      "step": 734
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8008974403337925e-05,
+      "loss": 0.579,
+      "step": 735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800294475516882e-05,
+      "loss": 0.5536,
+      "step": 736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7996907003292708e-05,
+      "loss": 0.5253,
+      "step": 737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990861153823373e-05,
+      "loss": 0.5513,
+      "step": 738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.798480721288277e-05,
+      "loss": 0.546,
+      "step": 739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7978745186601078e-05,
+      "loss": 0.5371,
+      "step": 740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797267508111664e-05,
+      "loss": 0.5497,
+      "step": 741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966596902575985e-05,
+      "loss": 0.5769,
+      "step": 742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7960510657133828e-05,
+      "loss": 0.5577,
+      "step": 743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7954416350953037e-05,
+      "loss": 0.5397,
+      "step": 744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7948313990204654e-05,
+      "loss": 0.5537,
+      "step": 745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7942203581067875e-05,
+      "loss": 0.5668,
+      "step": 746
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7936085129730036e-05,
+      "loss": 0.5519,
+      "step": 747
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7929958642386627e-05,
+      "loss": 0.5847,
+      "step": 748
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7923824125241273e-05,
+      "loss": 0.5397,
+      "step": 749
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791768158450573e-05,
+      "loss": 0.5309,
+      "step": 750
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791153102639987e-05,
+      "loss": 0.5717,
+      "step": 751
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7905372457151698e-05,
+      "loss": 0.5753,
+      "step": 752
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7899205882997322e-05,
+      "loss": 0.512,
+      "step": 753
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789303131018095e-05,
+      "loss": 0.5237,
+      "step": 754
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.788684874495491e-05,
+      "loss": 0.5583,
+      "step": 755
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.78806581935796e-05,
+      "loss": 0.5314,
+      "step": 756
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7874459662323516e-05,
+      "loss": 0.5226,
+      "step": 757
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7868253157463236e-05,
+      "loss": 0.5374,
+      "step": 758
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7862038685283404e-05,
+      "loss": 0.5646,
+      "step": 759
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.785581625207674e-05,
+      "loss": 0.5503,
+      "step": 760
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784958586414402e-05,
+      "loss": 0.5492,
+      "step": 761
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.784334752779408e-05,
+      "loss": 0.5593,
+      "step": 762
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7837101249343798e-05,
+      "loss": 0.5256,
+      "step": 763
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7830847035118095e-05,
+      "loss": 0.5336,
+      "step": 764
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7824584891449934e-05,
+      "loss": 0.5325,
+      "step": 765
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78183148246803e-05,
+      "loss": 0.526,
+      "step": 766
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7812036841158205e-05,
+      "loss": 0.5466,
+      "step": 767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7805750947240677e-05,
+      "loss": 0.5447,
+      "step": 768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7799457149292752e-05,
+      "loss": 0.574,
+      "step": 769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7793155453687477e-05,
+      "loss": 0.5262,
+      "step": 770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7786845866805878e-05,
+      "loss": 0.5687,
+      "step": 771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7780528395036997e-05,
+      "loss": 0.5437,
+      "step": 772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777420304477784e-05,
+      "loss": 0.5735,
+      "step": 773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.776786982243339e-05,
+      "loss": 0.5472,
+      "step": 774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7761528734416623e-05,
+      "loss": 0.5551,
+      "step": 775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.775517978714846e-05,
+      "loss": 0.5292,
+      "step": 776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7748822987057782e-05,
+      "loss": 0.5155,
+      "step": 777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.774245834058143e-05,
+      "loss": 0.558,
+      "step": 778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773608585416418e-05,
+      "loss": 0.5471,
+      "step": 779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729705534258758e-05,
+      "loss": 0.5442,
+      "step": 780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.772331738732582e-05,
+      "loss": 0.5186,
+      "step": 781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7716921419833927e-05,
+      "loss": 0.5138,
+      "step": 782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7710517638259593e-05,
+      "loss": 0.5426,
+      "step": 783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7704106049087218e-05,
+      "loss": 0.5107,
+      "step": 784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7697686658809124e-05,
+      "loss": 0.5331,
+      "step": 785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7691259473925518e-05,
+      "loss": 0.53,
+      "step": 786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684824500944514e-05,
+      "loss": 0.5543,
+      "step": 787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7678381746382097e-05,
+      "loss": 0.5635,
+      "step": 788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767193121676215e-05,
+      "loss": 0.5292,
+      "step": 789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7665472918616412e-05,
+      "loss": 0.5701,
+      "step": 790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76590068584845e-05,
+      "loss": 0.542,
+      "step": 791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7652533042913885e-05,
+      "loss": 0.5698,
+      "step": 792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764605147845989e-05,
+      "loss": 0.5414,
+      "step": 793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7639562171685683e-05,
+      "loss": 0.5673,
+      "step": 794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633065129162282e-05,
+      "loss": 0.5546,
+      "step": 795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7626560357468533e-05,
+      "loss": 0.5245,
+      "step": 796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.76200478631911e-05,
+      "loss": 0.5475,
+      "step": 797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7613527652924475e-05,
+      "loss": 0.5256,
+      "step": 798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7606999733270967e-05,
+      "loss": 0.5294,
+      "step": 799
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7600464110840682e-05,
+      "loss": 0.5295,
+      "step": 800
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759392079225153e-05,
+      "loss": 0.5444,
+      "step": 801
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7587369784129217e-05,
+      "loss": 0.5377,
+      "step": 802
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.758081109310723e-05,
+      "loss": 0.5341,
+      "step": 803
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757424472582684e-05,
+      "loss": 0.5376,
+      "step": 804
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.756767068893709e-05,
+      "loss": 0.5597,
+      "step": 805
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7561088989094786e-05,
+      "loss": 0.5747,
+      "step": 806
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7554499632964496e-05,
+      "loss": 0.5361,
+      "step": 807
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754790262721854e-05,
+      "loss": 0.5444,
+      "step": 808
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7541297978536987e-05,
+      "loss": 0.5305,
+      "step": 809
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753468569360764e-05,
+      "loss": 0.5399,
+      "step": 810
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7528065779126035e-05,
+      "loss": 0.5409,
+      "step": 811
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752143824179544e-05,
+      "loss": 0.5767,
+      "step": 812
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7514803088326836e-05,
+      "loss": 0.5966,
+      "step": 813
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.750816032543891e-05,
+      "loss": 0.552,
+      "step": 814
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7501509959858078e-05,
+      "loss": 0.5619,
+      "step": 815
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7494851998318422e-05,
+      "loss": 0.551,
+      "step": 816
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7488186447561738e-05,
+      "loss": 0.5487,
+      "step": 817
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.74815133143375e-05,
+      "loss": 0.5385,
+      "step": 818
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7474832605402862e-05,
+      "loss": 0.5352,
+      "step": 819
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7468144327522643e-05,
+      "loss": 0.5745,
+      "step": 820
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7461448487469338e-05,
+      "loss": 0.5474,
+      "step": 821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7454745092023083e-05,
+      "loss": 0.5546,
+      "step": 822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7448034147971678e-05,
+      "loss": 0.5568,
+      "step": 823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7441315662110564e-05,
+      "loss": 0.5551,
+      "step": 824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434589641242814e-05,
+      "loss": 0.5719,
+      "step": 825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7427856092179136e-05,
+      "loss": 0.5299,
+      "step": 826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7421115021737853e-05,
+      "loss": 0.5408,
+      "step": 827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7414366436744917e-05,
+      "loss": 0.5422,
+      "step": 828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7407610344033883e-05,
+      "loss": 0.5364,
+      "step": 829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7400846750445903e-05,
+      "loss": 0.5464,
+      "step": 830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7394075662829727e-05,
+      "loss": 0.5477,
+      "step": 831
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7387297088041696e-05,
+      "loss": 0.524,
+      "step": 832
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7380511032945734e-05,
+      "loss": 0.5332,
+      "step": 833
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7373717504413334e-05,
+      "loss": 0.5304,
+      "step": 834
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7366916509323563e-05,
+      "loss": 0.5771,
+      "step": 835
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7360108054563042e-05,
+      "loss": 0.5364,
+      "step": 836
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7353292147025953e-05,
+      "loss": 0.5277,
+      "step": 837
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7346468793614015e-05,
+      "loss": 0.542,
+      "step": 838
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7339638001236495e-05,
+      "loss": 0.5627,
+      "step": 839
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7332799776810184e-05,
+      "loss": 0.5114,
+      "step": 840
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7325954127259413e-05,
+      "loss": 0.5519,
+      "step": 841
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7319101059516015e-05,
+      "loss": 0.5525,
+      "step": 842
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7312240580519346e-05,
+      "loss": 0.5361,
+      "step": 843
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7305372697216264e-05,
+      "loss": 0.54,
+      "step": 844
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729849741656112e-05,
+      "loss": 0.5885,
+      "step": 845
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729161474551576e-05,
+      "loss": 0.5431,
+      "step": 846
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7284724691049516e-05,
+      "loss": 0.5441,
+      "step": 847
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7277827260139183e-05,
+      "loss": 0.5305,
+      "step": 848
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7270922459769044e-05,
+      "loss": 0.538,
+      "step": 849
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.726401029693084e-05,
+      "loss": 0.541,
+      "step": 850
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7257090778623745e-05,
+      "loss": 0.546,
+      "step": 851
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7250163911854415e-05,
+      "loss": 0.537,
+      "step": 852
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7243229703636924e-05,
+      "loss": 0.5394,
+      "step": 853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7236288160992788e-05,
+      "loss": 0.5439,
+      "step": 854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229339290950943e-05,
+      "loss": 0.5553,
+      "step": 855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7222383100547754e-05,
+      "loss": 0.5493,
+      "step": 856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7215419596826994e-05,
+      "loss": 0.541,
+      "step": 857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720844878683984e-05,
+      "loss": 0.5388,
+      "step": 858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7201470677644873e-05,
+      "loss": 0.5444,
+      "step": 859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7194485276308058e-05,
+      "loss": 0.5184,
+      "step": 860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718749258990275e-05,
+      "loss": 0.5421,
+      "step": 861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7180492625509675e-05,
+      "loss": 0.541,
+      "step": 862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717348539021693e-05,
+      "loss": 0.5487,
+      "step": 863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.716647089111998e-05,
+      "loss": 0.5554,
+      "step": 864
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7159449135321636e-05,
+      "loss": 0.5488,
+      "step": 865
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.715242012993207e-05,
+      "loss": 0.5807,
+      "step": 866
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714538388206878e-05,
+      "loss": 0.4967,
+      "step": 867
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7138340398856605e-05,
+      "loss": 0.5549,
+      "step": 868
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7131289687427718e-05,
+      "loss": 0.5674,
+      "step": 869
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124231754921593e-05,
+      "loss": 0.5383,
+      "step": 870
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7117166608485033e-05,
+      "loss": 0.552,
+      "step": 871
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711009425527214e-05,
+      "loss": 0.5415,
+      "step": 872
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710301470244431e-05,
+      "loss": 0.5369,
+      "step": 873
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709592795717023e-05,
+      "loss": 0.5323,
+      "step": 874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7088834026625872e-05,
+      "loss": 0.5452,
+      "step": 875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7081732917994488e-05,
+      "loss": 0.5404,
+      "step": 876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7074624638466595e-05,
+      "loss": 0.49,
+      "step": 877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7067509195239966e-05,
+      "loss": 0.5426,
+      "step": 878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.706038659551963e-05,
+      "loss": 0.5363,
+      "step": 879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7053256846517874e-05,
+      "loss": 0.517,
+      "step": 880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704611995545421e-05,
+      "loss": 0.5549,
+      "step": 881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7038975929555382e-05,
+      "loss": 0.5184,
+      "step": 882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703182477605537e-05,
+      "loss": 0.5642,
+      "step": 883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.702466650219537e-05,
+      "loss": 0.5405,
+      "step": 884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701750111522377e-05,
+      "loss": 0.5403,
+      "step": 885
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7010328622396183e-05,
+      "loss": 0.5498,
+      "step": 886
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7003149030975406e-05,
+      "loss": 0.5342,
+      "step": 887
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6995962348231424e-05,
+      "loss": 0.524,
+      "step": 888
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.698876858144141e-05,
+      "loss": 0.5325,
+      "step": 889
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.69815677378897e-05,
+      "loss": 0.5381,
+      "step": 890
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6974359824867802e-05,
+      "loss": 0.5326,
+      "step": 891
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6967144849674385e-05,
+      "loss": 0.5217,
+      "step": 892
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6959922819615263e-05,
+      "loss": 0.5758,
+      "step": 893
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.695269374200339e-05,
+      "loss": 0.52,
+      "step": 894
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694545762415887e-05,
+      "loss": 0.5618,
+      "step": 895
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6938214473408928e-05,
+      "loss": 0.5406,
+      "step": 896
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6930964297087905e-05,
+      "loss": 0.5503,
+      "step": 897
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692370710253727e-05,
+      "loss": 0.5197,
+      "step": 898
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6916442897105586e-05,
+      "loss": 0.5127,
+      "step": 899
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909171688148514e-05,
+      "loss": 0.4993,
+      "step": 900
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6901893483028818e-05,
+      "loss": 0.5201,
+      "step": 901
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6894608289116344e-05,
+      "loss": 0.5569,
+      "step": 902
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6887316113788004e-05,
+      "loss": 0.5297,
+      "step": 903
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6880016964427792e-05,
+      "loss": 0.5459,
+      "step": 904
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6872710848426753e-05,
+      "loss": 0.561,
+      "step": 905
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6865397773183e-05,
+      "loss": 0.533,
+      "step": 906
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6858077746101674e-05,
+      "loss": 0.5292,
+      "step": 907
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6850750774594972e-05,
+      "loss": 0.5444,
+      "step": 908
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6843416866082118e-05,
+      "loss": 0.5646,
+      "step": 909
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6836076027989354e-05,
+      "loss": 0.5388,
+      "step": 910
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.682872826774995e-05,
+      "loss": 0.5544,
+      "step": 911
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6821373592804174e-05,
+      "loss": 0.5306,
+      "step": 912
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6814012010599308e-05,
+      "loss": 0.5544,
+      "step": 913
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680664352858961e-05,
+      "loss": 0.5526,
+      "step": 914
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799268154236347e-05,
+      "loss": 0.5596,
+      "step": 915
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6791885895007748e-05,
+      "loss": 0.5756,
+      "step": 916
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6784496758379024e-05,
+      "loss": 0.5098,
+      "step": 917
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6777100751832342e-05,
+      "loss": 0.5692,
+      "step": 918
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6769697882856832e-05,
+      "loss": 0.5763,
+      "step": 919
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6762288158948563e-05,
+      "loss": 0.5536,
+      "step": 920
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6754871587610562e-05,
+      "loss": 0.5426,
+      "step": 921
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.674744817635277e-05,
+      "loss": 0.54,
+      "step": 922
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740017932692073e-05,
+      "loss": 0.5365,
+      "step": 923
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673258086415226e-05,
+      "loss": 0.5659,
+      "step": 924
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672513697826404e-05,
+      "loss": 0.557,
+      "step": 925
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6717686282565024e-05,
+      "loss": 0.5713,
+      "step": 926
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.671022878459971e-05,
+      "loss": 0.5333,
+      "step": 927
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6702764491919494e-05,
+      "loss": 0.5115,
+      "step": 928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6695293412082654e-05,
+      "loss": 0.5553,
+      "step": 929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687815552654328e-05,
+      "loss": 0.545,
+      "step": 930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.668033092120653e-05,
+      "loss": 0.5327,
+      "step": 931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6672839525318125e-05,
+      "loss": 0.582,
+      "step": 932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6665341372574838e-05,
+      "loss": 0.5385,
+      "step": 933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665783647056922e-05,
+      "loss": 0.5317,
+      "step": 934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665032482690067e-05,
+      "loss": 0.5542,
+      "step": 935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66428064491754e-05,
+      "loss": 0.5897,
+      "step": 936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.663528134500646e-05,
+      "loss": 0.5496,
+      "step": 937
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6627749522013697e-05,
+      "loss": 0.5311,
+      "step": 938
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662021098782376e-05,
+      "loss": 0.5382,
+      "step": 939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.66126657500701e-05,
+      "loss": 0.5521,
+      "step": 940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6605113816392954e-05,
+      "loss": 0.5233,
+      "step": 941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6597555194439342e-05,
+      "loss": 0.5513,
+      "step": 942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.658998989186305e-05,
+      "loss": 0.5322,
+      "step": 943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6582417916324636e-05,
+      "loss": 0.5238,
+      "step": 944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657483927549141e-05,
+      "loss": 0.5171,
+      "step": 945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6567253977037435e-05,
+      "loss": 0.5206,
+      "step": 946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655966202864351e-05,
+      "loss": 0.5349,
+      "step": 947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6552063437997175e-05,
+      "loss": 0.4974,
+      "step": 948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544458212792687e-05,
+      "loss": 0.5546,
+      "step": 949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6536846360731024e-05,
+      "loss": 0.5446,
+      "step": 950
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6529227889519884e-05,
+      "loss": 0.5724,
+      "step": 951
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652160280687366e-05,
+      "loss": 0.5582,
+      "step": 952
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.651397112051343e-05,
+      "loss": 0.5592,
+      "step": 953
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6506332838166974e-05,
+      "loss": 0.5107,
+      "step": 954
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6498687967568747e-05,
+      "loss": 0.5262,
+      "step": 955
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.649103651645987e-05,
+      "loss": 0.5321,
+      "step": 956
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6483378492588133e-05,
+      "loss": 0.5287,
+      "step": 957
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6475713903707977e-05,
+      "loss": 0.5211,
+      "step": 958
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.64680427575805e-05,
+      "loss": 0.5326,
+      "step": 959
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646036506197342e-05,
+      "loss": 0.5293,
+      "step": 960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6452680824661107e-05,
+      "loss": 0.5447,
+      "step": 961
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6444990053424552e-05,
+      "loss": 0.5007,
+      "step": 962
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6437292756051354e-05,
+      "loss": 0.5492,
+      "step": 963
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642958894033572e-05,
+      "loss": 0.5238,
+      "step": 964
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642187861407847e-05,
+      "loss": 0.5183,
+      "step": 965
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6414161785087e-05,
+      "loss": 0.5308,
+      "step": 966
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406438461175308e-05,
+      "loss": 0.5187,
+      "step": 967
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6398708650163952e-05,
+      "loss": 0.5671,
+      "step": 968
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639097235988007e-05,
+      "loss": 0.5559,
+      "step": 969
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6383229598157356e-05,
+      "loss": 0.5725,
+      "step": 970
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6375480372836054e-05,
+      "loss": 0.5723,
+      "step": 971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6367724691762967e-05,
+      "loss": 0.543,
+      "step": 972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359962562791416e-05,
+      "loss": 0.5201,
+      "step": 973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352193993781263e-05,
+      "loss": 0.5657,
+      "step": 974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634441899259889e-05,
+      "loss": 0.5116,
+      "step": 975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633663756711719e-05,
+      "loss": 0.5594,
+      "step": 976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632884972521556e-05,
+      "loss": 0.5524,
+      "step": 977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63210554747799e-05,
+      "loss": 0.545,
+      "step": 978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631325482370259e-05,
+      "loss": 0.5239,
+      "step": 979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63054477798825e-05,
+      "loss": 0.5514,
+      "step": 980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6297634351224964e-05,
+      "loss": 0.532,
+      "step": 981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6289814545641792e-05,
+      "loss": 0.5425,
+      "step": 982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6281988371051244e-05,
+      "loss": 0.5458,
+      "step": 983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6274155835378032e-05,
+      "loss": 0.5492,
+      "step": 984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6266316946553312e-05,
+      "loss": 0.5161,
+      "step": 985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625847171251466e-05,
+      "loss": 0.5458,
+      "step": 986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6250620141206096e-05,
+      "loss": 0.5466,
+      "step": 987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.624276224057804e-05,
+      "loss": 0.5328,
+      "step": 988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.5483,
+      "step": 989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622702748319722e-05,
+      "loss": 0.5488,
+      "step": 990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6219150642377314e-05,
+      "loss": 0.5533,
+      "step": 991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6211267504103647e-05,
+      "loss": 0.5552,
+      "step": 992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6203378076358602e-05,
+      "loss": 0.5687,
+      "step": 993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619548236713094e-05,
+      "loss": 0.5269,
+      "step": 994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6187580384415785e-05,
+      "loss": 0.5162,
+      "step": 995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617967213621461e-05,
+      "loss": 0.5306,
+      "step": 996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617175763053523e-05,
+      "loss": 0.4924,
+      "step": 997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6163836875391808e-05,
+      "loss": 0.5189,
+      "step": 998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615590987880482e-05,
+      "loss": 0.5537,
+      "step": 999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.614797664880107e-05,
+      "loss": 0.5275,
+      "step": 1000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6140037193413667e-05,
+      "loss": 0.5543,
+      "step": 1001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6132091520682042e-05,
+      "loss": 0.5375,
+      "step": 1002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6124139638651896e-05,
+      "loss": 0.5254,
+      "step": 1003
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6116181555375236e-05,
+      "loss": 0.5493,
+      "step": 1004
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108217278910344e-05,
+      "loss": 0.5338,
+      "step": 1005
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.610024681732177e-05,
+      "loss": 0.5043,
+      "step": 1006
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609227017868033e-05,
+      "loss": 0.5197,
+      "step": 1007
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608428737106309e-05,
+      "loss": 0.5227,
+      "step": 1008
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6076298402553372e-05,
+      "loss": 0.5438,
+      "step": 1009
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6068303281240727e-05,
+      "loss": 0.5634,
+      "step": 1010
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.606030201522094e-05,
+      "loss": 0.5526,
+      "step": 1011
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6052294612596022e-05,
+      "loss": 0.5625,
+      "step": 1012
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6044281081474188e-05,
+      "loss": 0.5173,
+      "step": 1013
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.603626142996987e-05,
+      "loss": 0.551,
+      "step": 1014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6028235666203687e-05,
+      "loss": 0.537,
+      "step": 1015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6020203798302455e-05,
+      "loss": 0.533,
+      "step": 1016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6012165834399173e-05,
+      "loss": 0.5129,
+      "step": 1017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6004121782632997e-05,
+      "loss": 0.5067,
+      "step": 1018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5996071651149268e-05,
+      "loss": 0.5452,
+      "step": 1019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5988015448099475e-05,
+      "loss": 0.5272,
+      "step": 1020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5979953181641246e-05,
+      "loss": 0.5482,
+      "step": 1021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.597188485993836e-05,
+      "loss": 0.5255,
+      "step": 1022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5963810491160727e-05,
+      "loss": 0.5325,
+      "step": 1023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.595573008348437e-05,
+      "loss": 0.5369,
+      "step": 1024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5947643645091443e-05,
+      "loss": 0.549,
+      "step": 1025
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5939551184170196e-05,
+      "loss": 0.549,
+      "step": 1026
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5931452708914972e-05,
+      "loss": 0.5359,
+      "step": 1027
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5923348227526218e-05,
+      "loss": 0.5338,
+      "step": 1028
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5915237748210458e-05,
+      "loss": 0.5276,
+      "step": 1029
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5907121279180277e-05,
+      "loss": 0.524,
+      "step": 1030
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5898998828654347e-05,
+      "loss": 0.535,
+      "step": 1031
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.589087040485738e-05,
+      "loss": 0.5439,
+      "step": 1032
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5882736016020142e-05,
+      "loss": 0.5485,
+      "step": 1033
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5874595670379444e-05,
+      "loss": 0.5615,
+      "step": 1034
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5866449376178118e-05,
+      "loss": 0.5638,
+      "step": 1035
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5858297141665026e-05,
+      "loss": 0.5328,
+      "step": 1036
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5850138975095045e-05,
+      "loss": 0.5442,
+      "step": 1037
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5841974884729058e-05,
+      "loss": 0.5291,
+      "step": 1038
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.583380487883395e-05,
+      "loss": 0.5223,
+      "step": 1039
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5825628965682588e-05,
+      "loss": 0.5214,
+      "step": 1040
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5817447153553826e-05,
+      "loss": 0.5649,
+      "step": 1041
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5809259450732495e-05,
+      "loss": 0.5198,
+      "step": 1042
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.580106586550938e-05,
+      "loss": 0.5432,
+      "step": 1043
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5792866406181232e-05,
+      "loss": 0.5198,
+      "step": 1044
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5784661081050744e-05,
+      "loss": 0.498,
+      "step": 1045
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5776449898426557e-05,
+      "loss": 0.5344,
+      "step": 1046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5768232866623234e-05,
+      "loss": 0.5341,
+      "step": 1047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5760009993961264e-05,
+      "loss": 0.5176,
+      "step": 1048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5751781288767052e-05,
+      "loss": 0.5449,
+      "step": 1049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5743546759372906e-05,
+      "loss": 0.5487,
+      "step": 1050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.573530641411704e-05,
+      "loss": 0.5078,
+      "step": 1051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727060261343543e-05,
+      "loss": 0.5466,
+      "step": 1052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5718808309402397e-05,
+      "loss": 0.5284,
+      "step": 1053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5710550566649444e-05,
+      "loss": 0.5413,
+      "step": 1054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5702287041446408e-05,
+      "loss": 0.5327,
+      "step": 1055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569401774216085e-05,
+      "loss": 0.5577,
+      "step": 1056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5685742677166188e-05,
+      "loss": 0.5316,
+      "step": 1057
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567746185484167e-05,
+      "loss": 0.5403,
+      "step": 1058
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5669175283572388e-05,
+      "loss": 0.5514,
+      "step": 1059
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5660882971749237e-05,
+      "loss": 0.5485,
+      "step": 1060
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5652584927768942e-05,
+      "loss": 0.551,
+      "step": 1061
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5644281160034024e-05,
+      "loss": 0.5652,
+      "step": 1062
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.56359716769528e-05,
+      "loss": 0.5317,
+      "step": 1063
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562765648693937e-05,
+      "loss": 0.5115,
+      "step": 1064
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561933559841362e-05,
+      "loss": 0.5592,
+      "step": 1065
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5611009019801206e-05,
+      "loss": 0.5479,
+      "step": 1066
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5602676759533542e-05,
+      "loss": 0.5228,
+      "step": 1067
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5594338826047802e-05,
+      "loss": 0.5562,
+      "step": 1068
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5585995227786895e-05,
+      "loss": 0.5082,
+      "step": 1069
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5577645973199468e-05,
+      "loss": 0.5399,
+      "step": 1070
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5569291070739898e-05,
+      "loss": 0.5271,
+      "step": 1071
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5560930528868284e-05,
+      "loss": 0.5203,
+      "step": 1072
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5552564356050437e-05,
+      "loss": 0.5492,
+      "step": 1073
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5544192560757855e-05,
+      "loss": 0.5385,
+      "step": 1074
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553581515146775e-05,
+      "loss": 0.4989,
+      "step": 1075
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5527432136663e-05,
+      "loss": 0.4912,
+      "step": 1076
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.551904352483217e-05,
+      "loss": 0.4863,
+      "step": 1077
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5510649324469493e-05,
+      "loss": 0.4749,
+      "step": 1078
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5502249544074855e-05,
+      "loss": 0.4939,
+      "step": 1079
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5493844192153796e-05,
+      "loss": 0.4901,
+      "step": 1080
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5485433277217495e-05,
+      "loss": 0.4747,
+      "step": 1081
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5477016807782767e-05,
+      "loss": 0.5067,
+      "step": 1082
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546859479237205e-05,
+      "loss": 0.4921,
+      "step": 1083
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460167239513397e-05,
+      "loss": 0.5178,
+      "step": 1084
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5451734157740475e-05,
+      "loss": 0.476,
+      "step": 1085
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5443295555592537e-05,
+      "loss": 0.4987,
+      "step": 1086
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5434851441614433e-05,
+      "loss": 0.4885,
+      "step": 1087
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5426401824356596e-05,
+      "loss": 0.4983,
+      "step": 1088
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.541794671237503e-05,
+      "loss": 0.4966,
+      "step": 1089
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.54094861142313e-05,
+      "loss": 0.4908,
+      "step": 1090
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540102003849253e-05,
+      "loss": 0.4855,
+      "step": 1091
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5392548493731392e-05,
+      "loss": 0.5128,
+      "step": 1092
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5384071488526088e-05,
+      "loss": 0.4764,
+      "step": 1093
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5375589031460354e-05,
+      "loss": 0.4854,
+      "step": 1094
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536710113112345e-05,
+      "loss": 0.4875,
+      "step": 1095
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5358607796110144e-05,
+      "loss": 0.4996,
+      "step": 1096
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5350109035020708e-05,
+      "loss": 0.4576,
+      "step": 1097
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5341604856460905e-05,
+      "loss": 0.4745,
+      "step": 1098
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533309526904199e-05,
+      "loss": 0.4885,
+      "step": 1099
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.532458028138069e-05,
+      "loss": 0.4658,
+      "step": 1100
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5316059902099198e-05,
+      "loss": 0.4835,
+      "step": 1101
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.530753413982518e-05,
+      "loss": 0.4869,
+      "step": 1102
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5299003003191734e-05,
+      "loss": 0.5178,
+      "step": 1103
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5290466500837414e-05,
+      "loss": 0.4684,
+      "step": 1104
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52819246414062e-05,
+      "loss": 0.4946,
+      "step": 1105
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.52733774335475e-05,
+      "loss": 0.5,
+      "step": 1106
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5264824885916136e-05,
+      "loss": 0.5107,
+      "step": 1107
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5256267007172343e-05,
+      "loss": 0.4965,
+      "step": 1108
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5247703805981742e-05,
+      "loss": 0.4869,
+      "step": 1109
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.523913529101535e-05,
+      "loss": 0.528,
+      "step": 1110
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5230561470949573e-05,
+      "loss": 0.4813,
+      "step": 1111
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5221982354466172e-05,
+      "loss": 0.5195,
+      "step": 1112
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5213397950252287e-05,
+      "loss": 0.515,
+      "step": 1113
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5204808267000402e-05,
+      "loss": 0.4998,
+      "step": 1114
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5196213313408348e-05,
+      "loss": 0.5034,
+      "step": 1115
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5187613098179296e-05,
+      "loss": 0.4907,
+      "step": 1116
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5179007630021747e-05,
+      "loss": 0.486,
+      "step": 1117
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5170396917649515e-05,
+      "loss": 0.4894,
+      "step": 1118
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5161780969781728e-05,
+      "loss": 0.4828,
+      "step": 1119
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5153159795142811e-05,
+      "loss": 0.4955,
+      "step": 1120
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5144533402462486e-05,
+      "loss": 0.4954,
+      "step": 1121
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5135901800475762e-05,
+      "loss": 0.4834,
+      "step": 1122
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5127264997922912e-05,
+      "loss": 0.484,
+      "step": 1123
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5118623003549489e-05,
+      "loss": 0.4946,
+      "step": 1124
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5109975826106286e-05,
+      "loss": 0.4756,
+      "step": 1125
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510132347434936e-05,
+      "loss": 0.4971,
+      "step": 1126
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5092665957040002e-05,
+      "loss": 0.4502,
+      "step": 1127
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.508400328294473e-05,
+      "loss": 0.5109,
+      "step": 1128
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.507533546083529e-05,
+      "loss": 0.4919,
+      "step": 1129
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5066662499488636e-05,
+      "loss": 0.4821,
+      "step": 1130
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5057984407686927e-05,
+      "loss": 0.4812,
+      "step": 1131
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5049301194217515e-05,
+      "loss": 0.5016,
+      "step": 1132
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5040612867872945e-05,
+      "loss": 0.4878,
+      "step": 1133
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5031919437450934e-05,
+      "loss": 0.53,
+      "step": 1134
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.502322091175437e-05,
+      "loss": 0.4896,
+      "step": 1135
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5014517299591296e-05,
+      "loss": 0.5035,
+      "step": 1136
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.500580860977491e-05,
+      "loss": 0.5063,
+      "step": 1137
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4997094851123548e-05,
+      "loss": 0.5081,
+      "step": 1138
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498837603246068e-05,
+      "loss": 0.4924,
+      "step": 1139
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4979652162614903e-05,
+      "loss": 0.4858,
+      "step": 1140
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4970923250419926e-05,
+      "loss": 0.5175,
+      "step": 1141
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.496218930471456e-05,
+      "loss": 0.486,
+      "step": 1142
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495345033434272e-05,
+      "loss": 0.4951,
+      "step": 1143
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944706348153405e-05,
+      "loss": 0.4769,
+      "step": 1144
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4935957355000695e-05,
+      "loss": 0.4663,
+      "step": 1145
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4927203363743737e-05,
+      "loss": 0.514,
+      "step": 1146
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4918444383246738e-05,
+      "loss": 0.4795,
+      "step": 1147
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4909680422378965e-05,
+      "loss": 0.5168,
+      "step": 1148
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490091149001472e-05,
+      "loss": 0.4934,
+      "step": 1149
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.489213759503334e-05,
+      "loss": 0.4729,
+      "step": 1150
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4883358746319192e-05,
+      "loss": 0.4799,
+      "step": 1151
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4874574952761655e-05,
+      "loss": 0.4891,
+      "step": 1152
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.486578622325512e-05,
+      "loss": 0.4899,
+      "step": 1153
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4856992566698967e-05,
+      "loss": 0.5159,
+      "step": 1154
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4848193991997573e-05,
+      "loss": 0.4841,
+      "step": 1155
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4839390508060287e-05,
+      "loss": 0.4901,
+      "step": 1156
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4830582123801445e-05,
+      "loss": 0.4941,
+      "step": 1157
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4821768848140326e-05,
+      "loss": 0.5134,
+      "step": 1158
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.481295069000117e-05,
+      "loss": 0.5217,
+      "step": 1159
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.480412765831317e-05,
+      "loss": 0.5262,
+      "step": 1160
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479529976201044e-05,
+      "loss": 0.493,
+      "step": 1161
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4786467010032022e-05,
+      "loss": 0.4669,
+      "step": 1162
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.477762941132188e-05,
+      "loss": 0.4716,
+      "step": 1163
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4768786974828887e-05,
+      "loss": 0.5084,
+      "step": 1164
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.475993970950681e-05,
+      "loss": 0.4551,
+      "step": 1165
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4751087624314307e-05,
+      "loss": 0.4762,
+      "step": 1166
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4742230728214912e-05,
+      "loss": 0.5079,
+      "step": 1167
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4733369030177043e-05,
+      "loss": 0.5098,
+      "step": 1168
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4724502539173969e-05,
+      "loss": 0.4786,
+      "step": 1169
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715631264183815e-05,
+      "loss": 0.468,
+      "step": 1170
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4706755214189551e-05,
+      "loss": 0.4817,
+      "step": 1171
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4697874398178983e-05,
+      "loss": 0.4998,
+      "step": 1172
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4688988825144743e-05,
+      "loss": 0.5199,
+      "step": 1173
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.468009850408428e-05,
+      "loss": 0.499,
+      "step": 1174
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4671203443999847e-05,
+      "loss": 0.4951,
+      "step": 1175
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.46623036538985e-05,
+      "loss": 0.4946,
+      "step": 1176
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4653399142792084e-05,
+      "loss": 0.4885,
+      "step": 1177
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464448991969722e-05,
+      "loss": 0.4926,
+      "step": 1178
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.463557599363531e-05,
+      "loss": 0.4807,
+      "step": 1179
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4626657373632507e-05,
+      "loss": 0.4775,
+      "step": 1180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4617734068719727e-05,
+      "loss": 0.5055,
+      "step": 1181
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.460880608793262e-05,
+      "loss": 0.5248,
+      "step": 1182
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4599873440311577e-05,
+      "loss": 0.4852,
+      "step": 1183
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4590936134901714e-05,
+      "loss": 0.4766,
+      "step": 1184
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581994180752866e-05,
+      "loss": 0.4877,
+      "step": 1185
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4573047586919566e-05,
+      "loss": 0.5067,
+      "step": 1186
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.456409636246106e-05,
+      "loss": 0.5024,
+      "step": 1187
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4555140516441263e-05,
+      "loss": 0.4968,
+      "step": 1188
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546180057928792e-05,
+      "loss": 0.4911,
+      "step": 1189
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4537214995996915e-05,
+      "loss": 0.5189,
+      "step": 1190
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4528245339723577e-05,
+      "loss": 0.4923,
+      "step": 1191
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4519271098191362e-05,
+      "loss": 0.4943,
+      "step": 1192
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510292280487504e-05,
+      "loss": 0.4763,
+      "step": 1193
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4501308895703872e-05,
+      "loss": 0.4723,
+      "step": 1194
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4492320952936956e-05,
+      "loss": 0.4939,
+      "step": 1195
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4483328461287862e-05,
+      "loss": 0.4896,
+      "step": 1196
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4474331429862307e-05,
+      "loss": 0.4927,
+      "step": 1197
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4465329867770597e-05,
+      "loss": 0.5048,
+      "step": 1198
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4456323784127629e-05,
+      "loss": 0.5256,
+      "step": 1199
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.444731318805288e-05,
+      "loss": 0.4857,
+      "step": 1200
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4438298088670397e-05,
+      "loss": 0.4749,
+      "step": 1201
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4429278495108785e-05,
+      "loss": 0.4739,
+      "step": 1202
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4420254416501198e-05,
+      "loss": 0.5007,
+      "step": 1203
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4411225861985339e-05,
+      "loss": 0.4904,
+      "step": 1204
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4402192840703431e-05,
+      "loss": 0.4689,
+      "step": 1205
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4393155361802229e-05,
+      "loss": 0.4978,
+      "step": 1206
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4384113434433001e-05,
+      "loss": 0.5035,
+      "step": 1207
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4375067067751518e-05,
+      "loss": 0.5033,
+      "step": 1208
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4366016270918043e-05,
+      "loss": 0.4489,
+      "step": 1209
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4356961053097334e-05,
+      "loss": 0.4647,
+      "step": 1210
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4347901423458614e-05,
+      "loss": 0.4873,
+      "step": 1211
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4338837391175582e-05,
+      "loss": 0.4678,
+      "step": 1212
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4329768965426393e-05,
+      "loss": 0.4957,
+      "step": 1213
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4320696155393651e-05,
+      "loss": 0.4935,
+      "step": 1214
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4311618970264394e-05,
+      "loss": 0.4962,
+      "step": 1215
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4302537419230098e-05,
+      "loss": 0.5225,
+      "step": 1216
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4293451511486658e-05,
+      "loss": 0.4867,
+      "step": 1217
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4284361256234377e-05,
+      "loss": 0.4963,
+      "step": 1218
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.427526666267796e-05,
+      "loss": 0.4858,
+      "step": 1219
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4266167740026515e-05,
+      "loss": 0.495,
+      "step": 1220
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4257064497493518e-05,
+      "loss": 0.4921,
+      "step": 1221
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.424795694429683e-05,
+      "loss": 0.4813,
+      "step": 1222
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4238845089658678e-05,
+      "loss": 0.49,
+      "step": 1223
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4229728942805636e-05,
+      "loss": 0.5221,
+      "step": 1224
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4220608512968627e-05,
+      "loss": 0.5059,
+      "step": 1225
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.421148380938292e-05,
+      "loss": 0.5135,
+      "step": 1226
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4202354841288101e-05,
+      "loss": 0.5094,
+      "step": 1227
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4193221617928076e-05,
+      "loss": 0.5106,
+      "step": 1228
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4184084148551063e-05,
+      "loss": 0.4895,
+      "step": 1229
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174942442409583e-05,
+      "loss": 0.4978,
+      "step": 1230
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.416579650876043e-05,
+      "loss": 0.5242,
+      "step": 1231
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4156646356864703e-05,
+      "loss": 0.4933,
+      "step": 1232
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4147491995987752e-05,
+      "loss": 0.5177,
+      "step": 1233
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41383334353992e-05,
+      "loss": 0.5074,
+      "step": 1234
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4129170684372924e-05,
+      "loss": 0.4917,
+      "step": 1235
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4120003752187035e-05,
+      "loss": 0.4647,
+      "step": 1236
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4110832648123885e-05,
+      "loss": 0.4911,
+      "step": 1237
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4101657381470046e-05,
+      "loss": 0.481,
+      "step": 1238
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4092477961516306e-05,
+      "loss": 0.4836,
+      "step": 1239
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4083294397557668e-05,
+      "loss": 0.489,
+      "step": 1240
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4074106698893313e-05,
+      "loss": 0.4973,
+      "step": 1241
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4064914874826626e-05,
+      "loss": 0.4813,
+      "step": 1242
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4055718934665157e-05,
+      "loss": 0.4723,
+      "step": 1243
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.404651888772063e-05,
+      "loss": 0.4998,
+      "step": 1244
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.403731474330893e-05,
+      "loss": 0.4856,
+      "step": 1245
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4028106510750083e-05,
+      "loss": 0.4935,
+      "step": 1246
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4018894199368265e-05,
+      "loss": 0.4989,
+      "step": 1247
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4009677818491767e-05,
+      "loss": 0.5096,
+      "step": 1248
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000457377453018e-05,
+      "loss": 0.4878,
+      "step": 1249
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3991232885588549e-05,
+      "loss": 0.467,
+      "step": 1250
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3982004352238992e-05,
+      "loss": 0.4893,
+      "step": 1251
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3972771786749074e-05,
+      "loss": 0.4618,
+      "step": 1252
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3963535198467609e-05,
+      "loss": 0.4934,
+      "step": 1253
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.395429459674747e-05,
+      "loss": 0.5043,
+      "step": 1254
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3945049990945614e-05,
+      "loss": 0.4924,
+      "step": 1255
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3935801390423036e-05,
+      "loss": 0.4742,
+      "step": 1256
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3926548804544788e-05,
+      "loss": 0.478,
+      "step": 1257
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3917292242679947e-05,
+      "loss": 0.4906,
+      "step": 1258
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3908031714201621e-05,
+      "loss": 0.5029,
+      "step": 1259
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898767228486937e-05,
+      "loss": 0.4541,
+      "step": 1260
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3889498794917027e-05,
+      "loss": 0.4891,
+      "step": 1261
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3880226422877017e-05,
+      "loss": 0.4978,
+      "step": 1262
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3870950121756029e-05,
+      "loss": 0.4908,
+      "step": 1263
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.386166990094715e-05,
+      "loss": 0.4935,
+      "step": 1264
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3852385769847453e-05,
+      "loss": 0.4905,
+      "step": 1265
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384309773785796e-05,
+      "loss": 0.4812,
+      "step": 1266
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3833805814383645e-05,
+      "loss": 0.5062,
+      "step": 1267
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3824510008833427e-05,
+      "loss": 0.5044,
+      "step": 1268
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3815210330620143e-05,
+      "loss": 0.5103,
+      "step": 1269
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3805906789160565e-05,
+      "loss": 0.5051,
+      "step": 1270
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3796599393875374e-05,
+      "loss": 0.4669,
+      "step": 1271
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3787288154189146e-05,
+      "loss": 0.4707,
+      "step": 1272
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3777973079530362e-05,
+      "loss": 0.4818,
+      "step": 1273
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3768654179331376e-05,
+      "loss": 0.4866,
+      "step": 1274
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3759331463028416e-05,
+      "loss": 0.476,
+      "step": 1275
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3750004940061581e-05,
+      "loss": 0.4725,
+      "step": 1276
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3740674619874822e-05,
+      "loss": 0.5076,
+      "step": 1277
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3731340511915932e-05,
+      "loss": 0.4778,
+      "step": 1278
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3722002625636542e-05,
+      "loss": 0.5114,
+      "step": 1279
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3712660970492108e-05,
+      "loss": 0.4888,
+      "step": 1280
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3703315555941904e-05,
+      "loss": 0.4927,
+      "step": 1281
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3693966391449006e-05,
+      "loss": 0.4753,
+      "step": 1282
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3684613486480292e-05,
+      "loss": 0.5122,
+      "step": 1283
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3675256850506428e-05,
+      "loss": 0.5038,
+      "step": 1284
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3665896493001854e-05,
+      "loss": 0.4924,
+      "step": 1285
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3656532423444779e-05,
+      "loss": 0.4735,
+      "step": 1286
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3647164651317178e-05,
+      "loss": 0.4985,
+      "step": 1287
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3637793186104764e-05,
+      "loss": 0.4851,
+      "step": 1288
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3628418037296998e-05,
+      "loss": 0.4928,
+      "step": 1289
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3619039214387066e-05,
+      "loss": 0.4833,
+      "step": 1290
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3609656726871878e-05,
+      "loss": 0.512,
+      "step": 1291
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3600270584252052e-05,
+      "loss": 0.489,
+      "step": 1292
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3590880796031909e-05,
+      "loss": 0.4731,
+      "step": 1293
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3581487371719457e-05,
+      "loss": 0.5111,
+      "step": 1294
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3572090320826396e-05,
+      "loss": 0.4879,
+      "step": 1295
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356268965286809e-05,
+      "loss": 0.5239,
+      "step": 1296
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3553285377363563e-05,
+      "loss": 0.4916,
+      "step": 1297
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35438775038355e-05,
+      "loss": 0.4872,
+      "step": 1298
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3534466041810227e-05,
+      "loss": 0.4785,
+      "step": 1299
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35250510008177e-05,
+      "loss": 0.5006,
+      "step": 1300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.35156323903915e-05,
+      "loss": 0.4897,
+      "step": 1301
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3506210220068828e-05,
+      "loss": 0.4927,
+      "step": 1302
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.349678449939048e-05,
+      "loss": 0.4778,
+      "step": 1303
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.348735523790086e-05,
+      "loss": 0.4869,
+      "step": 1304
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3477922445147944e-05,
+      "loss": 0.4662,
+      "step": 1305
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3468486130683294e-05,
+      "loss": 0.5021,
+      "step": 1306
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.345904630406203e-05,
+      "loss": 0.4623,
+      "step": 1307
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3449602974842832e-05,
+      "loss": 0.4789,
+      "step": 1308
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.344015615258793e-05,
+      "loss": 0.491,
+      "step": 1309
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3430705846863089e-05,
+      "loss": 0.4835,
+      "step": 1310
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3421252067237594e-05,
+      "loss": 0.5095,
+      "step": 1311
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3411794823284259e-05,
+      "loss": 0.4828,
+      "step": 1312
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402334124579399e-05,
+      "loss": 0.4797,
+      "step": 1313
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3392869980702829e-05,
+      "loss": 0.499,
+      "step": 1314
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.338340240123785e-05,
+      "loss": 0.5011,
+      "step": 1315
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3373931395771251e-05,
+      "loss": 0.5015,
+      "step": 1316
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3364456973893279e-05,
+      "loss": 0.485,
+      "step": 1317
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3354979145197643e-05,
+      "loss": 0.5134,
+      "step": 1318
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3345497919281506e-05,
+      "loss": 0.4833,
+      "step": 1319
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3336013305745471e-05,
+      "loss": 0.4957,
+      "step": 1320
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3326525314193569e-05,
+      "loss": 0.4855,
+      "step": 1321
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3317033954233246e-05,
+      "loss": 0.4852,
+      "step": 1322
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3307539235475369e-05,
+      "loss": 0.4846,
+      "step": 1323
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3298041167534202e-05,
+      "loss": 0.533,
+      "step": 1324
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3288539760027393e-05,
+      "loss": 0.4965,
+      "step": 1325
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3279035022575988e-05,
+      "loss": 0.4918,
+      "step": 1326
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.326952696480439e-05,
+      "loss": 0.5131,
+      "step": 1327
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260015596340368e-05,
+      "loss": 0.4978,
+      "step": 1328
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3250500926815046e-05,
+      "loss": 0.4924,
+      "step": 1329
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3240982965862893e-05,
+      "loss": 0.4711,
+      "step": 1330
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3231461723121699e-05,
+      "loss": 0.4534,
+      "step": 1331
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3221937208232593e-05,
+      "loss": 0.4973,
+      "step": 1332
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3212409430840012e-05,
+      "loss": 0.5017,
+      "step": 1333
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3202878400591683e-05,
+      "loss": 0.4895,
+      "step": 1334
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3193344127138647e-05,
+      "loss": 0.4862,
+      "step": 1335
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3183806620135218e-05,
+      "loss": 0.5126,
+      "step": 1336
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3174265889238984e-05,
+      "loss": 0.4791,
+      "step": 1337
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3164721944110802e-05,
+      "loss": 0.4779,
+      "step": 1338
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155174794414787e-05,
+      "loss": 0.4758,
+      "step": 1339
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3145624449818284e-05,
+      "loss": 0.4902,
+      "step": 1340
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3136070919991885e-05,
+      "loss": 0.494,
+      "step": 1341
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3126514214609408e-05,
+      "loss": 0.4714,
+      "step": 1342
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3116954343347882e-05,
+      "loss": 0.5135,
+      "step": 1343
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3107391315887544e-05,
+      "loss": 0.4691,
+      "step": 1344
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3097825141911824e-05,
+      "loss": 0.5005,
+      "step": 1345
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3088255831107344e-05,
+      "loss": 0.4875,
+      "step": 1346
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3078683393163894e-05,
+      "loss": 0.4838,
+      "step": 1347
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3069107837774437e-05,
+      "loss": 0.4904,
+      "step": 1348
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3059529174635095e-05,
+      "loss": 0.5078,
+      "step": 1349
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049947413445125e-05,
+      "loss": 0.4717,
+      "step": 1350
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3040362563906932e-05,
+      "loss": 0.4794,
+      "step": 1351
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.303077463572605e-05,
+      "loss": 0.47,
+      "step": 1352
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3021183638611117e-05,
+      "loss": 0.4835,
+      "step": 1353
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3011589582273891e-05,
+      "loss": 0.5216,
+      "step": 1354
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3001992476429224e-05,
+      "loss": 0.4905,
+      "step": 1355
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2992392330795058e-05,
+      "loss": 0.5037,
+      "step": 1356
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2982789155092407e-05,
+      "loss": 0.491,
+      "step": 1357
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973182959045356e-05,
+      "loss": 0.485,
+      "step": 1358
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2963573752381056e-05,
+      "loss": 0.46,
+      "step": 1359
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.295396154482969e-05,
+      "loss": 0.494,
+      "step": 1360
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2944346346124494e-05,
+      "loss": 0.4691,
+      "step": 1361
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.293472816600173e-05,
+      "loss": 0.47,
+      "step": 1362
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.292510701420068e-05,
+      "loss": 0.4835,
+      "step": 1363
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2915482900463622e-05,
+      "loss": 0.4851,
+      "step": 1364
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905855834535852e-05,
+      "loss": 0.4626,
+      "step": 1365
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2896225826165647e-05,
+      "loss": 0.4751,
+      "step": 1366
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2886592885104259e-05,
+      "loss": 0.5051,
+      "step": 1367
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2876957021105916e-05,
+      "loss": 0.4995,
+      "step": 1368
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2867318243927802e-05,
+      "loss": 0.5056,
+      "step": 1369
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2857676563330058e-05,
+      "loss": 0.4857,
+      "step": 1370
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2848031989075754e-05,
+      "loss": 0.4828,
+      "step": 1371
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2838384530930899e-05,
+      "loss": 0.4839,
+      "step": 1372
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2828734198664416e-05,
+      "loss": 0.5013,
+      "step": 1373
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2819081002048143e-05,
+      "loss": 0.4854,
+      "step": 1374
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2809424950856815e-05,
+      "loss": 0.4952,
+      "step": 1375
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2799766054868066e-05,
+      "loss": 0.4984,
+      "step": 1376
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790104323862394e-05,
+      "loss": 0.4789,
+      "step": 1377
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.278043976762318e-05,
+      "loss": 0.5129,
+      "step": 1378
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.277077239593667e-05,
+      "loss": 0.4839,
+      "step": 1379
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2761102218591945e-05,
+      "loss": 0.5,
+      "step": 1380
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275142924538094e-05,
+      "loss": 0.4877,
+      "step": 1381
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.274175348609842e-05,
+      "loss": 0.4735,
+      "step": 1382
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2732074950541966e-05,
+      "loss": 0.4928,
+      "step": 1383
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2722393648511968e-05,
+      "loss": 0.4823,
+      "step": 1384
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2712709589811629e-05,
+      "loss": 0.4757,
+      "step": 1385
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2703022784246932e-05,
+      "loss": 0.4808,
+      "step": 1386
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.269333324162665e-05,
+      "loss": 0.5009,
+      "step": 1387
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2683640971762316e-05,
+      "loss": 0.4938,
+      "step": 1388
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2673945984468241e-05,
+      "loss": 0.4921,
+      "step": 1389
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2664248289561472e-05,
+      "loss": 0.4828,
+      "step": 1390
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2654547896861804e-05,
+      "loss": 0.498,
+      "step": 1391
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.264484481619177e-05,
+      "loss": 0.4797,
+      "step": 1392
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2635139057376611e-05,
+      "loss": 0.4833,
+      "step": 1393
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2625430630244288e-05,
+      "loss": 0.4727,
+      "step": 1394
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.261571954462547e-05,
+      "loss": 0.5095,
+      "step": 1395
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.260600581035351e-05,
+      "loss": 0.5083,
+      "step": 1396
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.259628943726444e-05,
+      "loss": 0.4708,
+      "step": 1397
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2586570435196972e-05,
+      "loss": 0.4983,
+      "step": 1398
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2576848813992475e-05,
+      "loss": 0.5004,
+      "step": 1399
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2567124583494972e-05,
+      "loss": 0.476,
+      "step": 1400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2557397753551128e-05,
+      "loss": 0.4565,
+      "step": 1401
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2547668334010239e-05,
+      "loss": 0.4856,
+      "step": 1402
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2537936334724225e-05,
+      "loss": 0.4966,
+      "step": 1403
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2528201765547618e-05,
+      "loss": 0.495,
+      "step": 1404
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2518464636337556e-05,
+      "loss": 0.4706,
+      "step": 1405
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2508724956953755e-05,
+      "loss": 0.4923,
+      "step": 1406
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2498982737258532e-05,
+      "loss": 0.4838,
+      "step": 1407
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.248923798711676e-05,
+      "loss": 0.4863,
+      "step": 1408
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.247949071639589e-05,
+      "loss": 0.5002,
+      "step": 1409
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2469740934965912e-05,
+      "loss": 0.4768,
+      "step": 1410
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2459988652699364e-05,
+      "loss": 0.4869,
+      "step": 1411
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.245023387947132e-05,
+      "loss": 0.5057,
+      "step": 1412
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2440476625159363e-05,
+      "loss": 0.4871,
+      "step": 1413
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2430716899643605e-05,
+      "loss": 0.4705,
+      "step": 1414
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2420954712806654e-05,
+      "loss": 0.4816,
+      "step": 1415
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.24111900745336e-05,
+      "loss": 0.4895,
+      "step": 1416
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2401422994712035e-05,
+      "loss": 0.4785,
+      "step": 1417
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2391653483232007e-05,
+      "loss": 0.4963,
+      "step": 1418
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.238188154998603e-05,
+      "loss": 0.4976,
+      "step": 1419
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2372107204869078e-05,
+      "loss": 0.4712,
+      "step": 1420
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2362330457778557e-05,
+      "loss": 0.474,
+      "step": 1421
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2352551318614315e-05,
+      "loss": 0.506,
+      "step": 1422
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2342769797278608e-05,
+      "loss": 0.4762,
+      "step": 1423
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233298590367612e-05,
+      "loss": 0.5038,
+      "step": 1424
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2323199647713928e-05,
+      "loss": 0.5111,
+      "step": 1425
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2313411039301504e-05,
+      "loss": 0.5217,
+      "step": 1426
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.23036200883507e-05,
+      "loss": 0.48,
+      "step": 1427
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2293826804775741e-05,
+      "loss": 0.49,
+      "step": 1428
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2284031198493217e-05,
+      "loss": 0.474,
+      "step": 1429
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2274233279422068e-05,
+      "loss": 0.4968,
+      "step": 1430
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2264433057483568e-05,
+      "loss": 0.4755,
+      "step": 1431
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.225463054260134e-05,
+      "loss": 0.4629,
+      "step": 1432
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.224482574470131e-05,
+      "loss": 0.4791,
+      "step": 1433
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2235018673711729e-05,
+      "loss": 0.4832,
+      "step": 1434
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.4814,
+      "step": 1435
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2215397752188398e-05,
+      "loss": 0.4763,
+      "step": 1436
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2205583921522611e-05,
+      "loss": 0.4855,
+      "step": 1437
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2195767857503173e-05,
+      "loss": 0.499,
+      "step": 1438
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2185949570069744e-05,
+      "loss": 0.4911,
+      "step": 1439
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2176129069164226e-05,
+      "loss": 0.4996,
+      "step": 1440
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2166306364730766e-05,
+      "loss": 0.5021,
+      "step": 1441
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2156481466715748e-05,
+      "loss": 0.4749,
+      "step": 1442
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2146654385067764e-05,
+      "loss": 0.4972,
+      "step": 1443
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.213682512973763e-05,
+      "loss": 0.4551,
+      "step": 1444
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2126993710678361e-05,
+      "loss": 0.4956,
+      "step": 1445
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2117160137845151e-05,
+      "loss": 0.5038,
+      "step": 1446
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210732442119539e-05,
+      "loss": 0.47,
+      "step": 1447
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097486570688635e-05,
+      "loss": 0.4996,
+      "step": 1448
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2087646596286594e-05,
+      "loss": 0.5153,
+      "step": 1449
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2077804507953137e-05,
+      "loss": 0.4944,
+      "step": 1450
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2067960315654272e-05,
+      "loss": 0.4885,
+      "step": 1451
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058114029358134e-05,
+      "loss": 0.5168,
+      "step": 1452
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2048265659034981e-05,
+      "loss": 0.507,
+      "step": 1453
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2038415214657179e-05,
+      "loss": 0.485,
+      "step": 1454
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2028562706199201e-05,
+      "loss": 0.5116,
+      "step": 1455
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2018708143637598e-05,
+      "loss": 0.472,
+      "step": 1456
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2008851536951015e-05,
+      "loss": 0.4652,
+      "step": 1457
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1998992896120154e-05,
+      "loss": 0.5034,
+      "step": 1458
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1989132231127784e-05,
+      "loss": 0.4908,
+      "step": 1459
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1979269551958723e-05,
+      "loss": 0.4746,
+      "step": 1460
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1969404868599828e-05,
+      "loss": 0.5114,
+      "step": 1461
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1959538191039986e-05,
+      "loss": 0.4832,
+      "step": 1462
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1949669529270098e-05,
+      "loss": 0.4891,
+      "step": 1463
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.193979889328308e-05,
+      "loss": 0.4984,
+      "step": 1464
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1929926293073852e-05,
+      "loss": 0.495,
+      "step": 1465
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1920051738639309e-05,
+      "loss": 0.4854,
+      "step": 1466
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1910175239978334e-05,
+      "loss": 0.52,
+      "step": 1467
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190029680709178e-05,
+      "loss": 0.4906,
+      "step": 1468
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1890416449982451e-05,
+      "loss": 0.4823,
+      "step": 1469
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1880534178655112e-05,
+      "loss": 0.4786,
+      "step": 1470
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187065000311645e-05,
+      "loss": 0.5122,
+      "step": 1471
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1860763933375097e-05,
+      "loss": 0.467,
+      "step": 1472
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1850875979441585e-05,
+      "loss": 0.5143,
+      "step": 1473
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1840986151328371e-05,
+      "loss": 0.4989,
+      "step": 1474
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1831094459049803e-05,
+      "loss": 0.4918,
+      "step": 1475
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1821200912622111e-05,
+      "loss": 0.4979,
+      "step": 1476
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1811305522063409e-05,
+      "loss": 0.4553,
+      "step": 1477
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801408297393682e-05,
+      "loss": 0.4912,
+      "step": 1478
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1791509248634758e-05,
+      "loss": 0.4713,
+      "step": 1479
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1781608385810329e-05,
+      "loss": 0.5017,
+      "step": 1480
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1771705718945912e-05,
+      "loss": 0.4841,
+      "step": 1481
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1761801258068856e-05,
+      "loss": 0.4926,
+      "step": 1482
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1751895013208325e-05,
+      "loss": 0.5109,
+      "step": 1483
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1741986994395291e-05,
+      "loss": 0.4686,
+      "step": 1484
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1732077211662523e-05,
+      "loss": 0.5092,
+      "step": 1485
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1722165675044568e-05,
+      "loss": 0.4893,
+      "step": 1486
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.171225239457776e-05,
+      "loss": 0.4739,
+      "step": 1487
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1702337380300196e-05,
+      "loss": 0.4917,
+      "step": 1488
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692420642251726e-05,
+      "loss": 0.4932,
+      "step": 1489
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.168250219047394e-05,
+      "loss": 0.5057,
+      "step": 1490
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1672582035010174e-05,
+      "loss": 0.4896,
+      "step": 1491
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1662660185905485e-05,
+      "loss": 0.4792,
+      "step": 1492
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1652736653206641e-05,
+      "loss": 0.4907,
+      "step": 1493
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1642811446962125e-05,
+      "loss": 0.5081,
+      "step": 1494
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1632884577222106e-05,
+      "loss": 0.4705,
+      "step": 1495
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1622956054038432e-05,
+      "loss": 0.4845,
+      "step": 1496
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1613025887464642e-05,
+      "loss": 0.4694,
+      "step": 1497
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1603094087555928e-05,
+      "loss": 0.4822,
+      "step": 1498
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1593160664369132e-05,
+      "loss": 0.4781,
+      "step": 1499
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.158322562796275e-05,
+      "loss": 0.4982,
+      "step": 1500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1573288988396907e-05,
+      "loss": 0.4714,
+      "step": 1501
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1563350755733351e-05,
+      "loss": 0.4707,
+      "step": 1502
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1553410940035441e-05,
+      "loss": 0.5024,
+      "step": 1503
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1543469551368146e-05,
+      "loss": 0.4916,
+      "step": 1504
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.153352659979802e-05,
+      "loss": 0.4703,
+      "step": 1505
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1523582095393201e-05,
+      "loss": 0.4863,
+      "step": 1506
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1513636048223406e-05,
+      "loss": 0.4869,
+      "step": 1507
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1503688468359907e-05,
+      "loss": 0.4979,
+      "step": 1508
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1493739365875529e-05,
+      "loss": 0.5048,
+      "step": 1509
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1483788750844642e-05,
+      "loss": 0.5035,
+      "step": 1510
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1473836633343145e-05,
+      "loss": 0.4939,
+      "step": 1511
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1463883023448456e-05,
+      "loss": 0.4789,
+      "step": 1512
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1453927931239514e-05,
+      "loss": 0.4843,
+      "step": 1513
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1443971366796747e-05,
+      "loss": 0.4716,
+      "step": 1514
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.143401334020208e-05,
+      "loss": 0.4998,
+      "step": 1515
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.142405386153892e-05,
+      "loss": 0.5007,
+      "step": 1516
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1414092940892139e-05,
+      "loss": 0.4948,
+      "step": 1517
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1404130588348072e-05,
+      "loss": 0.5031,
+      "step": 1518
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1394166813994506e-05,
+      "loss": 0.5188,
+      "step": 1519
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1384201627920663e-05,
+      "loss": 0.4929,
+      "step": 1520
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1374235040217203e-05,
+      "loss": 0.4816,
+      "step": 1521
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1364267060976193e-05,
+      "loss": 0.4603,
+      "step": 1522
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1354297700291119e-05,
+      "loss": 0.4966,
+      "step": 1523
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.134432696825686e-05,
+      "loss": 0.4963,
+      "step": 1524
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.133435487496969e-05,
+      "loss": 0.4874,
+      "step": 1525
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1324381430527254e-05,
+      "loss": 0.4847,
+      "step": 1526
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1314406645028569e-05,
+      "loss": 0.5016,
+      "step": 1527
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1304430528574011e-05,
+      "loss": 0.4818,
+      "step": 1528
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1294453091265301e-05,
+      "loss": 0.4952,
+      "step": 1529
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1284474343205496e-05,
+      "loss": 0.5011,
+      "step": 1530
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.127449429449899e-05,
+      "loss": 0.483,
+      "step": 1531
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1264512955251479e-05,
+      "loss": 0.4998,
+      "step": 1532
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.125453033556998e-05,
+      "loss": 0.4973,
+      "step": 1533
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244546445562794e-05,
+      "loss": 0.4867,
+      "step": 1534
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.123456129533952e-05,
+      "loss": 0.4885,
+      "step": 1535
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.122457489501103e-05,
+      "loss": 0.4813,
+      "step": 1536
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.121458725468945e-05,
+      "loss": 0.4854,
+      "step": 1537
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204598384488181e-05,
+      "loss": 0.4735,
+      "step": 1538
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1194608294521853e-05,
+      "loss": 0.4694,
+      "step": 1539
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1184616994906343e-05,
+      "loss": 0.5106,
+      "step": 1540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1174624495758743e-05,
+      "loss": 0.4802,
+      "step": 1541
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1164630807197366e-05,
+      "loss": 0.493,
+      "step": 1542
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1154635939341728e-05,
+      "loss": 0.4976,
+      "step": 1543
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1144639902312537e-05,
+      "loss": 0.4624,
+      "step": 1544
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1134642706231687e-05,
+      "loss": 0.4915,
+      "step": 1545
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1124644361222246e-05,
+      "loss": 0.4904,
+      "step": 1546
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1114644877408446e-05,
+      "loss": 0.5291,
+      "step": 1547
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1104644264915667e-05,
+      "loss": 0.4673,
+      "step": 1548
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094642533870439e-05,
+      "loss": 0.476,
+      "step": 1549
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1084639694400421e-05,
+      "loss": 0.5038,
+      "step": 1550
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.107463575663439e-05,
+      "loss": 0.4945,
+      "step": 1551
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1064630730702248e-05,
+      "loss": 0.4857,
+      "step": 1552
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1054624626734985e-05,
+      "loss": 0.4987,
+      "step": 1553
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1044617454864693e-05,
+      "loss": 0.4945,
+      "step": 1554
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1034609225224533e-05,
+      "loss": 0.4873,
+      "step": 1555
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1024599947948749e-05,
+      "loss": 0.5074,
+      "step": 1556
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1014589633172643e-05,
+      "loss": 0.4764,
+      "step": 1557
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1004578291032566e-05,
+      "loss": 0.494,
+      "step": 1558
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0994565931665907e-05,
+      "loss": 0.4706,
+      "step": 1559
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0984552565211088e-05,
+      "loss": 0.4891,
+      "step": 1560
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0974538201807552e-05,
+      "loss": 0.499,
+      "step": 1561
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0964522851595748e-05,
+      "loss": 0.4938,
+      "step": 1562
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0954506524717125e-05,
+      "loss": 0.4904,
+      "step": 1563
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944489231314127e-05,
+      "loss": 0.4677,
+      "step": 1564
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0934470981530163e-05,
+      "loss": 0.4813,
+      "step": 1565
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0924451785509627e-05,
+      "loss": 0.4652,
+      "step": 1566
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0914431653397856e-05,
+      "loss": 0.4718,
+      "step": 1567
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090441059534115e-05,
+      "loss": 0.4885,
+      "step": 1568
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0894388621486733e-05,
+      "loss": 0.4707,
+      "step": 1569
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0884365741982765e-05,
+      "loss": 0.4889,
+      "step": 1570
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.087434196697832e-05,
+      "loss": 0.5131,
+      "step": 1571
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0864317306623377e-05,
+      "loss": 0.4939,
+      "step": 1572
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0854291771068812e-05,
+      "loss": 0.4845,
+      "step": 1573
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0844265370466393e-05,
+      "loss": 0.5026,
+      "step": 1574
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0834238114968755e-05,
+      "loss": 0.5044,
+      "step": 1575
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0824210014729406e-05,
+      "loss": 0.5049,
+      "step": 1576
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0814181079902709e-05,
+      "loss": 0.5193,
+      "step": 1577
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0804151320643861e-05,
+      "loss": 0.472,
+      "step": 1578
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0794120747108908e-05,
+      "loss": 0.4788,
+      "step": 1579
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0784089369454715e-05,
+      "loss": 0.4783,
+      "step": 1580
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0774057197838963e-05,
+      "loss": 0.4897,
+      "step": 1581
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.076402424242013e-05,
+      "loss": 0.4738,
+      "step": 1582
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0753990513357498e-05,
+      "loss": 0.521,
+      "step": 1583
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0743956020811122e-05,
+      "loss": 0.4788,
+      "step": 1584
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073392077494184e-05,
+      "loss": 0.4884,
+      "step": 1585
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0723884785911245e-05,
+      "loss": 0.4658,
+      "step": 1586
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0713848063881689e-05,
+      "loss": 0.4669,
+      "step": 1587
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070381061901626e-05,
+      "loss": 0.4835,
+      "step": 1588
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.069377246147878e-05,
+      "loss": 0.4872,
+      "step": 1589
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0683733601433794e-05,
+      "loss": 0.4857,
+      "step": 1590
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0673694049046561e-05,
+      "loss": 0.4817,
+      "step": 1591
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0663653814483035e-05,
+      "loss": 0.4813,
+      "step": 1592
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0653612907909864e-05,
+      "loss": 0.4639,
+      "step": 1593
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0643571339494378e-05,
+      "loss": 0.4532,
+      "step": 1594
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0633529119404571e-05,
+      "loss": 0.5048,
+      "step": 1595
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0623486257809103e-05,
+      "loss": 0.5009,
+      "step": 1596
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0613442764877286e-05,
+      "loss": 0.4929,
+      "step": 1597
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.060339865077906e-05,
+      "loss": 0.4883,
+      "step": 1598
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0593353925685005e-05,
+      "loss": 0.4996,
+      "step": 1599
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0583308599766316e-05,
+      "loss": 0.4658,
+      "step": 1600
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0573262683194793e-05,
+      "loss": 0.4597,
+      "step": 1601
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.056321618614284e-05,
+      "loss": 0.4752,
+      "step": 1602
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0553169118783445e-05,
+      "loss": 0.4895,
+      "step": 1603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0543121491290177e-05,
+      "loss": 0.4813,
+      "step": 1604
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0533073313837163e-05,
+      "loss": 0.4677,
+      "step": 1605
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0523024596599101e-05,
+      "loss": 0.4917,
+      "step": 1606
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0512975349751224e-05,
+      "loss": 0.4966,
+      "step": 1607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0502925583469305e-05,
+      "loss": 0.4956,
+      "step": 1608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0492875307929643e-05,
+      "loss": 0.4702,
+      "step": 1609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0482824533309057e-05,
+      "loss": 0.4916,
+      "step": 1610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0472773269784866e-05,
+      "loss": 0.4861,
+      "step": 1611
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0462721527534882e-05,
+      "loss": 0.491,
+      "step": 1612
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452669316737407e-05,
+      "loss": 0.4619,
+      "step": 1613
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044261664757122e-05,
+      "loss": 0.4917,
+      "step": 1614
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0432563530215552e-05,
+      "loss": 0.478,
+      "step": 1615
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0422509974850099e-05,
+      "loss": 0.4664,
+      "step": 1616
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0412455991654998e-05,
+      "loss": 0.4883,
+      "step": 1617
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0402401590810817e-05,
+      "loss": 0.4885,
+      "step": 1618
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.039234678249855e-05,
+      "loss": 0.494,
+      "step": 1619
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0382291576899606e-05,
+      "loss": 0.4774,
+      "step": 1620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.037223598419578e-05,
+      "loss": 0.5057,
+      "step": 1621
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0362180014569281e-05,
+      "loss": 0.4844,
+      "step": 1622
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0352123678202686e-05,
+      "loss": 0.5067,
+      "step": 1623
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.034206698527895e-05,
+      "loss": 0.5426,
+      "step": 1624
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0332009945981385e-05,
+      "loss": 0.4961,
+      "step": 1625
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0321952570493652e-05,
+      "loss": 0.4508,
+      "step": 1626
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031189486899976e-05,
+      "loss": 0.4988,
+      "step": 1627
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0301836851684038e-05,
+      "loss": 0.4674,
+      "step": 1628
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0291778528731144e-05,
+      "loss": 0.5122,
+      "step": 1629
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0281719910326043e-05,
+      "loss": 0.4819,
+      "step": 1630
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0271661006653995e-05,
+      "loss": 0.4995,
+      "step": 1631
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0261601827900551e-05,
+      "loss": 0.4944,
+      "step": 1632
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0251542384251545e-05,
+      "loss": 0.4761,
+      "step": 1633
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0241482685893072e-05,
+      "loss": 0.4806,
+      "step": 1634
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023142274301149e-05,
+      "loss": 0.4618,
+      "step": 1635
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0221362565793402e-05,
+      "loss": 0.5021,
+      "step": 1636
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0211302164425657e-05,
+      "loss": 0.4764,
+      "step": 1637
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.020124154909531e-05,
+      "loss": 0.4883,
+      "step": 1638
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0191180729989654e-05,
+      "loss": 0.4753,
+      "step": 1639
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0181119717296182e-05,
+      "loss": 0.4968,
+      "step": 1640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0171058521202578e-05,
+      "loss": 0.4938,
+      "step": 1641
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0160997151896712e-05,
+      "loss": 0.4889,
+      "step": 1642
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150935619566637e-05,
+      "loss": 0.4984,
+      "step": 1643
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0140873934400568e-05,
+      "loss": 0.4899,
+      "step": 1644
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.013081210658687e-05,
+      "loss": 0.4901,
+      "step": 1645
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0120750146314053e-05,
+      "loss": 0.4913,
+      "step": 1646
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0110688063770773e-05,
+      "loss": 0.4817,
+      "step": 1647
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.010062586914579e-05,
+      "loss": 0.4655,
+      "step": 1648
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0090563572627994e-05,
+      "loss": 0.4841,
+      "step": 1649
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0080501184406373e-05,
+      "loss": 0.4778,
+      "step": 1650
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0070438714670004e-05,
+      "loss": 0.4863,
+      "step": 1651
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0060376173608053e-05,
+      "loss": 0.4718,
+      "step": 1652
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0050313571409751e-05,
+      "loss": 0.5013,
+      "step": 1653
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.00402509182644e-05,
+      "loss": 0.4868,
+      "step": 1654
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0030188224361347e-05,
+      "loss": 0.4813,
+      "step": 1655
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0020125499889976e-05,
+      "loss": 0.4716,
+      "step": 1656
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0010062755039716e-05,
+      "loss": 0.4666,
+      "step": 1657
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1e-05,
+      "loss": 0.4889,
+      "step": 1658
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.989937244960287e-06,
+      "loss": 0.4592,
+      "step": 1659
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.979874500110026e-06,
+      "loss": 0.4711,
+      "step": 1660
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.969811775638655e-06,
+      "loss": 0.4536,
+      "step": 1661
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.959749081735604e-06,
+      "loss": 0.5002,
+      "step": 1662
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.949686428590252e-06,
+      "loss": 0.5,
+      "step": 1663
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.93962382639195e-06,
+      "loss": 0.4973,
+      "step": 1664
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.929561285329998e-06,
+      "loss": 0.4681,
+      "step": 1665
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.91949881559363e-06,
+      "loss": 0.4747,
+      "step": 1666
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.909436427372007e-06,
+      "loss": 0.4569,
+      "step": 1667
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.899374130854215e-06,
+      "loss": 0.4725,
+      "step": 1668
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.889311936229234e-06,
+      "loss": 0.4633,
+      "step": 1669
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.87924985368595e-06,
+      "loss": 0.4666,
+      "step": 1670
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.869187893413134e-06,
+      "loss": 0.5184,
+      "step": 1671
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.859126065599435e-06,
+      "loss": 0.4993,
+      "step": 1672
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.849064380433364e-06,
+      "loss": 0.5129,
+      "step": 1673
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.83900284810329e-06,
+      "loss": 0.5027,
+      "step": 1674
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.828941478797429e-06,
+      "loss": 0.4976,
+      "step": 1675
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.818880282703823e-06,
+      "loss": 0.467,
+      "step": 1676
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.808819270010348e-06,
+      "loss": 0.4734,
+      "step": 1677
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.798758450904693e-06,
+      "loss": 0.5001,
+      "step": 1678
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.788697835574348e-06,
+      "loss": 0.5046,
+      "step": 1679
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.778637434206596e-06,
+      "loss": 0.5261,
+      "step": 1680
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.76857725698851e-06,
+      "loss": 0.4698,
+      "step": 1681
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.758517314106932e-06,
+      "loss": 0.4678,
+      "step": 1682
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.748457615748459e-06,
+      "loss": 0.4915,
+      "step": 1683
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.738398172099452e-06,
+      "loss": 0.5317,
+      "step": 1684
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.728338993346009e-06,
+      "loss": 0.5005,
+      "step": 1685
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.718280089673959e-06,
+      "loss": 0.524,
+      "step": 1686
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.708221471268856e-06,
+      "loss": 0.4869,
+      "step": 1687
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.698163148315967e-06,
+      "loss": 0.4742,
+      "step": 1688
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.688105131000245e-06,
+      "loss": 0.5087,
+      "step": 1689
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.678047429506352e-06,
+      "loss": 0.5171,
+      "step": 1690
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.667990054018619e-06,
+      "loss": 0.4734,
+      "step": 1691
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.657933014721051e-06,
+      "loss": 0.4789,
+      "step": 1692
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.647876321797314e-06,
+      "loss": 0.4561,
+      "step": 1693
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.637819985430719e-06,
+      "loss": 0.4617,
+      "step": 1694
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.627764015804225e-06,
+      "loss": 0.463,
+      "step": 1695
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.6177084231004e-06,
+      "loss": 0.4907,
+      "step": 1696
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.607653217501452e-06,
+      "loss": 0.498,
+      "step": 1697
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.597598409189185e-06,
+      "loss": 0.5052,
+      "step": 1698
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.587544008345004e-06,
+      "loss": 0.5002,
+      "step": 1699
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.577490025149904e-06,
+      "loss": 0.5007,
+      "step": 1700
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.567436469784453e-06,
+      "loss": 0.4932,
+      "step": 1701
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.557383352428786e-06,
+      "loss": 0.4634,
+      "step": 1702
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.547330683262596e-06,
+      "loss": 0.5017,
+      "step": 1703
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.537278472465121e-06,
+      "loss": 0.4814,
+      "step": 1704
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.527226730215137e-06,
+      "loss": 0.4521,
+      "step": 1705
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.517175466690943e-06,
+      "loss": 0.4856,
+      "step": 1706
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.507124692070356e-06,
+      "loss": 0.5043,
+      "step": 1707
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.497074416530701e-06,
+      "loss": 0.4869,
+      "step": 1708
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.48702465024878e-06,
+      "loss": 0.5093,
+      "step": 1709
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.476975403400904e-06,
+      "loss": 0.4772,
+      "step": 1710
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.466926686162839e-06,
+      "loss": 0.4954,
+      "step": 1711
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.456878508709827e-06,
+      "loss": 0.4917,
+      "step": 1712
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.446830881216557e-06,
+      "loss": 0.4829,
+      "step": 1713
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436783813857161e-06,
+      "loss": 0.4805,
+      "step": 1714
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.42673731680521e-06,
+      "loss": 0.4971,
+      "step": 1715
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.416691400233687e-06,
+      "loss": 0.5077,
+      "step": 1716
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.406646074314998e-06,
+      "loss": 0.4781,
+      "step": 1717
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.396601349220942e-06,
+      "loss": 0.4674,
+      "step": 1718
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.386557235122716e-06,
+      "loss": 0.467,
+      "step": 1719
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.376513742190897e-06,
+      "loss": 0.4988,
+      "step": 1720
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.366470880595434e-06,
+      "loss": 0.4939,
+      "step": 1721
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.356428660505627e-06,
+      "loss": 0.4757,
+      "step": 1722
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.346387092090139e-06,
+      "loss": 0.476,
+      "step": 1723
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.336346185516966e-06,
+      "loss": 0.4758,
+      "step": 1724
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.32630595095344e-06,
+      "loss": 0.5076,
+      "step": 1725
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.316266398566206e-06,
+      "loss": 0.4935,
+      "step": 1726
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.306227538521221e-06,
+      "loss": 0.4718,
+      "step": 1727
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.296189380983747e-06,
+      "loss": 0.4706,
+      "step": 1728
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.286151936118316e-06,
+      "loss": 0.4948,
+      "step": 1729
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.276115214088758e-06,
+      "loss": 0.4744,
+      "step": 1730
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.266079225058164e-06,
+      "loss": 0.493,
+      "step": 1731
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.256043979188881e-06,
+      "loss": 0.483,
+      "step": 1732
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.246009486642506e-06,
+      "loss": 0.4916,
+      "step": 1733
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.235975757579872e-06,
+      "loss": 0.5054,
+      "step": 1734
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.225942802161042e-06,
+      "loss": 0.4965,
+      "step": 1735
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.215910630545288e-06,
+      "loss": 0.4733,
+      "step": 1736
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.205879252891094e-06,
+      "loss": 0.4917,
+      "step": 1737
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.195848679356144e-06,
+      "loss": 0.4967,
+      "step": 1738
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.185818920097296e-06,
+      "loss": 0.4762,
+      "step": 1739
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.175789985270594e-06,
+      "loss": 0.4755,
+      "step": 1740
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16576188503125e-06,
+      "loss": 0.4945,
+      "step": 1741
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.155734629533612e-06,
+      "loss": 0.4917,
+      "step": 1742
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.145708228931191e-06,
+      "loss": 0.4788,
+      "step": 1743
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.135682693376626e-06,
+      "loss": 0.4666,
+      "step": 1744
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.125658033021683e-06,
+      "loss": 0.4783,
+      "step": 1745
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.115634258017235e-06,
+      "loss": 0.4973,
+      "step": 1746
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.105611378513267e-06,
+      "loss": 0.4706,
+      "step": 1747
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095589404658852e-06,
+      "loss": 0.486,
+      "step": 1748
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.085568346602146e-06,
+      "loss": 0.4911,
+      "step": 1749
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.075548214490377e-06,
+      "loss": 0.4888,
+      "step": 1750
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.065529018469839e-06,
+      "loss": 0.4697,
+      "step": 1751
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055510768685878e-06,
+      "loss": 0.4885,
+      "step": 1752
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.045493475282876e-06,
+      "loss": 0.5062,
+      "step": 1753
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.035477148404257e-06,
+      "loss": 0.5118,
+      "step": 1754
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.025461798192453e-06,
+      "loss": 0.4999,
+      "step": 1755
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.015447434788915e-06,
+      "loss": 0.4828,
+      "step": 1756
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.005434068334095e-06,
+      "loss": 0.4844,
+      "step": 1757
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.995421708967437e-06,
+      "loss": 0.4932,
+      "step": 1758
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.985410366827358e-06,
+      "loss": 0.4764,
+      "step": 1759
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.97540005205125e-06,
+      "loss": 0.468,
+      "step": 1760
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.965390774775472e-06,
+      "loss": 0.4917,
+      "step": 1761
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.955382545135314e-06,
+      "loss": 0.4856,
+      "step": 1762
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945375373265017e-06,
+      "loss": 0.4703,
+      "step": 1763
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.935369269297755e-06,
+      "loss": 0.5012,
+      "step": 1764
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.925364243365611e-06,
+      "loss": 0.5096,
+      "step": 1765
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.915360305599584e-06,
+      "loss": 0.4845,
+      "step": 1766
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905357466129565e-06,
+      "loss": 0.4878,
+      "step": 1767
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.895355735084338e-06,
+      "loss": 0.5088,
+      "step": 1768
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88535512259156e-06,
+      "loss": 0.476,
+      "step": 1769
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.875355638777757e-06,
+      "loss": 0.4749,
+      "step": 1770
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.865357293768316e-06,
+      "loss": 0.5022,
+      "step": 1771
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.855360097687467e-06,
+      "loss": 0.4629,
+      "step": 1772
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.845364060658274e-06,
+      "loss": 0.4556,
+      "step": 1773
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.83536919280264e-06,
+      "loss": 0.4772,
+      "step": 1774
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.825375504241262e-06,
+      "loss": 0.4742,
+      "step": 1775
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.81538300509366e-06,
+      "loss": 0.4895,
+      "step": 1776
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.805391705478149e-06,
+      "loss": 0.4814,
+      "step": 1777
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.79540161551182e-06,
+      "loss": 0.4751,
+      "step": 1778
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.78541274531055e-06,
+      "loss": 0.4833,
+      "step": 1779
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.775425104988972e-06,
+      "loss": 0.485,
+      "step": 1780
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.765438704660481e-06,
+      "loss": 0.5011,
+      "step": 1781
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.755453554437207e-06,
+      "loss": 0.5001,
+      "step": 1782
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.745469664430025e-06,
+      "loss": 0.4855,
+      "step": 1783
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.735487044748523e-06,
+      "loss": 0.4766,
+      "step": 1784
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.725505705501013e-06,
+      "loss": 0.4736,
+      "step": 1785
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.715525656794504e-06,
+      "loss": 0.4845,
+      "step": 1786
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.705546908734704e-06,
+      "loss": 0.5095,
+      "step": 1787
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.695569471425992e-06,
+      "loss": 0.4771,
+      "step": 1788
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.685593354971433e-06,
+      "loss": 0.5049,
+      "step": 1789
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.675618569472747e-06,
+      "loss": 0.4753,
+      "step": 1790
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.665645125030312e-06,
+      "loss": 0.4674,
+      "step": 1791
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.65567303174314e-06,
+      "loss": 0.4532,
+      "step": 1792
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.645702299708881e-06,
+      "loss": 0.4807,
+      "step": 1793
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.63573293902381e-06,
+      "loss": 0.4863,
+      "step": 1794
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.6257649597828e-06,
+      "loss": 0.4609,
+      "step": 1795
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.615798372079338e-06,
+      "loss": 0.4843,
+      "step": 1796
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.605833186005496e-06,
+      "loss": 0.4919,
+      "step": 1797
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.595869411651931e-06,
+      "loss": 0.4707,
+      "step": 1798
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.585907059107864e-06,
+      "loss": 0.4705,
+      "step": 1799
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.575946138461084e-06,
+      "loss": 0.4812,
+      "step": 1800
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.565986659797924e-06,
+      "loss": 0.4913,
+      "step": 1801
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.556028633203256e-06,
+      "loss": 0.4984,
+      "step": 1802
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.54607206876049e-06,
+      "loss": 0.4924,
+      "step": 1803
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.536116976551546e-06,
+      "loss": 0.4902,
+      "step": 1804
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.526163366656858e-06,
+      "loss": 0.4731,
+      "step": 1805
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.51621124915536e-06,
+      "loss": 0.5125,
+      "step": 1806
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.506260634124476e-06,
+      "loss": 0.4844,
+      "step": 1807
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.496311531640098e-06,
+      "loss": 0.4769,
+      "step": 1808
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.486363951776597e-06,
+      "loss": 0.4717,
+      "step": 1809
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.4764179046068e-06,
+      "loss": 0.4762,
+      "step": 1810
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.466473400201984e-06,
+      "loss": 0.4635,
+      "step": 1811
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456530448631856e-06,
+      "loss": 0.4697,
+      "step": 1812
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.446589059964559e-06,
+      "loss": 0.5277,
+      "step": 1813
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.436649244266652e-06,
+      "loss": 0.4795,
+      "step": 1814
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.426711011603096e-06,
+      "loss": 0.5084,
+      "step": 1815
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.416774372037253e-06,
+      "loss": 0.5002,
+      "step": 1816
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.406839335630872e-06,
+      "loss": 0.5081,
+      "step": 1817
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.396905912444075e-06,
+      "loss": 0.4866,
+      "step": 1818
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.38697411253536e-06,
+      "loss": 0.4805,
+      "step": 1819
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.377043945961566e-06,
+      "loss": 0.4751,
+      "step": 1820
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3671154227779e-06,
+      "loss": 0.4735,
+      "step": 1821
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.357188553037878e-06,
+      "loss": 0.4608,
+      "step": 1822
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.34726334679336e-06,
+      "loss": 0.4965,
+      "step": 1823
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.337339814094518e-06,
+      "loss": 0.4598,
+      "step": 1824
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.32741796498983e-06,
+      "loss": 0.4645,
+      "step": 1825
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.317497809526062e-06,
+      "loss": 0.493,
+      "step": 1826
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307579357748281e-06,
+      "loss": 0.4827,
+      "step": 1827
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.297662619699807e-06,
+      "loss": 0.4529,
+      "step": 1828
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.287747605422241e-06,
+      "loss": 0.4829,
+      "step": 1829
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.277834324955434e-06,
+      "loss": 0.4926,
+      "step": 1830
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.267922788337482e-06,
+      "loss": 0.5046,
+      "step": 1831
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25801300560471e-06,
+      "loss": 0.4885,
+      "step": 1832
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.248104986791677e-06,
+      "loss": 0.4744,
+      "step": 1833
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.238198741931149e-06,
+      "loss": 0.5159,
+      "step": 1834
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.228294281054093e-06,
+      "loss": 0.5002,
+      "step": 1835
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.218391614189676e-06,
+      "loss": 0.494,
+      "step": 1836
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.208490751365245e-06,
+      "loss": 0.4874,
+      "step": 1837
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198591702606322e-06,
+      "loss": 0.4673,
+      "step": 1838
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.188694477936593e-06,
+      "loss": 0.4845,
+      "step": 1839
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.178799087377894e-06,
+      "loss": 0.4822,
+      "step": 1840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.168905540950202e-06,
+      "loss": 0.4781,
+      "step": 1841
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15901384867163e-06,
+      "loss": 0.4786,
+      "step": 1842
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.149124020558417e-06,
+      "loss": 0.4603,
+      "step": 1843
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.139236066624906e-06,
+      "loss": 0.494,
+      "step": 1844
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.129349996883551e-06,
+      "loss": 0.4678,
+      "step": 1845
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.11946582134489e-06,
+      "loss": 0.4865,
+      "step": 1846
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10958355001755e-06,
+      "loss": 0.4884,
+      "step": 1847
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.099703192908223e-06,
+      "loss": 0.4888,
+      "step": 1848
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089824760021668e-06,
+      "loss": 0.4869,
+      "step": 1849
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.079948261360695e-06,
+      "loss": 0.4908,
+      "step": 1850
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.070073706926151e-06,
+      "loss": 0.4723,
+      "step": 1851
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.06020110671692e-06,
+      "loss": 0.5149,
+      "step": 1852
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.050330470729904e-06,
+      "loss": 0.4891,
+      "step": 1853
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.04046180896002e-06,
+      "loss": 0.4608,
+      "step": 1854
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.030595131400176e-06,
+      "loss": 0.4731,
+      "step": 1855
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.020730448041279e-06,
+      "loss": 0.4738,
+      "step": 1856
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.010867768872218e-06,
+      "loss": 0.4856,
+      "step": 1857
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.001007103879847e-06,
+      "loss": 0.4651,
+      "step": 1858
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.991148463048987e-06,
+      "loss": 0.5167,
+      "step": 1859
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.981291856362405e-06,
+      "loss": 0.4924,
+      "step": 1860
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.971437293800804e-06,
+      "loss": 0.4817,
+      "step": 1861
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.961584785342824e-06,
+      "loss": 0.4523,
+      "step": 1862
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.951734340965022e-06,
+      "loss": 0.4878,
+      "step": 1863
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.94188597064187e-06,
+      "loss": 0.489,
+      "step": 1864
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.932039684345731e-06,
+      "loss": 0.4698,
+      "step": 1865
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.922195492046865e-06,
+      "loss": 0.4546,
+      "step": 1866
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.912353403713411e-06,
+      "loss": 0.4847,
+      "step": 1867
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90251342931137e-06,
+      "loss": 0.5052,
+      "step": 1868
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.892675578804611e-06,
+      "loss": 0.4795,
+      "step": 1869
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.88283986215485e-06,
+      "loss": 0.4746,
+      "step": 1870
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.873006289321642e-06,
+      "loss": 0.4771,
+      "step": 1871
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.863174870262369e-06,
+      "loss": 0.4995,
+      "step": 1872
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.85334561493224e-06,
+      "loss": 0.4854,
+      "step": 1873
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.843518533284257e-06,
+      "loss": 0.465,
+      "step": 1874
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.833693635269235e-06,
+      "loss": 0.4851,
+      "step": 1875
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.823870930835778e-06,
+      "loss": 0.5083,
+      "step": 1876
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.814050429930258e-06,
+      "loss": 0.4813,
+      "step": 1877
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.804232142496829e-06,
+      "loss": 0.4912,
+      "step": 1878
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.79441607847739e-06,
+      "loss": 0.495,
+      "step": 1879
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.784602247811605e-06,
+      "loss": 0.4771,
+      "step": 1880
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.4618,
+      "step": 1881
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.764981326288273e-06,
+      "loss": 0.4857,
+      "step": 1882
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.755174255298692e-06,
+      "loss": 0.4742,
+      "step": 1883
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.745369457398665e-06,
+      "loss": 0.5139,
+      "step": 1884
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.735566942516433e-06,
+      "loss": 0.4827,
+      "step": 1885
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.725766720577933e-06,
+      "loss": 0.5016,
+      "step": 1886
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.715968801506786e-06,
+      "loss": 0.4501,
+      "step": 1887
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.70617319522426e-06,
+      "loss": 0.4745,
+      "step": 1888
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.696379911649303e-06,
+      "loss": 0.475,
+      "step": 1889
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.686588960698498e-06,
+      "loss": 0.4774,
+      "step": 1890
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.676800352286074e-06,
+      "loss": 0.4929,
+      "step": 1891
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.667014096323882e-06,
+      "loss": 0.4912,
+      "step": 1892
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.657230202721397e-06,
+      "loss": 0.4866,
+      "step": 1893
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.647448681385692e-06,
+      "loss": 0.4838,
+      "step": 1894
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.637669542221446e-06,
+      "loss": 0.5137,
+      "step": 1895
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6278927951309255e-06,
+      "loss": 0.4669,
+      "step": 1896
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.6181184500139716e-06,
+      "loss": 0.4922,
+      "step": 1897
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.608346516767997e-06,
+      "loss": 0.4874,
+      "step": 1898
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.598577005287968e-06,
+      "loss": 0.4802,
+      "step": 1899
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.588809925466403e-06,
+      "loss": 0.473,
+      "step": 1900
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.579045287193352e-06,
+      "loss": 0.502,
+      "step": 1901
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.569283100356397e-06,
+      "loss": 0.4745,
+      "step": 1902
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.559523374840639e-06,
+      "loss": 0.4878,
+      "step": 1903
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.549766120528684e-06,
+      "loss": 0.4851,
+      "step": 1904
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.540011347300636e-06,
+      "loss": 0.4783,
+      "step": 1905
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.530259065034089e-06,
+      "loss": 0.4801,
+      "step": 1906
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.520509283604114e-06,
+      "loss": 0.4895,
+      "step": 1907
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.510762012883243e-06,
+      "loss": 0.5037,
+      "step": 1908
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.501017262741472e-06,
+      "loss": 0.4661,
+      "step": 1909
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.491275043046246e-06,
+      "loss": 0.4706,
+      "step": 1910
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4815353636624474e-06,
+      "loss": 0.4764,
+      "step": 1911
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.471798234452381e-06,
+      "loss": 0.4738,
+      "step": 1912
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.462063665275777e-06,
+      "loss": 0.4558,
+      "step": 1913
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.4523316659897645e-06,
+      "loss": 0.4849,
+      "step": 1914
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.442602246448875e-06,
+      "loss": 0.4825,
+      "step": 1915
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.432875416505031e-06,
+      "loss": 0.4685,
+      "step": 1916
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423151186007527e-06,
+      "loss": 0.4845,
+      "step": 1917
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.41342956480303e-06,
+      "loss": 0.4888,
+      "step": 1918
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.40371056273556e-06,
+      "loss": 0.486,
+      "step": 1919
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.393994189646494e-06,
+      "loss": 0.4905,
+      "step": 1920
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384280455374531e-06,
+      "loss": 0.4624,
+      "step": 1921
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.374569369755713e-06,
+      "loss": 0.4803,
+      "step": 1922
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.364860942623393e-06,
+      "loss": 0.4791,
+      "step": 1923
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.355155183808234e-06,
+      "loss": 0.4802,
+      "step": 1924
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.345452103138195e-06,
+      "loss": 0.4956,
+      "step": 1925
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.335751710438533e-06,
+      "loss": 0.5065,
+      "step": 1926
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.326054015531763e-06,
+      "loss": 0.4749,
+      "step": 1927
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.316359028237685e-06,
+      "loss": 0.4778,
+      "step": 1928
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.306666758373353e-06,
+      "loss": 0.4701,
+      "step": 1929
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.29697721575307e-06,
+      "loss": 0.4686,
+      "step": 1930
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.287290410188374e-06,
+      "loss": 0.4893,
+      "step": 1931
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2776063514880336e-06,
+      "loss": 0.5022,
+      "step": 1932
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.26792504945804e-06,
+      "loss": 0.4762,
+      "step": 1933
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2582465139015834e-06,
+      "loss": 0.4645,
+      "step": 1934
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.248570754619062e-06,
+      "loss": 0.4978,
+      "step": 1935
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.238897781408059e-06,
+      "loss": 0.4614,
+      "step": 1936
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2292276040633334e-06,
+      "loss": 0.4606,
+      "step": 1937
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.219560232376821e-06,
+      "loss": 0.5022,
+      "step": 1938
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209895676137608e-06,
+      "loss": 0.4688,
+      "step": 1939
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.2002339451319405e-06,
+      "loss": 0.4878,
+      "step": 1940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1905750491431865e-06,
+      "loss": 0.4982,
+      "step": 1941
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.18091899795186e-06,
+      "loss": 0.4904,
+      "step": 1942
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.171265801335587e-06,
+      "loss": 0.4575,
+      "step": 1943
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1616154690691055e-06,
+      "loss": 0.4834,
+      "step": 1944
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1519680109242486e-06,
+      "loss": 0.4718,
+      "step": 1945
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.142323436669946e-06,
+      "loss": 0.4763,
+      "step": 1946
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.132681756072199e-06,
+      "loss": 0.4696,
+      "step": 1947
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.123042978894088e-06,
+      "loss": 0.5006,
+      "step": 1948
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.113407114895744e-06,
+      "loss": 0.4971,
+      "step": 1949
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.103774173834356e-06,
+      "loss": 0.4906,
+      "step": 1950
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.094144165464147e-06,
+      "loss": 0.467,
+      "step": 1951
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.084517099536378e-06,
+      "loss": 0.4795,
+      "step": 1952
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0748929857993264e-06,
+      "loss": 0.4835,
+      "step": 1953
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.065271833998272e-06,
+      "loss": 0.4806,
+      "step": 1954
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.055653653875508e-06,
+      "loss": 0.4891,
+      "step": 1955
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.046038455170313e-06,
+      "loss": 0.5035,
+      "step": 1956
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.036426247618948e-06,
+      "loss": 0.486,
+      "step": 1957
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0268170409546444e-06,
+      "loss": 0.4801,
+      "step": 1958
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.017210844907598e-06,
+      "loss": 0.4897,
+      "step": 1959
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.007607669204946e-06,
+      "loss": 0.5162,
+      "step": 1960
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.998007523570777e-06,
+      "loss": 0.4646,
+      "step": 1961
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.988410417726113e-06,
+      "loss": 0.4651,
+      "step": 1962
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.978816361388887e-06,
+      "loss": 0.5276,
+      "step": 1963
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.969225364273955e-06,
+      "loss": 0.4771,
+      "step": 1964
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9596374360930695e-06,
+      "loss": 0.4946,
+      "step": 1965
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.950052586554881e-06,
+      "loss": 0.5074,
+      "step": 1966
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.940470825364911e-06,
+      "loss": 0.494,
+      "step": 1967
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.930892162225565e-06,
+      "loss": 0.4598,
+      "step": 1968
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.921316606836108e-06,
+      "loss": 0.4825,
+      "step": 1969
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.911744168892658e-06,
+      "loss": 0.4907,
+      "step": 1970
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.902174858088177e-06,
+      "loss": 0.4934,
+      "step": 1971
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.892608684112456e-06,
+      "loss": 0.4568,
+      "step": 1972
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.883045656652122e-06,
+      "loss": 0.4814,
+      "step": 1973
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.873485785390595e-06,
+      "loss": 0.4926,
+      "step": 1974
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.863929080008118e-06,
+      "loss": 0.4786,
+      "step": 1975
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.85437555018172e-06,
+      "loss": 0.4696,
+      "step": 1976
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8448252055852175e-06,
+      "loss": 0.4827,
+      "step": 1977
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8352780558891975e-06,
+      "loss": 0.4896,
+      "step": 1978
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.825734110761018e-06,
+      "loss": 0.48,
+      "step": 1979
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.816193379864786e-06,
+      "loss": 0.4777,
+      "step": 1980
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.806655872861357e-06,
+      "loss": 0.4755,
+      "step": 1981
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.797121599408319e-06,
+      "loss": 0.4846,
+      "step": 1982
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.787590569159992e-06,
+      "loss": 0.4767,
+      "step": 1983
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.778062791767406e-06,
+      "loss": 0.4919,
+      "step": 1984
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7685382768783e-06,
+      "loss": 0.4949,
+      "step": 1985
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.759017034137114e-06,
+      "loss": 0.476,
+      "step": 1986
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.749499073184957e-06,
+      "loss": 0.4705,
+      "step": 1987
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.739984403659636e-06,
+      "loss": 0.4795,
+      "step": 1988
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7304730351956145e-06,
+      "loss": 0.491,
+      "step": 1989
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.720964977424014e-06,
+      "loss": 0.4659,
+      "step": 1990
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.711460239972607e-06,
+      "loss": 0.4651,
+      "step": 1991
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7019588324658005e-06,
+      "loss": 0.4712,
+      "step": 1992
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6924607645246355e-06,
+      "loss": 0.47,
+      "step": 1993
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.682966045766758e-06,
+      "loss": 0.4998,
+      "step": 1994
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.673474685806436e-06,
+      "loss": 0.4847,
+      "step": 1995
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.66398669425453e-06,
+      "loss": 0.4889,
+      "step": 1996
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.654502080718494e-06,
+      "loss": 0.4747,
+      "step": 1997
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.64502085480236e-06,
+      "loss": 0.4644,
+      "step": 1998
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.6355430261067256e-06,
+      "loss": 0.4727,
+      "step": 1999
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.626068604228752e-06,
+      "loss": 0.4896,
+      "step": 2000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.616597598762151e-06,
+      "loss": 0.4574,
+      "step": 2001
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.607130019297174e-06,
+      "loss": 0.4983,
+      "step": 2002
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597665875420604e-06,
+      "loss": 0.4873,
+      "step": 2003
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.588205176715743e-06,
+      "loss": 0.4763,
+      "step": 2004
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.578747932762406e-06,
+      "loss": 0.5138,
+      "step": 2005
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5692941531369155e-06,
+      "loss": 0.4846,
+      "step": 2006
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.559843847412073e-06,
+      "loss": 0.4865,
+      "step": 2007
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.550397025157169e-06,
+      "loss": 0.469,
+      "step": 2008
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.540953695937973e-06,
+      "loss": 0.4452,
+      "step": 2009
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.531513869316708e-06,
+      "loss": 0.4847,
+      "step": 2010
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522077554852058e-06,
+      "loss": 0.4845,
+      "step": 2011
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.512644762099143e-06,
+      "loss": 0.4667,
+      "step": 2012
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.503215500609521e-06,
+      "loss": 0.4757,
+      "step": 2013
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.493789779931177e-06,
+      "loss": 0.4844,
+      "step": 2014
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.484367609608503e-06,
+      "loss": 0.4705,
+      "step": 2015
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.474948999182304e-06,
+      "loss": 0.4861,
+      "step": 2016
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.465533958189778e-06,
+      "loss": 0.4841,
+      "step": 2017
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.456122496164503e-06,
+      "loss": 0.449,
+      "step": 2018
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.446714622636442e-06,
+      "loss": 0.4863,
+      "step": 2019
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.437310347131916e-06,
+      "loss": 0.4536,
+      "step": 2020
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.427909679173607e-06,
+      "loss": 0.4846,
+      "step": 2021
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418512628280544e-06,
+      "loss": 0.4808,
+      "step": 2022
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.409119203968095e-06,
+      "loss": 0.4944,
+      "step": 2023
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.399729415747951e-06,
+      "loss": 0.5104,
+      "step": 2024
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.390343273128123e-06,
+      "loss": 0.4779,
+      "step": 2025
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.380960785612938e-06,
+      "loss": 0.4666,
+      "step": 2026
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3715819627030065e-06,
+      "loss": 0.4797,
+      "step": 2027
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.362206813895238e-06,
+      "loss": 0.485,
+      "step": 2028
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.352835348682824e-06,
+      "loss": 0.4637,
+      "step": 2029
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.343467576555222e-06,
+      "loss": 0.4736,
+      "step": 2030
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3341035069981486e-06,
+      "loss": 0.4846,
+      "step": 2031
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3247431494935774e-06,
+      "loss": 0.4692,
+      "step": 2032
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3153865135197125e-06,
+      "loss": 0.4862,
+      "step": 2033
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3060336085509985e-06,
+      "loss": 0.457,
+      "step": 2034
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2966844440581e-06,
+      "loss": 0.4941,
+      "step": 2035
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.287339029507894e-06,
+      "loss": 0.4955,
+      "step": 2036
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.277997374363458e-06,
+      "loss": 0.4744,
+      "step": 2037
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.268659488084068e-06,
+      "loss": 0.4696,
+      "step": 2038
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.259325380125181e-06,
+      "loss": 0.504,
+      "step": 2039
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.249995059938422e-06,
+      "loss": 0.5016,
+      "step": 2040
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.240668536971588e-06,
+      "loss": 0.4612,
+      "step": 2041
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.231345820668628e-06,
+      "loss": 0.4691,
+      "step": 2042
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.22202692046964e-06,
+      "loss": 0.4651,
+      "step": 2043
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.212711845810855e-06,
+      "loss": 0.4833,
+      "step": 2044
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.2034006061246295e-06,
+      "loss": 0.5123,
+      "step": 2045
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.194093210839438e-06,
+      "loss": 0.4816,
+      "step": 2046
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1847896693798606e-06,
+      "loss": 0.4506,
+      "step": 2047
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.175489991166577e-06,
+      "loss": 0.4811,
+      "step": 2048
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.166194185616356e-06,
+      "loss": 0.4747,
+      "step": 2049
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.156902262142042e-06,
+      "loss": 0.4957,
+      "step": 2050
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.147614230152548e-06,
+      "loss": 0.4866,
+      "step": 2051
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.138330099052854e-06,
+      "loss": 0.4689,
+      "step": 2052
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.129049878243978e-06,
+      "loss": 0.46,
+      "step": 2053
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.119773577122986e-06,
+      "loss": 0.4801,
+      "step": 2054
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1105012050829756e-06,
+      "loss": 0.4985,
+      "step": 2055
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.101232771513064e-06,
+      "loss": 0.4814,
+      "step": 2056
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.09196828579838e-06,
+      "loss": 0.4856,
+      "step": 2057
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.082707757320054e-06,
+      "loss": 0.4592,
+      "step": 2058
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.073451195455216e-06,
+      "loss": 0.5095,
+      "step": 2059
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.064198609576966e-06,
+      "loss": 0.4827,
+      "step": 2060
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.054950009054389e-06,
+      "loss": 0.4998,
+      "step": 2061
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.045705403252531e-06,
+      "loss": 0.4891,
+      "step": 2062
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.036464801532396e-06,
+      "loss": 0.4858,
+      "step": 2063
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.027228213250926e-06,
+      "loss": 0.4785,
+      "step": 2064
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0179956477610115e-06,
+      "loss": 0.4636,
+      "step": 2065
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.008767114411455e-06,
+      "loss": 0.4801,
+      "step": 2066
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.999542622546985e-06,
+      "loss": 0.4369,
+      "step": 2067
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.990322181508235e-06,
+      "loss": 0.4733,
+      "step": 2068
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.981105800631739e-06,
+      "loss": 0.4843,
+      "step": 2069
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.971893489249918e-06,
+      "loss": 0.4681,
+      "step": 2070
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.962685256691071e-06,
+      "loss": 0.4866,
+      "step": 2071
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.953481112279373e-06,
+      "loss": 0.4595,
+      "step": 2072
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.944281065334847e-06,
+      "loss": 0.481,
+      "step": 2073
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.935085125173379e-06,
+      "loss": 0.4594,
+      "step": 2074
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9258933011066885e-06,
+      "loss": 0.4846,
+      "step": 2075
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.916705602442334e-06,
+      "loss": 0.4669,
+      "step": 2076
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9075220384836925e-06,
+      "loss": 0.4901,
+      "step": 2077
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.898342618529956e-06,
+      "loss": 0.4761,
+      "step": 2078
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8891673518761185e-06,
+      "loss": 0.4443,
+      "step": 2079
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.879996247812969e-06,
+      "loss": 0.4601,
+      "step": 2080
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.870829315627079e-06,
+      "loss": 0.4932,
+      "step": 2081
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.861666564600801e-06,
+      "loss": 0.4823,
+      "step": 2082
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.852508004012251e-06,
+      "loss": 0.4507,
+      "step": 2083
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.843353643135301e-06,
+      "loss": 0.454,
+      "step": 2084
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.834203491239574e-06,
+      "loss": 0.4849,
+      "step": 2085
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825057557590423e-06,
+      "loss": 0.4797,
+      "step": 2086
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.815915851448939e-06,
+      "loss": 0.4828,
+      "step": 2087
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.8067783820719256e-06,
+      "loss": 0.4763,
+      "step": 2088
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.797645158711901e-06,
+      "loss": 0.481,
+      "step": 2089
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.788516190617081e-06,
+      "loss": 0.4773,
+      "step": 2090
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.779391487031372e-06,
+      "loss": 0.4617,
+      "step": 2091
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7702710571943695e-06,
+      "loss": 0.4634,
+      "step": 2092
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7611549103413264e-06,
+      "loss": 0.4891,
+      "step": 2093
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.752043055703171e-06,
+      "loss": 0.4846,
+      "step": 2094
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.742935502506485e-06,
+      "loss": 0.4574,
+      "step": 2095
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.7338322599734885e-06,
+      "loss": 0.4576,
+      "step": 2096
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.724733337322039e-06,
+      "loss": 0.4691,
+      "step": 2097
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.715638743765629e-06,
+      "loss": 0.4883,
+      "step": 2098
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.706548488513347e-06,
+      "loss": 0.4641,
+      "step": 2099
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.697462580769906e-06,
+      "loss": 0.4985,
+      "step": 2100
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.68838102973561e-06,
+      "loss": 0.4589,
+      "step": 2101
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.679303844606354e-06,
+      "loss": 0.4856,
+      "step": 2102
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.67023103457361e-06,
+      "loss": 0.4918,
+      "step": 2103
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.66116260882442e-06,
+      "loss": 0.4772,
+      "step": 2104
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.652098576541388e-06,
+      "loss": 0.4591,
+      "step": 2105
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.643038946902668e-06,
+      "loss": 0.4919,
+      "step": 2106
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6339837290819576e-06,
+      "loss": 0.4719,
+      "step": 2107
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.624932932248484e-06,
+      "loss": 0.481,
+      "step": 2108
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.615886565566999e-06,
+      "loss": 0.4764,
+      "step": 2109
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.60684463819777e-06,
+      "loss": 0.4675,
+      "step": 2110
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5978071592965685e-06,
+      "loss": 0.5035,
+      "step": 2111
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588774138014664e-06,
+      "loss": 0.4669,
+      "step": 2112
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.579745583498802e-06,
+      "loss": 0.4663,
+      "step": 2113
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.570721504891214e-06,
+      "loss": 0.4832,
+      "step": 2114
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.561701911329602e-06,
+      "loss": 0.4691,
+      "step": 2115
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.552686811947119e-06,
+      "loss": 0.4751,
+      "step": 2116
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.543676215872371e-06,
+      "loss": 0.498,
+      "step": 2117
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.534670132229408e-06,
+      "loss": 0.4728,
+      "step": 2118
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.525668570137697e-06,
+      "loss": 0.4767,
+      "step": 2119
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.51667153871214e-06,
+      "loss": 0.4811,
+      "step": 2120
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.507679047063048e-06,
+      "loss": 0.5074,
+      "step": 2121
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.498691104296131e-06,
+      "loss": 0.4618,
+      "step": 2122
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489707719512498e-06,
+      "loss": 0.4602,
+      "step": 2123
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.480728901808641e-06,
+      "loss": 0.4741,
+      "step": 2124
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4717546602764285e-06,
+      "loss": 0.4906,
+      "step": 2125
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.462785004003087e-06,
+      "loss": 0.4633,
+      "step": 2126
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453819942071212e-06,
+      "loss": 0.4819,
+      "step": 2127
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4448594835587375e-06,
+      "loss": 0.4977,
+      "step": 2128
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.435903637538943e-06,
+      "loss": 0.4904,
+      "step": 2129
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.426952413080434e-06,
+      "loss": 0.4474,
+      "step": 2130
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418005819247139e-06,
+      "loss": 0.5201,
+      "step": 2131
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.40906386509829e-06,
+      "loss": 0.4612,
+      "step": 2132
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.400126559688428e-06,
+      "loss": 0.4763,
+      "step": 2133
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.391193912067386e-06,
+      "loss": 0.4331,
+      "step": 2134
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.382265931280279e-06,
+      "loss": 0.474,
+      "step": 2135
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3733426263674955e-06,
+      "loss": 0.4681,
+      "step": 2136
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.364424006364693e-06,
+      "loss": 0.4945,
+      "step": 2137
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.355510080302781e-06,
+      "loss": 0.4869,
+      "step": 2138
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3466008572079184e-06,
+      "loss": 0.4509,
+      "step": 2139
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.3376963461015015e-06,
+      "loss": 0.4857,
+      "step": 2140
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.328796556000153e-06,
+      "loss": 0.4818,
+      "step": 2141
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.31990149591572e-06,
+      "loss": 0.4814,
+      "step": 2142
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.311011174855255e-06,
+      "loss": 0.4725,
+      "step": 2143
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.302125601821015e-06,
+      "loss": 0.4852,
+      "step": 2144
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.293244785810452e-06,
+      "loss": 0.4691,
+      "step": 2145
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284368735816188e-06,
+      "loss": 0.48,
+      "step": 2146
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2754974608260336e-06,
+      "loss": 0.4698,
+      "step": 2147
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.266630969822958e-06,
+      "loss": 0.4071,
+      "step": 2148
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.257769271785087e-06,
+      "loss": 0.4409,
+      "step": 2149
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.2489123756856956e-06,
+      "loss": 0.4294,
+      "step": 2150
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.240060290493194e-06,
+      "loss": 0.4359,
+      "step": 2151
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.231213025171116e-06,
+      "loss": 0.4475,
+      "step": 2152
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.2223705886781225e-06,
+      "loss": 0.4097,
+      "step": 2153
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.213532989967982e-06,
+      "loss": 0.4391,
+      "step": 2154
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.204700237989564e-06,
+      "loss": 0.4235,
+      "step": 2155
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.195872341686832e-06,
+      "loss": 0.4265,
+      "step": 2156
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1870493099988285e-06,
+      "loss": 0.4252,
+      "step": 2157
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.178231151859678e-06,
+      "loss": 0.4261,
+      "step": 2158
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.169417876198559e-06,
+      "loss": 0.4257,
+      "step": 2159
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1606094919397144e-06,
+      "loss": 0.4098,
+      "step": 2160
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.151806008002431e-06,
+      "loss": 0.4165,
+      "step": 2161
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.1430074333010346e-06,
+      "loss": 0.426,
+      "step": 2162
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 5.134213776744881e-06,
+      "loss": 0.45,
+      "step": 2163
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.125425047238344e-06,
+      "loss": 0.4396,
+      "step": 2164
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.116641253680812e-06,
+      "loss": 0.4342,
+      "step": 2165
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.107862404966664e-06,
+      "loss": 0.4318,
+      "step": 2166
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.099088509985286e-06,
+      "loss": 0.4344,
+      "step": 2167
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.090319577621039e-06,
+      "loss": 0.4075,
+      "step": 2168
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.081555616753264e-06,
+      "loss": 0.4125,
+      "step": 2169
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.072796636256267e-06,
+      "loss": 0.4446,
+      "step": 2170
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.064042644999308e-06,
+      "loss": 0.4109,
+      "step": 2171
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.0552936518465955e-06,
+      "loss": 0.4327,
+      "step": 2172
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 5.046549665657281e-06,
+      "loss": 0.4093,
+      "step": 2173
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.0378106952854414e-06,
+      "loss": 0.411,
+      "step": 2174
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.029076749580075e-06,
+      "loss": 0.4154,
+      "step": 2175
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.020347837385096e-06,
+      "loss": 0.405,
+      "step": 2176
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.011623967539318e-06,
+      "loss": 0.4017,
+      "step": 2177
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 5.002905148876456e-06,
+      "loss": 0.3992,
+      "step": 2178
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.994191390225093e-06,
+      "loss": 0.4315,
+      "step": 2179
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.985482700408705e-06,
+      "loss": 0.3871,
+      "step": 2180
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.97677908824563e-06,
+      "loss": 0.412,
+      "step": 2181
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.968080562549065e-06,
+      "loss": 0.417,
+      "step": 2182
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.959387132127054e-06,
+      "loss": 0.4146,
+      "step": 2183
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 4.950698805782488e-06,
+      "loss": 0.4111,
+      "step": 2184
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.942015592313079e-06,
+      "loss": 0.4218,
+      "step": 2185
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.933337500511368e-06,
+      "loss": 0.4099,
+      "step": 2186
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.924664539164713e-06,
+      "loss": 0.4221,
+      "step": 2187
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.915996717055271e-06,
+      "loss": 0.4053,
+      "step": 2188
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.907334042959999e-06,
+      "loss": 0.4427,
+      "step": 2189
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.8986765256506395e-06,
+      "loss": 0.4377,
+      "step": 2190
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.890024173893718e-06,
+      "loss": 0.4254,
+      "step": 2191
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.881376996450516e-06,
+      "loss": 0.4517,
+      "step": 2192
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.872735002077089e-06,
+      "loss": 0.4189,
+      "step": 2193
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.864098199524241e-06,
+      "loss": 0.4175,
+      "step": 2194
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 4.855466597537515e-06,
+      "loss": 0.4281,
+      "step": 2195
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.84684020485719e-06,
+      "loss": 0.4283,
+      "step": 2196
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.838219030218274e-06,
+      "loss": 0.4113,
+      "step": 2197
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.829603082350489e-06,
+      "loss": 0.4149,
+      "step": 2198
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.820992369978257e-06,
+      "loss": 0.4177,
+      "step": 2199
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.812386901820708e-06,
+      "loss": 0.4045,
+      "step": 2200
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.8037866865916574e-06,
+      "loss": 0.4229,
+      "step": 2201
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.795191732999604e-06,
+      "loss": 0.4115,
+      "step": 2202
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.786602049747718e-06,
+      "loss": 0.433,
+      "step": 2203
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.77801764553383e-06,
+      "loss": 0.4187,
+      "step": 2204
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.7694385290504306e-06,
+      "loss": 0.4207,
+      "step": 2205
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 4.76086470898465e-06,
+      "loss": 0.4166,
+      "step": 2206
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.75229619401826e-06,
+      "loss": 0.4093,
+      "step": 2207
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.743732992827657e-06,
+      "loss": 0.424,
+      "step": 2208
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.735175114083861e-06,
+      "loss": 0.4244,
+      "step": 2209
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.726622566452498e-06,
+      "loss": 0.4373,
+      "step": 2210
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.718075358593802e-06,
+      "loss": 0.4331,
+      "step": 2211
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.709533499162589e-06,
+      "loss": 0.4064,
+      "step": 2212
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.700996996808268e-06,
+      "loss": 0.4207,
+      "step": 2213
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6924658601748215e-06,
+      "loss": 0.4099,
+      "step": 2214
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.683940097900801e-06,
+      "loss": 0.4171,
+      "step": 2215
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 4.6754197186193114e-06,
+      "loss": 0.4279,
+      "step": 2216
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.666904730958014e-06,
+      "loss": 0.422,
+      "step": 2217
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6583951435390975e-06,
+      "loss": 0.4083,
+      "step": 2218
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.649890964979295e-06,
+      "loss": 0.4427,
+      "step": 2219
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.641392203889857e-06,
+      "loss": 0.4042,
+      "step": 2220
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.632898868876549e-06,
+      "loss": 0.4232,
+      "step": 2221
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.6244109685396465e-06,
+      "loss": 0.3976,
+      "step": 2222
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.615928511473914e-06,
+      "loss": 0.3989,
+      "step": 2223
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.607451506268612e-06,
+      "loss": 0.4282,
+      "step": 2224
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.598979961507472e-06,
+      "loss": 0.4254,
+      "step": 2225
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5905138857687025e-06,
+      "loss": 0.4236,
+      "step": 2226
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 4.5820532876249725e-06,
+      "loss": 0.413,
+      "step": 2227
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.573598175643406e-06,
+      "loss": 0.4426,
+      "step": 2228
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.565148558385569e-06,
+      "loss": 0.4335,
+      "step": 2229
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.556704444407466e-06,
+      "loss": 0.4104,
+      "step": 2230
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.548265842259531e-06,
+      "loss": 0.421,
+      "step": 2231
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5398327604866056e-06,
+      "loss": 0.4176,
+      "step": 2232
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.531405207627953e-06,
+      "loss": 0.4402,
+      "step": 2233
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.522983192217237e-06,
+      "loss": 0.4235,
+      "step": 2234
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.514566722782509e-06,
+      "loss": 0.4426,
+      "step": 2235
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.5061558078462076e-06,
+      "loss": 0.4297,
+      "step": 2236
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.497750455925147e-06,
+      "loss": 0.401,
+      "step": 2237
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 4.489350675530508e-06,
+      "loss": 0.4229,
+      "step": 2238
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.48095647516783e-06,
+      "loss": 0.4425,
+      "step": 2239
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4725678633370015e-06,
+      "loss": 0.4226,
+      "step": 2240
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.464184848532251e-06,
+      "loss": 0.4221,
+      "step": 2241
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.455807439242143e-06,
+      "loss": 0.4236,
+      "step": 2242
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4474356439495635e-06,
+      "loss": 0.4147,
+      "step": 2243
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.439069471131717e-06,
+      "loss": 0.4065,
+      "step": 2244
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.430708929260105e-06,
+      "loss": 0.4137,
+      "step": 2245
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.422354026800536e-06,
+      "loss": 0.4047,
+      "step": 2246
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.414004772213109e-06,
+      "loss": 0.4266,
+      "step": 2247
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 4.4056611739521985e-06,
+      "loss": 0.4001,
+      "step": 2248
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.397323240466455e-06,
+      "loss": 0.4105,
+      "step": 2249
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.388990980198796e-06,
+      "loss": 0.4218,
+      "step": 2250
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.380664401586382e-06,
+      "loss": 0.4393,
+      "step": 2251
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.372343513060634e-06,
+      "loss": 0.4041,
+      "step": 2252
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.364028323047205e-06,
+      "loss": 0.4238,
+      "step": 2253
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3557188399659775e-06,
+      "loss": 0.431,
+      "step": 2254
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.3474150722310584e-06,
+      "loss": 0.4349,
+      "step": 2255
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.339117028250763e-06,
+      "loss": 0.4167,
+      "step": 2256
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.330824716427618e-06,
+      "loss": 0.4136,
+      "step": 2257
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.322538145158334e-06,
+      "loss": 0.4152,
+      "step": 2258
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 4.314257322833818e-06,
+      "loss": 0.4496,
+      "step": 2259
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.305982257839154e-06,
+      "loss": 0.4384,
+      "step": 2260
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.297712958553595e-06,
+      "loss": 0.4301,
+      "step": 2261
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2894494333505565e-06,
+      "loss": 0.4103,
+      "step": 2262
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.281191690597607e-06,
+      "loss": 0.4104,
+      "step": 2263
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.272939738656462e-06,
+      "loss": 0.4058,
+      "step": 2264
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.264693585882965e-06,
+      "loss": 0.4336,
+      "step": 2265
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.256453240627096e-06,
+      "loss": 0.4215,
+      "step": 2266
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.248218711232952e-06,
+      "loss": 0.4332,
+      "step": 2267
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.239990006038739e-06,
+      "loss": 0.4226,
+      "step": 2268
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2317671333767695e-06,
+      "loss": 0.4199,
+      "step": 2269
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 4.2235501015734445e-06,
+      "loss": 0.402,
+      "step": 2270
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.215338918949256e-06,
+      "loss": 0.4386,
+      "step": 2271
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.207133593818769e-06,
+      "loss": 0.4313,
+      "step": 2272
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.198934134490622e-06,
+      "loss": 0.4451,
+      "step": 2273
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1907405492675065e-06,
+      "loss": 0.4137,
+      "step": 2274
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.182552846446173e-06,
+      "loss": 0.4275,
+      "step": 2275
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.174371034317411e-06,
+      "loss": 0.4404,
+      "step": 2276
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.166195121166054e-06,
+      "loss": 0.3856,
+      "step": 2277
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.158025115270944e-06,
+      "loss": 0.4218,
+      "step": 2278
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.149861024904957e-06,
+      "loss": 0.4248,
+      "step": 2279
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.1417028583349775e-06,
+      "loss": 0.4313,
+      "step": 2280
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 4.133550623821884e-06,
+      "loss": 0.4205,
+      "step": 2281
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.125404329620557e-06,
+      "loss": 0.4089,
+      "step": 2282
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.117263983979855e-06,
+      "loss": 0.4202,
+      "step": 2283
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.109129595142623e-06,
+      "loss": 0.4373,
+      "step": 2284
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.101001171345656e-06,
+      "loss": 0.4496,
+      "step": 2285
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.092878720819724e-06,
+      "loss": 0.4292,
+      "step": 2286
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.084762251789547e-06,
+      "loss": 0.4318,
+      "step": 2287
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.076651772473783e-06,
+      "loss": 0.4056,
+      "step": 2288
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.06854729108503e-06,
+      "loss": 0.4432,
+      "step": 2289
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.060448815829811e-06,
+      "loss": 0.4039,
+      "step": 2290
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 4.05235635490856e-06,
+      "loss": 0.4639,
+      "step": 2291
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.044269916515631e-06,
+      "loss": 0.4404,
+      "step": 2292
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0361895088392776e-06,
+      "loss": 0.4065,
+      "step": 2293
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.028115140061643e-06,
+      "loss": 0.4283,
+      "step": 2294
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.0200468183587556e-06,
+      "loss": 0.4244,
+      "step": 2295
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.011984551900528e-06,
+      "loss": 0.4104,
+      "step": 2296
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 4.003928348850734e-06,
+      "loss": 0.4329,
+      "step": 2297
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.995878217367005e-06,
+      "loss": 0.433,
+      "step": 2298
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.987834165600832e-06,
+      "loss": 0.457,
+      "step": 2299
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.9797962016975465e-06,
+      "loss": 0.4147,
+      "step": 2300
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.971764333796316e-06,
+      "loss": 0.4452,
+      "step": 2301
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.963738570030135e-06,
+      "loss": 0.4058,
+      "step": 2302
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.955718918525815e-06,
+      "loss": 0.4285,
+      "step": 2303
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.94770538740398e-06,
+      "loss": 0.4388,
+      "step": 2304
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.93969798477906e-06,
+      "loss": 0.4191,
+      "step": 2305
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.931696718759274e-06,
+      "loss": 0.4288,
+      "step": 2306
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.923701597446628e-06,
+      "loss": 0.4329,
+      "step": 2307
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.915712628936908e-06,
+      "loss": 0.4421,
+      "step": 2308
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.90772982131967e-06,
+      "loss": 0.4346,
+      "step": 2309
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.899753182678232e-06,
+      "loss": 0.4368,
+      "step": 2310
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.891782721089658e-06,
+      "loss": 0.3988,
+      "step": 2311
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.883818444624765e-06,
+      "loss": 0.4457,
+      "step": 2312
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.875860361348106e-06,
+      "loss": 0.4325,
+      "step": 2313
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.867908479317961e-06,
+      "loss": 0.403,
+      "step": 2314
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.859962806586332e-06,
+      "loss": 0.4363,
+      "step": 2315
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.852023351198933e-06,
+      "loss": 0.4323,
+      "step": 2316
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.844090121195185e-06,
+      "loss": 0.4337,
+      "step": 2317
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8361631246081945e-06,
+      "loss": 0.4246,
+      "step": 2318
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.828242369464769e-06,
+      "loss": 0.4412,
+      "step": 2319
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.820327863785392e-06,
+      "loss": 0.4082,
+      "step": 2320
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8124196155842175e-06,
+      "loss": 0.4365,
+      "step": 2321
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.8045176328690615e-06,
+      "loss": 0.4059,
+      "step": 2322
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.7966219236414036e-06,
+      "loss": 0.4175,
+      "step": 2323
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.788732495896358e-06,
+      "loss": 0.416,
+      "step": 2324
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7808493576226866e-06,
+      "loss": 0.432,
+      "step": 2325
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.772972516802785e-06,
+      "loss": 0.4241,
+      "step": 2326
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.4395,
+      "step": 2327
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.75723775942196e-06,
+      "loss": 0.4443,
+      "step": 2328
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7493798587939067e-06,
+      "loss": 0.3901,
+      "step": 2329
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7415282874853443e-06,
+      "loss": 0.4276,
+      "step": 2330
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.733683053446694e-06,
+      "loss": 0.4304,
+      "step": 2331
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.72584416462197e-06,
+      "loss": 0.4098,
+      "step": 2332
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7180116289487587e-06,
+      "loss": 0.424,
+      "step": 2333
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 3.7101854543582107e-06,
+      "loss": 0.4303,
+      "step": 2334
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.702365648775039e-06,
+      "loss": 0.4054,
+      "step": 2335
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6945522201175043e-06,
+      "loss": 0.4358,
+      "step": 2336
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6867451762974117e-06,
+      "loss": 0.4316,
+      "step": 2337
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6789445252201008e-06,
+      "loss": 0.4224,
+      "step": 2338
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6711502747844384e-06,
+      "loss": 0.4231,
+      "step": 2339
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6633624328828087e-06,
+      "loss": 0.4332,
+      "step": 2340
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.655581007401108e-06,
+      "loss": 0.4089,
+      "step": 2341
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.647806006218735e-06,
+      "loss": 0.4027,
+      "step": 2342
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6400374372085866e-06,
+      "loss": 0.4249,
+      "step": 2343
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.6322753082370364e-06,
+      "loss": 0.4125,
+      "step": 2344
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.624519627163946e-06,
+      "loss": 0.4102,
+      "step": 2345
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.6167704018426474e-06,
+      "loss": 0.4319,
+      "step": 2346
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.609027640119932e-06,
+      "loss": 0.4016,
+      "step": 2347
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.601291349836049e-06,
+      "loss": 0.4357,
+      "step": 2348
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5935615388246927e-06,
+      "loss": 0.4124,
+      "step": 2349
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.585838214913001e-06,
+      "loss": 0.4423,
+      "step": 2350
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5781213859215334e-06,
+      "loss": 0.4237,
+      "step": 2351
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5704110596642815e-06,
+      "loss": 0.4419,
+      "step": 2352
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5627072439486497e-06,
+      "loss": 0.4296,
+      "step": 2353
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5550099465754493e-06,
+      "loss": 0.3984,
+      "step": 2354
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.5473191753388925e-06,
+      "loss": 0.4274,
+      "step": 2355
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.539634938026585e-06,
+      "loss": 0.4338,
+      "step": 2356
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5319572424195082e-06,
+      "loss": 0.4275,
+      "step": 2357
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5242860962920254e-06,
+      "loss": 0.4448,
+      "step": 2358
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5166215074118703e-06,
+      "loss": 0.4197,
+      "step": 2359
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.508963483540132e-06,
+      "loss": 0.4247,
+      "step": 2360
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.5013120324312544e-06,
+      "loss": 0.4166,
+      "step": 2361
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4936671618330263e-06,
+      "loss": 0.4174,
+      "step": 2362
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4860288794865747e-06,
+      "loss": 0.4447,
+      "step": 2363
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.478397193126346e-06,
+      "loss": 0.4319,
+      "step": 2364
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4707721104801175e-06,
+      "loss": 0.4219,
+      "step": 2365
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.463153639268977e-06,
+      "loss": 0.42,
+      "step": 2366
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.4555417872073184e-06,
+      "loss": 0.4091,
+      "step": 2367
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4479365620028294e-06,
+      "loss": 0.4259,
+      "step": 2368
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.440337971356492e-06,
+      "loss": 0.4159,
+      "step": 2369
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4327460229625666e-06,
+      "loss": 0.424,
+      "step": 2370
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4251607245085904e-06,
+      "loss": 0.4026,
+      "step": 2371
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.417582083675365e-06,
+      "loss": 0.438,
+      "step": 2372
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4100101081369498e-06,
+      "loss": 0.4337,
+      "step": 2373
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.4024448055606596e-06,
+      "loss": 0.4491,
+      "step": 2374
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3948861836070468e-06,
+      "loss": 0.4136,
+      "step": 2375
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3873342499299043e-06,
+      "loss": 0.4261,
+      "step": 2376
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.3797890121762444e-06,
+      "loss": 0.4037,
+      "step": 2377
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3722504779863065e-06,
+      "loss": 0.4214,
+      "step": 2378
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3647186549935407e-06,
+      "loss": 0.4211,
+      "step": 2379
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3571935508245988e-06,
+      "loss": 0.4312,
+      "step": 2380
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3496751730993327e-06,
+      "loss": 0.3885,
+      "step": 2381
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3421635294307807e-06,
+      "loss": 0.4388,
+      "step": 2382
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.334658627425167e-06,
+      "loss": 0.422,
+      "step": 2383
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3271604746818764e-06,
+      "loss": 0.4338,
+      "step": 2384
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.319669078793474e-06,
+      "loss": 0.3994,
+      "step": 2385
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.3121844473456757e-06,
+      "loss": 0.4242,
+      "step": 2386
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.304706587917349e-06,
+      "loss": 0.4026,
+      "step": 2387
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.2972355080805063e-06,
+      "loss": 0.4335,
+      "step": 2388
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2897712154002948e-06,
+      "loss": 0.4464,
+      "step": 2389
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2823137174349806e-06,
+      "loss": 0.4149,
+      "step": 2390
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2748630217359613e-06,
+      "loss": 0.4005,
+      "step": 2391
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2674191358477414e-06,
+      "loss": 0.408,
+      "step": 2392
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2599820673079286e-06,
+      "loss": 0.431,
+      "step": 2393
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2525518236472296e-06,
+      "loss": 0.4294,
+      "step": 2394
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.24512841238944e-06,
+      "loss": 0.4044,
+      "step": 2395
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2377118410514407e-06,
+      "loss": 0.4136,
+      "step": 2396
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2303021171431737e-06,
+      "loss": 0.4254,
+      "step": 2397
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.222899248167661e-06,
+      "loss": 0.4075,
+      "step": 2398
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.2155032416209787e-06,
+      "loss": 0.4244,
+      "step": 2399
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2081141049922535e-06,
+      "loss": 0.4221,
+      "step": 2400
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.2007318457636547e-06,
+      "loss": 0.4332,
+      "step": 2401
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1933564714103917e-06,
+      "loss": 0.4146,
+      "step": 2402
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.185987989400696e-06,
+      "loss": 0.4266,
+      "step": 2403
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1786264071958274e-06,
+      "loss": 0.4148,
+      "step": 2404
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1712717322500517e-06,
+      "loss": 0.4182,
+      "step": 2405
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1639239720106464e-06,
+      "loss": 0.4254,
+      "step": 2406
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1565831339178844e-06,
+      "loss": 0.4094,
+      "step": 2407
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1492492254050313e-06,
+      "loss": 0.41,
+      "step": 2408
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.1419222538983298e-06,
+      "loss": 0.4319,
+      "step": 2409
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.134602226817005e-06,
+      "loss": 0.4169,
+      "step": 2410
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.127289151573247e-06,
+      "loss": 0.4058,
+      "step": 2411
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1199830355722095e-06,
+      "loss": 0.3894,
+      "step": 2412
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.112683886211996e-06,
+      "loss": 0.4213,
+      "step": 2413
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.1053917108836563e-06,
+      "loss": 0.4147,
+      "step": 2414
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.09810651697118e-06,
+      "loss": 0.4153,
+      "step": 2415
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.090828311851489e-06,
+      "loss": 0.402,
+      "step": 2416
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.083557102894419e-06,
+      "loss": 0.4152,
+      "step": 2417
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0762928974627325e-06,
+      "loss": 0.4388,
+      "step": 2418
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0690357029120955e-06,
+      "loss": 0.4257,
+      "step": 2419
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.0617855265910734e-06,
+      "loss": 0.4171,
+      "step": 2420
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0545423758411298e-06,
+      "loss": 0.437,
+      "step": 2421
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.047306257996614e-06,
+      "loss": 0.4192,
+      "step": 2422
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.040077180384743e-06,
+      "loss": 0.4575,
+      "step": 2423
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.032855150325619e-06,
+      "loss": 0.4196,
+      "step": 2424
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0256401751321994e-06,
+      "loss": 0.4009,
+      "step": 2425
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0184322621103024e-06,
+      "loss": 0.3933,
+      "step": 2426
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0112314185585923e-06,
+      "loss": 0.4158,
+      "step": 2427
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.0040376517685764e-06,
+      "loss": 0.416,
+      "step": 2428
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.9968509690245995e-06,
+      "loss": 0.4495,
+      "step": 2429
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.989671377603822e-06,
+      "loss": 0.4041,
+      "step": 2430
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.982498884776235e-06,
+      "loss": 0.4364,
+      "step": 2431
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.975333497804637e-06,
+      "loss": 0.413,
+      "step": 2432
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9681752239446317e-06,
+      "loss": 0.4355,
+      "step": 2433
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9610240704446203e-06,
+      "loss": 0.4351,
+      "step": 2434
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.953880044545795e-06,
+      "loss": 0.4021,
+      "step": 2435
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.946743153482128e-06,
+      "loss": 0.4286,
+      "step": 2436
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9396134044803694e-06,
+      "loss": 0.4265,
+      "step": 2437
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9324908047600363e-06,
+      "loss": 0.4131,
+      "step": 2438
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.925375361533409e-06,
+      "loss": 0.4275,
+      "step": 2439
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9182670820055136e-06,
+      "loss": 0.4384,
+      "step": 2440
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9111659733741305e-06,
+      "loss": 0.4191,
+      "step": 2441
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.9040720428297754e-06,
+      "loss": 0.4276,
+      "step": 2442
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8969852975556954e-06,
+      "loss": 0.4333,
+      "step": 2443
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.889905744727863e-06,
+      "loss": 0.4266,
+      "step": 2444
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8828333915149675e-06,
+      "loss": 0.4409,
+      "step": 2445
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8757682450784073e-06,
+      "loss": 0.4395,
+      "step": 2446
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.868710312572285e-06,
+      "loss": 0.4513,
+      "step": 2447
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8616596011433937e-06,
+      "loss": 0.4008,
+      "step": 2448
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8546161179312247e-06,
+      "loss": 0.4312,
+      "step": 2449
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8475798700679347e-06,
+      "loss": 0.4182,
+      "step": 2450
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8405508646783654e-06,
+      "loss": 0.4281,
+      "step": 2451
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.8335291088800234e-06,
+      "loss": 0.4172,
+      "step": 2452
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.826514609783072e-06,
+      "loss": 0.3909,
+      "step": 2453
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.819507374490328e-06,
+      "loss": 0.4153,
+      "step": 2454
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8125074100972514e-06,
+      "loss": 0.4194,
+      "step": 2455
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.8055147236919444e-06,
+      "loss": 0.4362,
+      "step": 2456
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.798529322355129e-06,
+      "loss": 0.4014,
+      "step": 2457
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7915512131601617e-06,
+      "loss": 0.4274,
+      "step": 2458
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7845804031730096e-06,
+      "loss": 0.4102,
+      "step": 2459
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.777616899452249e-06,
+      "loss": 0.4274,
+      "step": 2460
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.770660709049061e-06,
+      "loss": 0.4026,
+      "step": 2461
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7637118390072194e-06,
+      "loss": 0.4453,
+      "step": 2462
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.7567702963630805e-06,
+      "loss": 0.4008,
+      "step": 2463
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7498360881455876e-06,
+      "loss": 0.4143,
+      "step": 2464
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7429092213762553e-06,
+      "loss": 0.4163,
+      "step": 2465
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.735989703069165e-06,
+      "loss": 0.4157,
+      "step": 2466
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7290775402309542e-06,
+      "loss": 0.428,
+      "step": 2467
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7221727398608166e-06,
+      "loss": 0.4243,
+      "step": 2468
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.715275308950488e-06,
+      "loss": 0.4312,
+      "step": 2469
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.708385254484244e-06,
+      "loss": 0.4451,
+      "step": 2470
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.7015025834388842e-06,
+      "loss": 0.4242,
+      "step": 2471
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.69462730278374e-06,
+      "loss": 0.406,
+      "step": 2472
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.6877594194806567e-06,
+      "loss": 0.4573,
+      "step": 2473
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.680898940483988e-06,
+      "loss": 0.4342,
+      "step": 2474
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6740458727405907e-06,
+      "loss": 0.4343,
+      "step": 2475
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6672002231898177e-06,
+      "loss": 0.4392,
+      "step": 2476
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6603619987635087e-06,
+      "loss": 0.4203,
+      "step": 2477
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6535312063859874e-06,
+      "loss": 0.415,
+      "step": 2478
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6467078529740486e-06,
+      "loss": 0.4453,
+      "step": 2479
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6398919454369566e-06,
+      "loss": 0.4413,
+      "step": 2480
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.633083490676437e-06,
+      "loss": 0.4111,
+      "step": 2481
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6262824955866684e-06,
+      "loss": 0.41,
+      "step": 2482
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.6194889670542702e-06,
+      "loss": 0.422,
+      "step": 2483
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.612702911958308e-06,
+      "loss": 0.4303,
+      "step": 2484
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.605924337170278e-06,
+      "loss": 0.4182,
+      "step": 2485
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5991532495541016e-06,
+      "loss": 0.4286,
+      "step": 2486
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.592389655966119e-06,
+      "loss": 0.4363,
+      "step": 2487
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.585633563255082e-06,
+      "loss": 0.3917,
+      "step": 2488
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5788849782621494e-06,
+      "loss": 0.4396,
+      "step": 2489
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5721439078208686e-06,
+      "loss": 0.4246,
+      "step": 2490
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5654103587571887e-06,
+      "loss": 0.4576,
+      "step": 2491
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.558684337889439e-06,
+      "loss": 0.4003,
+      "step": 2492
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.551965852028323e-06,
+      "loss": 0.4219,
+      "step": 2493
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5452549079769195e-06,
+      "loss": 0.4264,
+      "step": 2494
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.5385515125306683e-06,
+      "loss": 0.4158,
+      "step": 2495
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5318556724773602e-06,
+      "loss": 0.4248,
+      "step": 2496
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.525167394597142e-06,
+      "loss": 0.4149,
+      "step": 2497
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.518486685662502e-06,
+      "loss": 0.406,
+      "step": 2498
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.511813552438265e-06,
+      "loss": 0.4125,
+      "step": 2499
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.5051480016815823e-06,
+      "loss": 0.418,
+      "step": 2500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4984900401419265e-06,
+      "loss": 0.4058,
+      "step": 2501
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4918396745610917e-06,
+      "loss": 0.4137,
+      "step": 2502
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4851969116731712e-06,
+      "loss": 0.3995,
+      "step": 2503
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.478561758204565e-06,
+      "loss": 0.4308,
+      "step": 2504
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4719342208739695e-06,
+      "loss": 0.4421,
+      "step": 2505
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.4653143063923647e-06,
+      "loss": 0.4171,
+      "step": 2506
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4587020214630165e-06,
+      "loss": 0.423,
+      "step": 2507
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4520973727814624e-06,
+      "loss": 0.4025,
+      "step": 2508
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.445500367035506e-06,
+      "loss": 0.4337,
+      "step": 2509
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.438911010905216e-06,
+      "loss": 0.4182,
+      "step": 2510
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.432329311062911e-06,
+      "loss": 0.4321,
+      "step": 2511
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4257552741731593e-06,
+      "loss": 0.4205,
+      "step": 2512
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4191889068927688e-06,
+      "loss": 0.4015,
+      "step": 2513
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.4126302158707825e-06,
+      "loss": 0.4017,
+      "step": 2514
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.406079207748473e-06,
+      "loss": 0.4181,
+      "step": 2515
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.399535889159321e-06,
+      "loss": 0.45,
+      "step": 2516
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.3930002667290363e-06,
+      "loss": 0.4198,
+      "step": 2517
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.386472347075527e-06,
+      "loss": 0.4053,
+      "step": 2518
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.379952136808903e-06,
+      "loss": 0.4363,
+      "step": 2519
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.37343964253147e-06,
+      "loss": 0.4463,
+      "step": 2520
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.366934870837717e-06,
+      "loss": 0.4221,
+      "step": 2521
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3604378283143194e-06,
+      "loss": 0.4175,
+      "step": 2522
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.353948521540115e-06,
+      "loss": 0.4407,
+      "step": 2523
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3474669570861185e-06,
+      "loss": 0.3888,
+      "step": 2524
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3409931415155006e-06,
+      "loss": 0.4202,
+      "step": 2525
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3345270813835883e-06,
+      "loss": 0.424,
+      "step": 2526
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3280687832378512e-06,
+      "loss": 0.4148,
+      "step": 2527
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.3216182536179054e-06,
+      "loss": 0.4357,
+      "step": 2528
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.315175499055492e-06,
+      "loss": 0.423,
+      "step": 2529
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3087405260744854e-06,
+      "loss": 0.4187,
+      "step": 2530
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.3023133411908805e-06,
+      "loss": 0.4222,
+      "step": 2531
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.295893950912784e-06,
+      "loss": 0.4242,
+      "step": 2532
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.2894823617404107e-06,
+      "loss": 0.4298,
+      "step": 2533
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.283078580166074e-06,
+      "loss": 0.433,
+      "step": 2534
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.276682612674188e-06,
+      "loss": 0.4437,
+      "step": 2535
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.270294465741244e-06,
+      "loss": 0.4229,
+      "step": 2536
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.263914145835823e-06,
+      "loss": 0.4034,
+      "step": 2537
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.257541659418574e-06,
+      "loss": 0.4095,
+      "step": 2538
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2511770129422206e-06,
+      "loss": 0.4346,
+      "step": 2539
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.244820212851544e-06,
+      "loss": 0.4418,
+      "step": 2540
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2384712655833784e-06,
+      "loss": 0.4335,
+      "step": 2541
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2321301775666103e-06,
+      "loss": 0.4266,
+      "step": 2542
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2257969552221657e-06,
+      "loss": 0.4384,
+      "step": 2543
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.219471604963006e-06,
+      "loss": 0.4292,
+      "step": 2544
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.213154133194122e-06,
+      "loss": 0.4409,
+      "step": 2545
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.2068445463125253e-06,
+      "loss": 0.4121,
+      "step": 2546
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.200542850707247e-06,
+      "loss": 0.4181,
+      "step": 2547
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1942490527593254e-06,
+      "loss": 0.4347,
+      "step": 2548
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.187963158841797e-06,
+      "loss": 0.4247,
+      "step": 2549
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1816851753197023e-06,
+      "loss": 0.4349,
+      "step": 2550
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1754151085500684e-06,
+      "loss": 0.4504,
+      "step": 2551
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1691529648819065e-06,
+      "loss": 0.4056,
+      "step": 2552
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1628987506562036e-06,
+      "loss": 0.4153,
+      "step": 2553
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.15665247220592e-06,
+      "loss": 0.4089,
+      "step": 2554
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1504141358559818e-06,
+      "loss": 0.4247,
+      "step": 2555
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.144183747923262e-06,
+      "loss": 0.427,
+      "step": 2556
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.1379613147165977e-06,
+      "loss": 0.4106,
+      "step": 2557
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.131746842536767e-06,
+      "loss": 0.4252,
+      "step": 2558
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.125540337676486e-06,
+      "loss": 0.4194,
+      "step": 2559
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.119341806420402e-06,
+      "loss": 0.4159,
+      "step": 2560
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.113151255045095e-06,
+      "loss": 0.4043,
+      "step": 2561
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.106968689819051e-06,
+      "loss": 0.427,
+      "step": 2562
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.1007941170026834e-06,
+      "loss": 0.4155,
+      "step": 2563
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.094627542848305e-06,
+      "loss": 0.442,
+      "step": 2564
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0884689736001317e-06,
+      "loss": 0.4309,
+      "step": 2565
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0823184154942733e-06,
+      "loss": 0.4275,
+      "step": 2566
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.076175874758728e-06,
+      "loss": 0.4414,
+      "step": 2567
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.070041357613376e-06,
+      "loss": 0.4313,
+      "step": 2568
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0639148702699674e-06,
+      "loss": 0.4369,
+      "step": 2569
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0577964189321288e-06,
+      "loss": 0.4111,
+      "step": 2570
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.051686009795347e-06,
+      "loss": 0.4128,
+      "step": 2571
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.045583649046965e-06,
+      "loss": 0.4078,
+      "step": 2572
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0394893428661765e-06,
+      "loss": 0.4238,
+      "step": 2573
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0334030974240183e-06,
+      "loss": 0.4092,
+      "step": 2574
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0273249188833656e-06,
+      "loss": 0.4077,
+      "step": 2575
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.021254813398926e-06,
+      "loss": 0.4076,
+      "step": 2576
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0151927871172306e-06,
+      "loss": 0.4434,
+      "step": 2577
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0091388461766314e-06,
+      "loss": 0.4272,
+      "step": 2578
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.003092996707292e-06,
+      "loss": 0.4412,
+      "step": 2579
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9970552448311818e-06,
+      "loss": 0.3904,
+      "step": 2580
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.9910255966620763e-06,
+      "loss": 0.425,
+      "step": 2581
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985004058305535e-06,
+      "loss": 0.3917,
+      "step": 2582
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.978990635858914e-06,
+      "loss": 0.4052,
+      "step": 2583
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9729853354113483e-06,
+      "loss": 0.4137,
+      "step": 2584
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9669881630437483e-06,
+      "loss": 0.4323,
+      "step": 2585
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9609991248287965e-06,
+      "loss": 0.4147,
+      "step": 2586
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.955018226830935e-06,
+      "loss": 0.423,
+      "step": 2587
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9490454751063693e-06,
+      "loss": 0.4114,
+      "step": 2588
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9430808757030452e-06,
+      "loss": 0.4132,
+      "step": 2589
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9371244346606645e-06,
+      "loss": 0.4259,
+      "step": 2590
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9311761580106637e-06,
+      "loss": 0.4266,
+      "step": 2591
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9252360517762102e-06,
+      "loss": 0.4275,
+      "step": 2592
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.919304121972202e-06,
+      "loss": 0.4301,
+      "step": 2593
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.913380374605257e-06,
+      "loss": 0.4392,
+      "step": 2594
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.907464815673702e-06,
+      "loss": 0.401,
+      "step": 2595
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9015574511675783e-06,
+      "loss": 0.4237,
+      "step": 2596
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.895658287068629e-06,
+      "loss": 0.4203,
+      "step": 2597
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8897673293502916e-06,
+      "loss": 0.4214,
+      "step": 2598
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8838845839776954e-06,
+      "loss": 0.3966,
+      "step": 2599
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.878010056907653e-06,
+      "loss": 0.4073,
+      "step": 2600
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.8721437540886589e-06,
+      "loss": 0.4384,
+      "step": 2601
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.866285681460871e-06,
+      "loss": 0.4097,
+      "step": 2602
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.860435844956121e-06,
+      "loss": 0.449,
+      "step": 2603
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8545942504978975e-06,
+      "loss": 0.4158,
+      "step": 2604
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8487609040013466e-06,
+      "loss": 0.4395,
+      "step": 2605
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8429358113732598e-06,
+      "loss": 0.4229,
+      "step": 2606
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8371189785120714e-06,
+      "loss": 0.4303,
+      "step": 2607
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8313104113078506e-06,
+      "loss": 0.4475,
+      "step": 2608
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8255101156422995e-06,
+      "loss": 0.4237,
+      "step": 2609
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.819718097388743e-06,
+      "loss": 0.4247,
+      "step": 2610
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8139343624121241e-06,
+      "loss": 0.4265,
+      "step": 2611
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.8081589165690005e-06,
+      "loss": 0.4326,
+      "step": 2612
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.802391765707533e-06,
+      "loss": 0.4015,
+      "step": 2613
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7966329156674867e-06,
+      "loss": 0.4336,
+      "step": 2614
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.790882372280216e-06,
+      "loss": 0.4329,
+      "step": 2615
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7851401413686686e-06,
+      "loss": 0.438,
+      "step": 2616
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7794062287473734e-06,
+      "loss": 0.4146,
+      "step": 2617
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7736806402224382e-06,
+      "loss": 0.3936,
+      "step": 2618
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7679633815915397e-06,
+      "loss": 0.4203,
+      "step": 2619
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7622544586439193e-06,
+      "loss": 0.427,
+      "step": 2620
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7565538771603841e-06,
+      "loss": 0.4307,
+      "step": 2621
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.750861642913282e-06,
+      "loss": 0.4113,
+      "step": 2622
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.74517776166652e-06,
+      "loss": 0.428,
+      "step": 2623
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7395022391755433e-06,
+      "loss": 0.4467,
+      "step": 2624
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7338350811873316e-06,
+      "loss": 0.4146,
+      "step": 2625
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7281762934403967e-06,
+      "loss": 0.427,
+      "step": 2626
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.722525881664775e-06,
+      "loss": 0.4323,
+      "step": 2627
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.716883851582023e-06,
+      "loss": 0.4266,
+      "step": 2628
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7112502089052009e-06,
+      "loss": 0.4086,
+      "step": 2629
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7056249593388863e-06,
+      "loss": 0.4281,
+      "step": 2630
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7000081085791541e-06,
+      "loss": 0.4167,
+      "step": 2631
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6943996623135761e-06,
+      "loss": 0.421,
+      "step": 2632
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6887996262212102e-06,
+      "loss": 0.4326,
+      "step": 2633
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.683208005972604e-06,
+      "loss": 0.4469,
+      "step": 2634
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.6776248072297762e-06,
+      "loss": 0.4089,
+      "step": 2635
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6720500356462221e-06,
+      "loss": 0.4371,
+      "step": 2636
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6664836968669041e-06,
+      "loss": 0.3944,
+      "step": 2637
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6609257965282456e-06,
+      "loss": 0.4139,
+      "step": 2638
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6553763402581236e-06,
+      "loss": 0.4026,
+      "step": 2639
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6498353336758654e-06,
+      "loss": 0.4069,
+      "step": 2640
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.644302782392243e-06,
+      "loss": 0.4523,
+      "step": 2641
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.638778692009465e-06,
+      "loss": 0.4026,
+      "step": 2642
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6332630681211759e-06,
+      "loss": 0.4186,
+      "step": 2643
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6277559163124424e-06,
+      "loss": 0.4131,
+      "step": 2644
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.622257242159756e-06,
+      "loss": 0.4379,
+      "step": 2645
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.6167670512310229e-06,
+      "loss": 0.4299,
+      "step": 2646
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6112853490855618e-06,
+      "loss": 0.4332,
+      "step": 2647
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6058121412740878e-06,
+      "loss": 0.4128,
+      "step": 2648
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.6003474333387214e-06,
+      "loss": 0.4137,
+      "step": 2649
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.594891230812976e-06,
+      "loss": 0.4347,
+      "step": 2650
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5894435392217512e-06,
+      "loss": 0.4113,
+      "step": 2651
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5840043640813274e-06,
+      "loss": 0.413,
+      "step": 2652
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5785737108993636e-06,
+      "loss": 0.4137,
+      "step": 2653
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5731515851748901e-06,
+      "loss": 0.4303,
+      "step": 2654
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.567737992398296e-06,
+      "loss": 0.4132,
+      "step": 2655
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.5623329380513386e-06,
+      "loss": 0.4286,
+      "step": 2656
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5569364276071231e-06,
+      "loss": 0.4136,
+      "step": 2657
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5515484665301062e-06,
+      "loss": 0.418,
+      "step": 2658
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5461690602760882e-06,
+      "loss": 0.3882,
+      "step": 2659
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.540798214292204e-06,
+      "loss": 0.4475,
+      "step": 2660
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5354359340169244e-06,
+      "loss": 0.4104,
+      "step": 2661
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5300822248800406e-06,
+      "loss": 0.4207,
+      "step": 2662
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5247370923026706e-06,
+      "loss": 0.4478,
+      "step": 2663
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5194005416972457e-06,
+      "loss": 0.4266,
+      "step": 2664
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.514072578467506e-06,
+      "loss": 0.4088,
+      "step": 2665
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.5087532080084976e-06,
+      "loss": 0.422,
+      "step": 2666
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.503442435706568e-06,
+      "loss": 0.4227,
+      "step": 2667
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4981402669393507e-06,
+      "loss": 0.4191,
+      "step": 2668
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4928467070757746e-06,
+      "loss": 0.4199,
+      "step": 2669
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4875617614760495e-06,
+      "loss": 0.4316,
+      "step": 2670
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4822854354916605e-06,
+      "loss": 0.4459,
+      "step": 2671
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.477017734465368e-06,
+      "loss": 0.407,
+      "step": 2672
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4717586637311943e-06,
+      "loss": 0.4363,
+      "step": 2673
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4665082286144272e-06,
+      "loss": 0.4337,
+      "step": 2674
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4612664344316074e-06,
+      "loss": 0.443,
+      "step": 2675
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.456033286490527e-06,
+      "loss": 0.4282,
+      "step": 2676
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4508087900902211e-06,
+      "loss": 0.4401,
+      "step": 2677
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.4455929505209687e-06,
+      "loss": 0.4495,
+      "step": 2678
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.440385773064278e-06,
+      "loss": 0.3954,
+      "step": 2679
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.435187262992891e-06,
+      "loss": 0.4088,
+      "step": 2680
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4299974255707661e-06,
+      "loss": 0.41,
+      "step": 2681
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4248162660530861e-06,
+      "loss": 0.4183,
+      "step": 2682
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4196437896862448e-06,
+      "loss": 0.4297,
+      "step": 2683
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4144800017078431e-06,
+      "loss": 0.432,
+      "step": 2684
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4093249073466853e-06,
+      "loss": 0.4208,
+      "step": 2685
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.4041785118227702e-06,
+      "loss": 0.417,
+      "step": 2686
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3990408203472938e-06,
+      "loss": 0.4117,
+      "step": 2687
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.393911838122629e-06,
+      "loss": 0.4651,
+      "step": 2688
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.3887915703423371e-06,
+      "loss": 0.4203,
+      "step": 2689
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3836800221911539e-06,
+      "loss": 0.4081,
+      "step": 2690
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3785771988449847e-06,
+      "loss": 0.4125,
+      "step": 2691
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.373483105470902e-06,
+      "loss": 0.4072,
+      "step": 2692
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.368397747227136e-06,
+      "loss": 0.4106,
+      "step": 2693
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3633211292630743e-06,
+      "loss": 0.4091,
+      "step": 2694
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3582532567192509e-06,
+      "loss": 0.3911,
+      "step": 2695
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3531941347273458e-06,
+      "loss": 0.4296,
+      "step": 2696
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3481437684101795e-06,
+      "loss": 0.4111,
+      "step": 2697
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3431021628817066e-06,
+      "loss": 0.4019,
+      "step": 2698
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.3380693232470087e-06,
+      "loss": 0.4208,
+      "step": 2699
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3330452546022942e-06,
+      "loss": 0.4042,
+      "step": 2700
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3280299620348847e-06,
+      "loss": 0.4164,
+      "step": 2701
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3230234506232198e-06,
+      "loss": 0.4111,
+      "step": 2702
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3180257254368456e-06,
+      "loss": 0.4258,
+      "step": 2703
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3130367915364127e-06,
+      "loss": 0.4215,
+      "step": 2704
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.3080566539736695e-06,
+      "loss": 0.4142,
+      "step": 2705
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.303085317791456e-06,
+      "loss": 0.4363,
+      "step": 2706
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2981227880237002e-06,
+      "loss": 0.415,
+      "step": 2707
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2931690696954135e-06,
+      "loss": 0.4336,
+      "step": 2708
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.288224167822686e-06,
+      "loss": 0.4469,
+      "step": 2709
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.2832880874126784e-06,
+      "loss": 0.4066,
+      "step": 2710
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2783608334636188e-06,
+      "loss": 0.4159,
+      "step": 2711
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2734424109647992e-06,
+      "loss": 0.4254,
+      "step": 2712
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.26853282489657e-06,
+      "loss": 0.3867,
+      "step": 2713
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2636320802303281e-06,
+      "loss": 0.4385,
+      "step": 2714
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.258740181928524e-06,
+      "loss": 0.4015,
+      "step": 2715
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.253857134944647e-06,
+      "loss": 0.4338,
+      "step": 2716
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.248982944223226e-06,
+      "loss": 0.4354,
+      "step": 2717
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2441176146998212e-06,
+      "loss": 0.4203,
+      "step": 2718
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.239261151301019e-06,
+      "loss": 0.4106,
+      "step": 2719
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.2344135589444316e-06,
+      "loss": 0.4011,
+      "step": 2720
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.229574842538681e-06,
+      "loss": 0.437,
+      "step": 2721
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2247450069834077e-06,
+      "loss": 0.4155,
+      "step": 2722
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2199240571692582e-06,
+      "loss": 0.4453,
+      "step": 2723
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2151119979778824e-06,
+      "loss": 0.4375,
+      "step": 2724
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2103088342819258e-06,
+      "loss": 0.4049,
+      "step": 2725
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2055145709450267e-06,
+      "loss": 0.4233,
+      "step": 2726
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.2007292128218128e-06,
+      "loss": 0.4082,
+      "step": 2727
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1959527647578895e-06,
+      "loss": 0.422,
+      "step": 2728
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1911852315898465e-06,
+      "loss": 0.3933,
+      "step": 2729
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1864266181452422e-06,
+      "loss": 0.4031,
+      "step": 2730
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1816769292426033e-06,
+      "loss": 0.4125,
+      "step": 2731
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.1769361696914217e-06,
+      "loss": 0.4231,
+      "step": 2732
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1722043442921483e-06,
+      "loss": 0.4261,
+      "step": 2733
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1674814578361804e-06,
+      "loss": 0.4059,
+      "step": 2734
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1627675151058703e-06,
+      "loss": 0.4134,
+      "step": 2735
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1580625208745145e-06,
+      "loss": 0.4386,
+      "step": 2736
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1533664799063427e-06,
+      "loss": 0.4341,
+      "step": 2737
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1486793969565246e-06,
+      "loss": 0.4266,
+      "step": 2738
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.144001276771156e-06,
+      "loss": 0.4296,
+      "step": 2739
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1393321240872578e-06,
+      "loss": 0.4147,
+      "step": 2740
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1346719436327701e-06,
+      "loss": 0.4197,
+      "step": 2741
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.1300207401265485e-06,
+      "loss": 0.4139,
+      "step": 2742
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1253785182783571e-06,
+      "loss": 0.4244,
+      "step": 2743
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1207452827888664e-06,
+      "loss": 0.4332,
+      "step": 2744
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1161210383496479e-06,
+      "loss": 0.4281,
+      "step": 2745
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.111505789643169e-06,
+      "loss": 0.4408,
+      "step": 2746
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.1068995413427886e-06,
+      "loss": 0.4218,
+      "step": 2747
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.102302298112745e-06,
+      "loss": 0.4305,
+      "step": 2748
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0977140646081685e-06,
+      "loss": 0.4124,
+      "step": 2749
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0931348454750602e-06,
+      "loss": 0.4165,
+      "step": 2750
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0885646453502952e-06,
+      "loss": 0.431,
+      "step": 2751
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0840034688616142e-06,
+      "loss": 0.427,
+      "step": 2752
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0794513206276237e-06,
+      "loss": 0.434,
+      "step": 2753
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0749082052577842e-06,
+      "loss": 0.4227,
+      "step": 2754
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0703741273524127e-06,
+      "loss": 0.444,
+      "step": 2755
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.065849091502673e-06,
+      "loss": 0.4514,
+      "step": 2756
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.061333102290576e-06,
+      "loss": 0.4084,
+      "step": 2757
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.05682616428897e-06,
+      "loss": 0.4015,
+      "step": 2758
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0523282820615365e-06,
+      "loss": 0.4235,
+      "step": 2759
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.047839460162793e-06,
+      "loss": 0.4126,
+      "step": 2760
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0433597031380726e-06,
+      "loss": 0.4006,
+      "step": 2761
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0388890155235398e-06,
+      "loss": 0.4041,
+      "step": 2762
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0344274018461687e-06,
+      "loss": 0.4017,
+      "step": 2763
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0299748666237485e-06,
+      "loss": 0.4088,
+      "step": 2764
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0255314143648753e-06,
+      "loss": 0.426,
+      "step": 2765
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0210970495689487e-06,
+      "loss": 0.4327,
+      "step": 2766
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0166717767261614e-06,
+      "loss": 0.4136,
+      "step": 2767
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0122556003175043e-06,
+      "loss": 0.439,
+      "step": 2768
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.0078485248147573e-06,
+      "loss": 0.4152,
+      "step": 2769
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.003450554680484e-06,
+      "loss": 0.4117,
+      "step": 2770
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.990616943680266e-07,
+      "loss": 0.4042,
+      "step": 2771
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.946819483215053e-07,
+      "loss": 0.4152,
+      "step": 2772
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.4174,
+      "step": 2773
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.85949816756595e-07,
+      "loss": 0.4182,
+      "step": 2774
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.815974400802808e-07,
+      "loss": 0.437,
+      "step": 2775
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.772541953540427e-07,
+      "loss": 0.4138,
+      "step": 2776
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.729200869758104e-07,
+      "loss": 0.4622,
+      "step": 2777
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.685951193342603e-07,
+      "loss": 0.4089,
+      "step": 2778
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.642792968088143e-07,
+      "loss": 0.4237,
+      "step": 2779
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.59972623769636e-07,
+      "loss": 0.4396,
+      "step": 2780
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.556751045776192e-07,
+      "loss": 0.4326,
+      "step": 2781
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.513867435843927e-07,
+      "loss": 0.4084,
+      "step": 2782
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.471075451323108e-07,
+      "loss": 0.4313,
+      "step": 2783
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.428375135544499e-07,
+      "loss": 0.4325,
+      "step": 2784
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.385766531746055e-07,
+      "loss": 0.4196,
+      "step": 2785
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.343249683072841e-07,
+      "loss": 0.4116,
+      "step": 2786
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.300824632576999e-07,
+      "loss": 0.3918,
+      "step": 2787
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.258491423217753e-07,
+      "loss": 0.4319,
+      "step": 2788
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.216250097861323e-07,
+      "loss": 0.4147,
+      "step": 2789
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.174100699280864e-07,
+      "loss": 0.4104,
+      "step": 2790
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.132043270156477e-07,
+      "loss": 0.3983,
+      "step": 2791
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.090077853075119e-07,
+      "loss": 0.4447,
+      "step": 2792
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.048204490530599e-07,
+      "loss": 0.4093,
+      "step": 2793
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.006423224923444e-07,
+      "loss": 0.432,
+      "step": 2794
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.964734098561001e-07,
+      "loss": 0.4019,
+      "step": 2795
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 8.92313715365728e-07,
+      "loss": 0.4026,
+      "step": 2796
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.881632432332976e-07,
+      "loss": 0.4303,
+      "step": 2797
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.840219976615361e-07,
+      "loss": 0.4156,
+      "step": 2798
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.798899828438334e-07,
+      "loss": 0.4159,
+      "step": 2799
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.75767202964224e-07,
+      "loss": 0.4243,
+      "step": 2800
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.716536621973992e-07,
+      "loss": 0.452,
+      "step": 2801
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.675493647086919e-07,
+      "loss": 0.422,
+      "step": 2802
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.634543146540753e-07,
+      "loss": 0.3992,
+      "step": 2803
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.593685161801579e-07,
+      "loss": 0.4247,
+      "step": 2804
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.552919734241827e-07,
+      "loss": 0.4266,
+      "step": 2805
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.512246905140164e-07,
+      "loss": 0.4028,
+      "step": 2806
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 8.471666715681537e-07,
+      "loss": 0.4426,
+      "step": 2807
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.431179206957052e-07,
+      "loss": 0.4093,
+      "step": 2808
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.390784419963993e-07,
+      "loss": 0.402,
+      "step": 2809
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.350482395605742e-07,
+      "loss": 0.4176,
+      "step": 2810
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.31027317469174e-07,
+      "loss": 0.4276,
+      "step": 2811
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.270156797937479e-07,
+      "loss": 0.3983,
+      "step": 2812
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.23013330596445e-07,
+      "loss": 0.4314,
+      "step": 2813
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.190202739300013e-07,
+      "loss": 0.4122,
+      "step": 2814
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.150365138377514e-07,
+      "loss": 0.4315,
+      "step": 2815
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.110620543536141e-07,
+      "loss": 0.4246,
+      "step": 2816
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 8.0709689950209e-07,
+      "loss": 0.4297,
+      "step": 2817
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 8.031410532982553e-07,
+      "loss": 0.4375,
+      "step": 2818
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.991945197477679e-07,
+      "loss": 0.407,
+      "step": 2819
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.952573028468458e-07,
+      "loss": 0.4133,
+      "step": 2820
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.913294065822797e-07,
+      "loss": 0.4244,
+      "step": 2821
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.874108349314202e-07,
+      "loss": 0.4141,
+      "step": 2822
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.835015918621769e-07,
+      "loss": 0.4386,
+      "step": 2823
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.79601681333012e-07,
+      "loss": 0.4221,
+      "step": 2824
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.757111072929402e-07,
+      "loss": 0.4193,
+      "step": 2825
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.7182987368152e-07,
+      "loss": 0.4259,
+      "step": 2826
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.679579844288509e-07,
+      "loss": 0.429,
+      "step": 2827
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 7.640954434555714e-07,
+      "loss": 0.4166,
+      "step": 2828
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.602422546728561e-07,
+      "loss": 0.4128,
+      "step": 2829
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.563984219824061e-07,
+      "loss": 0.4182,
+      "step": 2830
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.525639492764525e-07,
+      "loss": 0.4143,
+      "step": 2831
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.487388404377449e-07,
+      "loss": 0.441,
+      "step": 2832
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.449230993395551e-07,
+      "loss": 0.4124,
+      "step": 2833
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.411167298456634e-07,
+      "loss": 0.4094,
+      "step": 2834
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.373197358103657e-07,
+      "loss": 0.439,
+      "step": 2835
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.335321210784618e-07,
+      "loss": 0.4043,
+      "step": 2836
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.297538894852563e-07,
+      "loss": 0.4504,
+      "step": 2837
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.259850448565508e-07,
+      "loss": 0.4116,
+      "step": 2838
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 7.222255910086418e-07,
+      "loss": 0.4088,
+      "step": 2839
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.184755317483172e-07,
+      "loss": 0.4263,
+      "step": 2840
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.147348708728508e-07,
+      "loss": 0.3893,
+      "step": 2841
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.110036121700026e-07,
+      "loss": 0.4182,
+      "step": 2842
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.072817594180103e-07,
+      "loss": 0.4144,
+      "step": 2843
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 7.03569316385585e-07,
+      "loss": 0.403,
+      "step": 2844
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.998662868319139e-07,
+      "loss": 0.4241,
+      "step": 2845
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.96172674506651e-07,
+      "loss": 0.4116,
+      "step": 2846
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.924884831499113e-07,
+      "loss": 0.4225,
+      "step": 2847
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.888137164922725e-07,
+      "loss": 0.4187,
+      "step": 2848
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.851483782547697e-07,
+      "loss": 0.4254,
+      "step": 2849
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 6.814924721488892e-07,
+      "loss": 0.4399,
+      "step": 2850
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.778460018765698e-07,
+      "loss": 0.4113,
+      "step": 2851
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.742089711301924e-07,
+      "loss": 0.3949,
+      "step": 2852
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.705813835925779e-07,
+      "loss": 0.4097,
+      "step": 2853
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.669632429369888e-07,
+      "loss": 0.4111,
+      "step": 2854
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.633545528271213e-07,
+      "loss": 0.4108,
+      "step": 2855
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.597553169171001e-07,
+      "loss": 0.4124,
+      "step": 2856
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.561655388514787e-07,
+      "loss": 0.4098,
+      "step": 2857
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.525852222652329e-07,
+      "loss": 0.4122,
+      "step": 2858
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.490143707837593e-07,
+      "loss": 0.4216,
+      "step": 2859
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 6.454529880228644e-07,
+      "loss": 0.4265,
+      "step": 2860
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.419010775887747e-07,
+      "loss": 0.4082,
+      "step": 2861
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.383586430781196e-07,
+      "loss": 0.3986,
+      "step": 2862
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.348256880779358e-07,
+      "loss": 0.429,
+      "step": 2863
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.313022161656601e-07,
+      "loss": 0.447,
+      "step": 2864
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.277882309091287e-07,
+      "loss": 0.4243,
+      "step": 2865
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.242837358665698e-07,
+      "loss": 0.4102,
+      "step": 2866
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.207887345866003e-07,
+      "loss": 0.4142,
+      "step": 2867
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.173032306082271e-07,
+      "loss": 0.4274,
+      "step": 2868
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.138272274608404e-07,
+      "loss": 0.4154,
+      "step": 2869
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.103607286642055e-07,
+      "loss": 0.4415,
+      "step": 2870
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 6.069037377284681e-07,
+      "loss": 0.427,
+      "step": 2871
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.034562581541459e-07,
+      "loss": 0.4114,
+      "step": 2872
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 6.00018293432123e-07,
+      "loss": 0.418,
+      "step": 2873
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.965898470436504e-07,
+      "loss": 0.4131,
+      "step": 2874
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.931709224603422e-07,
+      "loss": 0.3987,
+      "step": 2875
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.89761523144169e-07,
+      "loss": 0.4364,
+      "step": 2876
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.863616525474547e-07,
+      "loss": 0.4207,
+      "step": 2877
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.82971314112879e-07,
+      "loss": 0.4106,
+      "step": 2878
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.795905112734679e-07,
+      "loss": 0.4072,
+      "step": 2879
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.762192474525862e-07,
+      "loss": 0.4276,
+      "step": 2880
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.728575260639469e-07,
+      "loss": 0.4095,
+      "step": 2881
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.695053505115977e-07,
+      "loss": 0.425,
+      "step": 2882
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.661627241899193e-07,
+      "loss": 0.4108,
+      "step": 2883
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.62829650483625e-07,
+      "loss": 0.4359,
+      "step": 2884
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.595061327677543e-07,
+      "loss": 0.4056,
+      "step": 2885
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.561921744076681e-07,
+      "loss": 0.4072,
+      "step": 2886
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.528877787590514e-07,
+      "loss": 0.4286,
+      "step": 2887
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.49592949167903e-07,
+      "loss": 0.4044,
+      "step": 2888
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.463076889705377e-07,
+      "loss": 0.4236,
+      "step": 2889
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.430320014935797e-07,
+      "loss": 0.4388,
+      "step": 2890
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.397658900539571e-07,
+      "loss": 0.4213,
+      "step": 2891
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.365093579589076e-07,
+      "loss": 0.4314,
+      "step": 2892
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.332624085059612e-07,
+      "loss": 0.3965,
+      "step": 2893
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.300250449829514e-07,
+      "loss": 0.4298,
+      "step": 2894
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.267972706679991e-07,
+      "loss": 0.421,
+      "step": 2895
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.235790888295222e-07,
+      "loss": 0.431,
+      "step": 2896
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.203705027262185e-07,
+      "loss": 0.4246,
+      "step": 2897
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.171715156070734e-07,
+      "loss": 0.4218,
+      "step": 2898
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.139821307113535e-07,
+      "loss": 0.4226,
+      "step": 2899
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.108023512685967e-07,
+      "loss": 0.4167,
+      "step": 2900
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.076321804986184e-07,
+      "loss": 0.399,
+      "step": 2901
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.044716216115053e-07,
+      "loss": 0.4323,
+      "step": 2902
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.013206778076086e-07,
+      "loss": 0.4019,
+      "step": 2903
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.981793522775458e-07,
+      "loss": 0.4164,
+      "step": 2904
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.950476482021915e-07,
+      "loss": 0.4003,
+      "step": 2905
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.919255687526836e-07,
+      "loss": 0.4064,
+      "step": 2906
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.888131170904065e-07,
+      "loss": 0.4233,
+      "step": 2907
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.857102963670024e-07,
+      "loss": 0.3961,
+      "step": 2908
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.826171097243582e-07,
+      "loss": 0.4232,
+      "step": 2909
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.795335602946049e-07,
+      "loss": 0.4203,
+      "step": 2910
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.7645965120011627e-07,
+      "loss": 0.417,
+      "step": 2911
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.733953855535045e-07,
+      "loss": 0.4225,
+      "step": 2912
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.703407664576143e-07,
+      "loss": 0.4413,
+      "step": 2913
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.6729579700552695e-07,
+      "loss": 0.4287,
+      "step": 2914
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6426048028054726e-07,
+      "loss": 0.411,
+      "step": 2915
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6123481935621104e-07,
+      "loss": 0.4148,
+      "step": 2916
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5821881729627315e-07,
+      "loss": 0.4404,
+      "step": 2917
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.552124771547095e-07,
+      "loss": 0.3828,
+      "step": 2918
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.522158019757128e-07,
+      "loss": 0.4241,
+      "step": 2919
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.492287947936857e-07,
+      "loss": 0.4193,
+      "step": 2920
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.462514586332456e-07,
+      "loss": 0.4098,
+      "step": 2921
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4328379650921407e-07,
+      "loss": 0.4263,
+      "step": 2922
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.4032581142661955e-07,
+      "loss": 0.4214,
+      "step": 2923
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.3737750638068956e-07,
+      "loss": 0.4532,
+      "step": 2924
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.344388843568503e-07,
+      "loss": 0.4236,
+      "step": 2925
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.315099483307228e-07,
+      "loss": 0.3947,
+      "step": 2926
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2859070126811897e-07,
+      "loss": 0.4122,
+      "step": 2927
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.256811461250432e-07,
+      "loss": 0.4253,
+      "step": 2928
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.2278128584768207e-07,
+      "loss": 0.3942,
+      "step": 2929
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.198911233724079e-07,
+      "loss": 0.4114,
+      "step": 2930
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.1701066162577186e-07,
+      "loss": 0.4067,
+      "step": 2931
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.141399035245053e-07,
+      "loss": 0.4263,
+      "step": 2932
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.112788519755062e-07,
+      "loss": 0.4127,
+      "step": 2933
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0842750987585057e-07,
+      "loss": 0.4287,
+      "step": 2934
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.05585880112781e-07,
+      "loss": 0.4246,
+      "step": 2935
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.0275396556370584e-07,
+      "loss": 0.4218,
+      "step": 2936
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.999317690961946e-07,
+      "loss": 0.408,
+      "step": 2937
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.9711929356797686e-07,
+      "loss": 0.4668,
+      "step": 2938
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.943165418269401e-07,
+      "loss": 0.4331,
+      "step": 2939
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.915235167111231e-07,
+      "loss": 0.418,
+      "step": 2940
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.887402210487179e-07,
+      "loss": 0.4324,
+      "step": 2941
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.859666576580623e-07,
+      "loss": 0.3956,
+      "step": 2942
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.8320282934764306e-07,
+      "loss": 0.4155,
+      "step": 2943
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.804487389160849e-07,
+      "loss": 0.3979,
+      "step": 2944
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.777043891521559e-07,
+      "loss": 0.4274,
+      "step": 2945
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.749697828347565e-07,
+      "loss": 0.4155,
+      "step": 2946
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.722449227329239e-07,
+      "loss": 0.4281,
+      "step": 2947
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6952981160582767e-07,
+      "loss": 0.4305,
+      "step": 2948
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6682445220276195e-07,
+      "loss": 0.4347,
+      "step": 2949
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.64128847263151e-07,
+      "loss": 0.4219,
+      "step": 2950
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.6144299951653584e-07,
+      "loss": 0.429,
+      "step": 2951
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5876691168258536e-07,
+      "loss": 0.4126,
+      "step": 2952
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.561005864710754e-07,
+      "loss": 0.4194,
+      "step": 2953
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.5344402658190614e-07,
+      "loss": 0.405,
+      "step": 2954
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.507972347050825e-07,
+      "loss": 0.428,
+      "step": 2955
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.4816021352072403e-07,
+      "loss": 0.4192,
+      "step": 2956
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.455329656990514e-07,
+      "loss": 0.422,
+      "step": 2957
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.429154939003931e-07,
+      "loss": 0.425,
+      "step": 2958
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.4030780077517477e-07,
+      "loss": 0.4109,
+      "step": 2959
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.37709888963923e-07,
+      "loss": 0.4064,
+      "step": 2960
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.351217610972579e-07,
+      "loss": 0.4261,
+      "step": 2961
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.3254341979589545e-07,
+      "loss": 0.4137,
+      "step": 2962
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2997486767063824e-07,
+      "loss": 0.4362,
+      "step": 2963
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.2741610732237917e-07,
+      "loss": 0.4109,
+      "step": 2964
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.248671413420956e-07,
+      "loss": 0.4248,
+      "step": 2965
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.223279723108441e-07,
+      "loss": 0.416,
+      "step": 2966
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.197986027997657e-07,
+      "loss": 0.4156,
+      "step": 2967
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.172790353700761e-07,
+      "loss": 0.4181,
+      "step": 2968
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1476927257306444e-07,
+      "loss": 0.4257,
+      "step": 2969
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.1226931695009455e-07,
+      "loss": 0.4192,
+      "step": 2970
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.09779171032597e-07,
+      "loss": 0.4165,
+      "step": 2971
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0729883734207047e-07,
+      "loss": 0.4244,
+      "step": 2972
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.048283183900791e-07,
+      "loss": 0.4308,
+      "step": 2973
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.0236761667824523e-07,
+      "loss": 0.4288,
+      "step": 2974
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.999167346982534e-07,
+      "loss": 0.4106,
+      "step": 2975
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.9747567493184415e-07,
+      "loss": 0.4496,
+      "step": 2976
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.950444398508112e-07,
+      "loss": 0.4316,
+      "step": 2977
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 2.926230319170009e-07,
+      "loss": 0.4352,
+      "step": 2978
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.902114535823075e-07,
+      "loss": 0.4152,
+      "step": 2979
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.878097072886721e-07,
+      "loss": 0.4083,
+      "step": 2980
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.8541779546808255e-07,
+      "loss": 0.4355,
+      "step": 2981
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.830357205425649e-07,
+      "loss": 0.4125,
+      "step": 2982
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.806634849241863e-07,
+      "loss": 0.3937,
+      "step": 2983
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.7830109101505075e-07,
+      "loss": 0.407,
+      "step": 2984
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.75948541207296e-07,
+      "loss": 0.4163,
+      "step": 2985
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.736058378830919e-07,
+      "loss": 0.4208,
+      "step": 2986
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.712729834146366e-07,
+      "loss": 0.4148,
+      "step": 2987
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.689499801641593e-07,
+      "loss": 0.4397,
+      "step": 2988
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 2.6663683048390955e-07,
+      "loss": 0.4059,
+      "step": 2989
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.643335367161615e-07,
+      "loss": 0.4402,
+      "step": 2990
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.620401011932106e-07,
+      "loss": 0.3975,
+      "step": 2991
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5975652623736467e-07,
+      "loss": 0.428,
+      "step": 2992
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.574828141609531e-07,
+      "loss": 0.4066,
+      "step": 2993
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.5521896726631414e-07,
+      "loss": 0.4044,
+      "step": 2994
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.529649878457985e-07,
+      "loss": 0.4606,
+      "step": 2995
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.507208781817638e-07,
+      "loss": 0.4121,
+      "step": 2996
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4848664054657687e-07,
+      "loss": 0.4138,
+      "step": 2997
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.4626227720260465e-07,
+      "loss": 0.412,
+      "step": 2998
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.440477904022143e-07,
+      "loss": 0.4158,
+      "step": 2999
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 2.418431823877776e-07,
+      "loss": 0.4226,
+      "step": 3000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3964845539165894e-07,
+      "loss": 0.4442,
+      "step": 3001
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3746361163621723e-07,
+      "loss": 0.3911,
+      "step": 3002
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3528865333380724e-07,
+      "loss": 0.4354,
+      "step": 3003
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3312358268676837e-07,
+      "loss": 0.4025,
+      "step": 3004
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.3096840188743365e-07,
+      "loss": 0.4008,
+      "step": 3005
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2882311311811845e-07,
+      "loss": 0.4175,
+      "step": 3006
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.266877185511207e-07,
+      "loss": 0.4247,
+      "step": 3007
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2456222034872076e-07,
+      "loss": 0.3983,
+      "step": 3008
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2244662066318146e-07,
+      "loss": 0.4075,
+      "step": 3009
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.2034092163673582e-07,
+      "loss": 0.4192,
+      "step": 3010
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 2.1824512540159715e-07,
+      "loss": 0.436,
+      "step": 3011
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.1615923407994678e-07,
+      "loss": 0.4053,
+      "step": 3012
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.140832497839407e-07,
+      "loss": 0.4403,
+      "step": 3013
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.120171746157007e-07,
+      "loss": 0.4095,
+      "step": 3014
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0996101066731555e-07,
+      "loss": 0.4056,
+      "step": 3015
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0791476002083642e-07,
+      "loss": 0.4216,
+      "step": 3016
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0587842474827924e-07,
+      "loss": 0.4081,
+      "step": 3017
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.0385200691161787e-07,
+      "loss": 0.4153,
+      "step": 3018
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 2.018355085627821e-07,
+      "loss": 0.4022,
+      "step": 3019
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.998289317436608e-07,
+      "loss": 0.4401,
+      "step": 3020
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.978322784860942e-07,
+      "loss": 0.4198,
+      "step": 3021
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9584555081187507e-07,
+      "loss": 0.41,
+      "step": 3022
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9386875073274636e-07,
+      "loss": 0.3997,
+      "step": 3023
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.9190188025039692e-07,
+      "loss": 0.417,
+      "step": 3024
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8994494135645912e-07,
+      "loss": 0.4342,
+      "step": 3025
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8799793603251349e-07,
+      "loss": 0.4156,
+      "step": 3026
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8606086625007846e-07,
+      "loss": 0.4347,
+      "step": 3027
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8413373397061284e-07,
+      "loss": 0.4263,
+      "step": 3028
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8221654114551234e-07,
+      "loss": 0.438,
+      "step": 3029
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.8030928971610963e-07,
+      "loss": 0.4192,
+      "step": 3030
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.7841198161366868e-07,
+      "loss": 0.416,
+      "step": 3031
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.765246187593872e-07,
+      "loss": 0.4064,
+      "step": 3032
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7464720306438865e-07,
+      "loss": 0.4272,
+      "step": 3033
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.72779736429729e-07,
+      "loss": 0.4246,
+      "step": 3034
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.7092222074638675e-07,
+      "loss": 0.4213,
+      "step": 3035
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.69074657895264e-07,
+      "loss": 0.4167,
+      "step": 3036
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6723704974718758e-07,
+      "loss": 0.4277,
+      "step": 3037
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6540939816290014e-07,
+      "loss": 0.424,
+      "step": 3038
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6359170499306687e-07,
+      "loss": 0.4377,
+      "step": 3039
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6178397207826435e-07,
+      "loss": 0.4487,
+      "step": 3040
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5998620124898722e-07,
+      "loss": 0.4336,
+      "step": 3041
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5819839432564266e-07,
+      "loss": 0.4156,
+      "step": 3042
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.5642055311854476e-07,
+      "loss": 0.4163,
+      "step": 3043
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5465267942792129e-07,
+      "loss": 0.3947,
+      "step": 3044
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.528947750439036e-07,
+      "loss": 0.42,
+      "step": 3045
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.5114684174652783e-07,
+      "loss": 0.4128,
+      "step": 3046
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4940888130573705e-07,
+      "loss": 0.4209,
+      "step": 3047
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4768089548137242e-07,
+      "loss": 0.4446,
+      "step": 3048
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4596288602317766e-07,
+      "loss": 0.4154,
+      "step": 3049
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4425485467079114e-07,
+      "loss": 0.4338,
+      "step": 3050
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4255680315375164e-07,
+      "loss": 0.42,
+      "step": 3051
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.4086873319148707e-07,
+      "loss": 0.4102,
+      "step": 3052
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.391906464933246e-07,
+      "loss": 0.443,
+      "step": 3053
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.3752254475847716e-07,
+      "loss": 0.4175,
+      "step": 3054
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3586442967604918e-07,
+      "loss": 0.4296,
+      "step": 3055
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.342163029250332e-07,
+      "loss": 0.4044,
+      "step": 3056
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.3257816617430753e-07,
+      "loss": 0.4332,
+      "step": 3057
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.30950021082632e-07,
+      "loss": 0.4216,
+      "step": 3058
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2933186929865226e-07,
+      "loss": 0.4283,
+      "step": 3059
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2772371246089432e-07,
+      "loss": 0.4028,
+      "step": 3060
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2612555219776223e-07,
+      "loss": 0.4214,
+      "step": 3061
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2453739012753707e-07,
+      "loss": 0.425,
+      "step": 3062
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2295922785837799e-07,
+      "loss": 0.411,
+      "step": 3063
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.2139106698831892e-07,
+      "loss": 0.4055,
+      "step": 3064
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.198329091052608e-07,
+      "loss": 0.4121,
+      "step": 3065
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1828475578698373e-07,
+      "loss": 0.4099,
+      "step": 3066
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.167466086011304e-07,
+      "loss": 0.3921,
+      "step": 3067
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1521846910521606e-07,
+      "loss": 0.4415,
+      "step": 3068
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1370033884661957e-07,
+      "loss": 0.4342,
+      "step": 3069
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1219221936258684e-07,
+      "loss": 0.4056,
+      "step": 3070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.1069411218022297e-07,
+      "loss": 0.4232,
+      "step": 3071
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0920601881650006e-07,
+      "loss": 0.4366,
+      "step": 3072
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0772794077824611e-07,
+      "loss": 0.3954,
+      "step": 3073
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0625987956214834e-07,
+      "loss": 0.4,
+      "step": 3074
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.0480183665475319e-07,
+      "loss": 0.4153,
+      "step": 3075
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0335381353246076e-07,
+      "loss": 0.4032,
+      "step": 3076
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0191581166152708e-07,
+      "loss": 0.4359,
+      "step": 3077
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.0048783249805627e-07,
+      "loss": 0.4456,
+      "step": 3078
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.906987748800945e-08,
+      "loss": 0.415,
+      "step": 3079
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.766194806719364e-08,
+      "loss": 0.4036,
+      "step": 3080
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.626404566126402e-08,
+      "loss": 0.4205,
+      "step": 3081
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.487617168572605e-08,
+      "loss": 0.3975,
+      "step": 3082
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.349832754592559e-08,
+      "loss": 0.4265,
+      "step": 3083
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.213051463705547e-08,
+      "loss": 0.4557,
+      "step": 3084
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.077273434415002e-08,
+      "loss": 0.4376,
+      "step": 3085
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.9424988042085e-08,
+      "loss": 0.423,
+      "step": 3086
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.808727709557652e-08,
+      "loss": 0.4221,
+      "step": 3087
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.675960285917662e-08,
+      "loss": 0.4407,
+      "step": 3088
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.544196667727767e-08,
+      "loss": 0.4161,
+      "step": 3089
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.413436988410351e-08,
+      "loss": 0.4264,
+      "step": 3090
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.283681380371722e-08,
+      "loss": 0.4315,
+      "step": 3091
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.154929975001002e-08,
+      "loss": 0.4036,
+      "step": 3092
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.027182902670571e-08,
+      "loss": 0.4374,
+      "step": 3093
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.900440292736178e-08,
+      "loss": 0.4364,
+      "step": 3094
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.774702273535938e-08,
+      "loss": 0.4189,
+      "step": 3095
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.64996897239123e-08,
+      "loss": 0.4176,
+      "step": 3096
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 7.526240515605799e-08,
+      "loss": 0.4429,
+      "step": 3097
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.403517028465757e-08,
+      "loss": 0.4248,
+      "step": 3098
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.281798635239812e-08,
+      "loss": 0.4233,
+      "step": 3099
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.161085459178929e-08,
+      "loss": 0.4061,
+      "step": 3100
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.04137762251611e-08,
+      "loss": 0.4028,
+      "step": 3101
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.922675246466392e-08,
+      "loss": 0.4037,
+      "step": 3102
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.804978451226741e-08,
+      "loss": 0.4141,
+      "step": 3103
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.68828735597582e-08,
+      "loss": 0.4154,
+      "step": 3104
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.57260207887389e-08,
+      "loss": 0.4128,
+      "step": 3105
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.457922737062694e-08,
+      "loss": 0.4281,
+      "step": 3106
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 6.344249446665673e-08,
+      "loss": 0.4357,
+      "step": 3107
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.2315823227872e-08,
+      "loss": 0.4076,
+      "step": 3108
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.119921479513013e-08,
+      "loss": 0.4052,
+      "step": 3109
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 6.009267029909893e-08,
+      "loss": 0.4365,
+      "step": 3110
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.899619086025321e-08,
+      "loss": 0.4296,
+      "step": 3111
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.790977758887817e-08,
+      "loss": 0.4302,
+      "step": 3112
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.6833431585067155e-08,
+      "loss": 0.4301,
+      "step": 3113
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.576715393871612e-08,
+      "loss": 0.4307,
+      "step": 3114
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.4710945729530285e-08,
+      "loss": 0.3995,
+      "step": 3115
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.366480802701413e-08,
+      "loss": 0.4207,
+      "step": 3116
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.262874189047806e-08,
+      "loss": 0.4315,
+      "step": 3117
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 5.160274836903179e-08,
+      "loss": 0.4271,
+      "step": 3118
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 5.058682850158758e-08,
+      "loss": 0.4221,
+      "step": 3119
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.958098331685701e-08,
+      "loss": 0.4247,
+      "step": 3120
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.8585213833348686e-08,
+      "loss": 0.4249,
+      "step": 3121
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.759952105936827e-08,
+      "loss": 0.41,
+      "step": 3122
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.66239059930218e-08,
+      "loss": 0.3898,
+      "step": 3123
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.565836962220571e-08,
+      "loss": 0.4325,
+      "step": 3124
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.470291292461459e-08,
+      "loss": 0.4148,
+      "step": 3125
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.375753686773343e-08,
+      "loss": 0.4208,
+      "step": 3126
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.282224240884314e-08,
+      "loss": 0.4114,
+      "step": 3127
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.18970304950117e-08,
+      "loss": 0.4181,
+      "step": 3128
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.098190206310193e-08,
+      "loss": 0.4323,
+      "step": 3129
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.00768580397648e-08,
+      "loss": 0.4142,
+      "step": 3130
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.918189934143724e-08,
+      "loss": 0.4501,
+      "step": 3131
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.8297026874346554e-08,
+      "loss": 0.4506,
+      "step": 3132
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.742224153450824e-08,
+      "loss": 0.4581,
+      "step": 3133
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.655754420771929e-08,
+      "loss": 0.4061,
+      "step": 3134
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.570293576956596e-08,
+      "loss": 0.4001,
+      "step": 3135
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4858417085416044e-08,
+      "loss": 0.4275,
+      "step": 3136
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.4023989010421034e-08,
+      "loss": 0.4197,
+      "step": 3137
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.319965238951617e-08,
+      "loss": 0.4294,
+      "step": 3138
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.238540805741486e-08,
+      "loss": 0.4034,
+      "step": 3139
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.158125683861535e-08,
+      "loss": 0.4293,
+      "step": 3140
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0787199547392955e-08,
+      "loss": 0.4071,
+      "step": 3141
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.0003236987802276e-08,
+      "loss": 0.4037,
+      "step": 3142
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.92293699536772e-08,
+      "loss": 0.4171,
+      "step": 3143
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.8465599228627572e-08,
+      "loss": 0.4188,
+      "step": 3144
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.77119255860403e-08,
+      "loss": 0.4432,
+      "step": 3145
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.696834978907825e-08,
+      "loss": 0.424,
+      "step": 3146
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.623487259068136e-08,
+      "loss": 0.4241,
+      "step": 3147
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.551149473355885e-08,
+      "loss": 0.4003,
+      "step": 3148
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.479821695019813e-08,
+      "loss": 0.4327,
+      "step": 3149
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 2.4095039962857004e-08,
+      "loss": 0.4,
+      "step": 3150
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.340196448356591e-08,
+      "loss": 0.4364,
+      "step": 3151
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2718991214126796e-08,
+      "loss": 0.4572,
+      "step": 3152
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.2046120846112017e-08,
+      "loss": 0.4089,
+      "step": 3153
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.1383354060864337e-08,
+      "loss": 0.4119,
+      "step": 3154
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0730691529493586e-08,
+      "loss": 0.417,
+      "step": 3155
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 2.0088133912881114e-08,
+      "loss": 0.4189,
+      "step": 3156
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.9455681861675348e-08,
+      "loss": 0.4287,
+      "step": 3157
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.883333601628956e-08,
+      "loss": 0.42,
+      "step": 3158
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.822109700690633e-08,
+      "loss": 0.4153,
+      "step": 3159
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.761896545347308e-08,
+      "loss": 0.3975,
+      "step": 3160
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.7026941965702093e-08,
+      "loss": 0.4249,
+      "step": 3161
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.6445027143071612e-08,
+      "loss": 0.4209,
+      "step": 3162
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.587322157482252e-08,
+      "loss": 0.4512,
+      "step": 3163
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.531152583996054e-08,
+      "loss": 0.4297,
+      "step": 3164
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.4759940507251825e-08,
+      "loss": 0.4077,
+      "step": 3165
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.421846613522848e-08,
+      "loss": 0.4004,
+      "step": 3166
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3687103272181923e-08,
+      "loss": 0.4058,
+      "step": 3167
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.3165852456163975e-08,
+      "loss": 0.3915,
+      "step": 3168
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2654714214990205e-08,
+      "loss": 0.4062,
+      "step": 3169
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.2153689066233266e-08,
+      "loss": 0.4304,
+      "step": 3170
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.166277751722622e-08,
+      "loss": 0.4189,
+      "step": 3171
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.1181980065062547e-08,
+      "loss": 0.426,
+      "step": 3172
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0711297196592806e-08,
+      "loss": 0.4191,
+      "step": 3173
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.0250729388425751e-08,
+      "loss": 0.4211,
+      "step": 3174
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.800277106927215e-09,
+      "loss": 0.4105,
+      "step": 3175
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.359940808223444e-09,
+      "loss": 0.4193,
+      "step": 3176
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.929720938193331e-09,
+      "loss": 0.4099,
+      "step": 3177
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.509617932472846e-09,
+      "loss": 0.4279,
+      "step": 3178
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.099632216456154e-09,
+      "loss": 0.4187,
+      "step": 3179
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.69976420528895e-09,
+      "loss": 0.4221,
+      "step": 3180
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.310014303875124e-09,
+      "loss": 0.4189,
+      "step": 3181
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.930382906872313e-09,
+      "loss": 0.4336,
+      "step": 3182
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.56087039868969e-09,
+      "loss": 0.4121,
+      "step": 3183
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 6.201477153493507e-09,
+      "loss": 0.4236,
+      "step": 3184
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.852203535202661e-09,
+      "loss": 0.4138,
+      "step": 3185
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.513049897486467e-09,
+      "loss": 0.4327,
+      "step": 3186
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.1840165837702124e-09,
+      "loss": 0.4082,
+      "step": 3187
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.865103927229609e-09,
+      "loss": 0.4304,
+      "step": 3188
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.556312250793005e-09,
+      "loss": 0.4205,
+      "step": 3189
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.257641867139173e-09,
+      "loss": 0.4297,
+      "step": 3190
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.9690930786995266e-09,
+      "loss": 0.4229,
+      "step": 3191
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.690666177654789e-09,
+      "loss": 0.418,
+      "step": 3192
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.422361445938327e-09,
+      "loss": 0.4515,
+      "step": 3193
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.1641791552339265e-09,
+      "loss": 0.4023,
+      "step": 3194
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.916119566973574e-09,
+      "loss": 0.4254,
+      "step": 3195
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.678182932339679e-09,
+      "loss": 0.3982,
+      "step": 3196
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.45036949226396e-09,
+      "loss": 0.4034,
+      "step": 3197
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.232679477430777e-09,
+      "loss": 0.4332,
+      "step": 3198
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.025113108269361e-09,
+      "loss": 0.4208,
+      "step": 3199
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.8276705949593631e-09,
+      "loss": 0.4254,
+      "step": 3200
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6403521374286357e-09,
+      "loss": 0.415,
+      "step": 3201
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.4631579253565619e-09,
+      "loss": 0.4297,
+      "step": 3202
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.296088138166285e-09,
+      "loss": 0.4237,
+      "step": 3203
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.1391429450302583e-09,
+      "loss": 0.4174,
+      "step": 3204
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.923225048724672e-10,
+      "loss": 0.4062,
+      "step": 3205
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.556269663595462e-10,
+      "loss": 0.4141,
+      "step": 3206
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.290564679096612e-10,
+      "loss": 0.3955,
+      "step": 3207
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 6.126111376858479e-10,
+      "loss": 0.4041,
+      "step": 3208
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 5.062910936004528e-10,
+      "loss": 0.4425,
+      "step": 3209
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.1009644331180264e-10,
+      "loss": 0.4471,
+      "step": 3210
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.240272842253145e-10,
+      "loss": 0.4259,
+      "step": 3211
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.48083703494606e-10,
+      "loss": 0.4255,
+      "step": 3212
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.822657780181647e-10,
+      "loss": 0.4269,
+      "step": 3213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2657357444378903e-10,
+      "loss": 0.4131,
+      "step": 3214
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.100714916414731e-11,
+      "loss": 0.449,
+      "step": 3215
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.556654832010843e-11,
+      "loss": 0.4144,
+      "step": 3216
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 2.0251807797411204e-11,
+      "loss": 0.4164,
+      "step": 3217
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.062953231105283e-12,
+      "loss": 0.4226,
+      "step": 3218
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 0.0,
+      "loss": 0.4039,
+      "step": 3219
+    },
+    {
+      "epoch": 3.0,
+      "step": 3219,
+      "total_flos": 3.5328358468691165e+19,
+      "train_loss": 0.5010387353106069,
+      "train_runtime": 53338.978,
+      "train_samples_per_second": 15.453,
+      "train_steps_per_second": 0.06
+    }
+  ],
+  "max_steps": 3219,
+  "num_train_epochs": 3,
+  "total_flos": 3.5328358468691165e+19,
+  "trial_name": null,
+  "trial_params": null
+}