diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16821 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.8687558175612783,
+  "eval_steps": 500,
+  "global_step": 2800,
+  "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": 1.5256,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.123711340206186e-07,
+      "loss": 1.479,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.185567010309279e-07,
+      "loss": 1.5353,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.247422680412372e-07,
+      "loss": 1.5129,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0309278350515464e-06,
+      "loss": 1.5303,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2371134020618557e-06,
+      "loss": 1.4805,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4432989690721649e-06,
+      "loss": 1.4738,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6494845360824744e-06,
+      "loss": 1.4154,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8556701030927837e-06,
+      "loss": 1.4077,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.061855670103093e-06,
+      "loss": 1.3999,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.268041237113402e-06,
+      "loss": 1.3334,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4742268041237115e-06,
+      "loss": 1.3346,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.680412371134021e-06,
+      "loss": 1.3407,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8865979381443297e-06,
+      "loss": 1.3016,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0927835051546395e-06,
+      "loss": 1.3023,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.298969072164949e-06,
+      "loss": 1.294,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5051546391752577e-06,
+      "loss": 1.2748,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7113402061855674e-06,
+      "loss": 1.2387,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.917525773195877e-06,
+      "loss": 1.2392,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.123711340206186e-06,
+      "loss": 1.2732,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.329896907216495e-06,
+      "loss": 1.2198,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.536082474226804e-06,
+      "loss": 1.2312,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.742268041237113e-06,
+      "loss": 1.2241,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.948453608247423e-06,
+      "loss": 1.1967,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.154639175257732e-06,
+      "loss": 1.2168,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.360824742268042e-06,
+      "loss": 1.1827,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.567010309278351e-06,
+      "loss": 1.1516,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7731958762886594e-06,
+      "loss": 1.1732,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.979381443298969e-06,
+      "loss": 1.1851,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.185567010309279e-06,
+      "loss": 1.1797,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.391752577319588e-06,
+      "loss": 1.148,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.597938144329898e-06,
+      "loss": 1.1582,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.804123711340207e-06,
+      "loss": 1.1267,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.010309278350515e-06,
+      "loss": 1.1363,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.216494845360825e-06,
+      "loss": 1.1357,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.422680412371135e-06,
+      "loss": 1.1158,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.628865979381444e-06,
+      "loss": 1.0985,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.835051546391754e-06,
+      "loss": 1.1028,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.041237113402063e-06,
+      "loss": 1.1132,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.247422680412371e-06,
+      "loss": 1.094,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.453608247422681e-06,
+      "loss": 1.0998,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.65979381443299e-06,
+      "loss": 1.0958,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.865979381443299e-06,
+      "loss": 1.1033,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.072164948453609e-06,
+      "loss": 1.1022,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.278350515463918e-06,
+      "loss": 1.0859,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.484536082474226e-06,
+      "loss": 1.0848,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.690721649484536e-06,
+      "loss": 1.0666,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.896907216494846e-06,
+      "loss": 1.07,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0103092783505156e-05,
+      "loss": 1.0678,
+      "step": 49
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0309278350515464e-05,
+      "loss": 1.0785,
+      "step": 50
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0515463917525775e-05,
+      "loss": 1.0392,
+      "step": 51
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0721649484536083e-05,
+      "loss": 1.0617,
+      "step": 52
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0927835051546391e-05,
+      "loss": 1.0524,
+      "step": 53
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1134020618556703e-05,
+      "loss": 1.0533,
+      "step": 54
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.134020618556701e-05,
+      "loss": 1.057,
+      "step": 55
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1546391752577319e-05,
+      "loss": 1.0353,
+      "step": 56
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.175257731958763e-05,
+      "loss": 1.041,
+      "step": 57
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1958762886597938e-05,
+      "loss": 1.0717,
+      "step": 58
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2164948453608248e-05,
+      "loss": 1.0502,
+      "step": 59
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2371134020618558e-05,
+      "loss": 1.0353,
+      "step": 60
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2577319587628866e-05,
+      "loss": 1.0399,
+      "step": 61
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2783505154639176e-05,
+      "loss": 1.0419,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2989690721649485e-05,
+      "loss": 1.0375,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3195876288659795e-05,
+      "loss": 1.0527,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3402061855670103e-05,
+      "loss": 1.0277,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3608247422680415e-05,
+      "loss": 1.0399,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3814432989690723e-05,
+      "loss": 0.9983,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.402061855670103e-05,
+      "loss": 1.0395,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4226804123711342e-05,
+      "loss": 1.0405,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.443298969072165e-05,
+      "loss": 1.0105,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4639175257731958e-05,
+      "loss": 1.0247,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.484536082474227e-05,
+      "loss": 0.9939,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5051546391752578e-05,
+      "loss": 1.0359,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5257731958762888e-05,
+      "loss": 1.0261,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5463917525773197e-05,
+      "loss": 1.01,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5670103092783507e-05,
+      "loss": 1.0128,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5876288659793813e-05,
+      "loss": 1.0484,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6082474226804127e-05,
+      "loss": 1.027,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6288659793814433e-05,
+      "loss": 1.034,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6494845360824743e-05,
+      "loss": 1.0047,
+      "step": 80
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6701030927835052e-05,
+      "loss": 1.0186,
+      "step": 81
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6907216494845362e-05,
+      "loss": 1.0098,
+      "step": 82
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7113402061855672e-05,
+      "loss": 1.0261,
+      "step": 83
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.731958762886598e-05,
+      "loss": 0.9943,
+      "step": 84
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.752577319587629e-05,
+      "loss": 1.0104,
+      "step": 85
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7731958762886598e-05,
+      "loss": 0.9975,
+      "step": 86
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.793814432989691e-05,
+      "loss": 1.0054,
+      "step": 87
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8144329896907217e-05,
+      "loss": 0.9957,
+      "step": 88
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8350515463917527e-05,
+      "loss": 1.0204,
+      "step": 89
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8556701030927837e-05,
+      "loss": 0.9864,
+      "step": 90
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8762886597938147e-05,
+      "loss": 0.9994,
+      "step": 91
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8969072164948453e-05,
+      "loss": 1.0086,
+      "step": 92
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9175257731958766e-05,
+      "loss": 1.0109,
+      "step": 93
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9381443298969072e-05,
+      "loss": 1.0343,
+      "step": 94
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9587628865979382e-05,
+      "loss": 1.0224,
+      "step": 95
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9793814432989692e-05,
+      "loss": 0.9803,
+      "step": 96
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9876,
+      "step": 97
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994949995493e-05,
+      "loss": 1.0318,
+      "step": 98
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997979998707e-05,
+      "loss": 0.9679,
+      "step": 99
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999954549990032e-05,
+      "loss": 0.9992,
+      "step": 100
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999919200029883e-05,
+      "loss": 0.9846,
+      "step": 101
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999873750142327e-05,
+      "loss": 0.9704,
+      "step": 102
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999818200373267e-05,
+      "loss": 0.9893,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999752550778807e-05,
+      "loss": 1.0042,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999676801425253e-05,
+      "loss": 0.993,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999590952389118e-05,
+      "loss": 0.9888,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999495003757103e-05,
+      "loss": 0.9736,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999938895562612e-05,
+      "loss": 0.9768,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999272808103276e-05,
+      "loss": 0.973,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999146561305883e-05,
+      "loss": 0.9898,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999010215361444e-05,
+      "loss": 0.9903,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999886377040767e-05,
+      "loss": 0.9664,
+      "step": 112
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999870722659248e-05,
+      "loss": 0.9889,
+      "step": 113
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998540584073972e-05,
+      "loss": 0.9733,
+      "step": 114
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998363843020458e-05,
+      "loss": 0.9774,
+      "step": 115
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999817700361045e-05,
+      "loss": 0.9731,
+      "step": 116
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997980066032655e-05,
+      "loss": 1.0069,
+      "step": 117
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997773030485976e-05,
+      "loss": 0.9803,
+      "step": 118
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997555897179524e-05,
+      "loss": 1.0022,
+      "step": 119
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997328666332602e-05,
+      "loss": 0.9563,
+      "step": 120
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997091338174713e-05,
+      "loss": 0.9784,
+      "step": 121
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996843912945557e-05,
+      "loss": 1.004,
+      "step": 122
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999658639089504e-05,
+      "loss": 0.9774,
+      "step": 123
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999631877228325e-05,
+      "loss": 0.9675,
+      "step": 124
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996041057380494e-05,
+      "loss": 1.0001,
+      "step": 125
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995753246467252e-05,
+      "loss": 0.9719,
+      "step": 126
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999545533983422e-05,
+      "loss": 0.9871,
+      "step": 127
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995147337782285e-05,
+      "loss": 0.96,
+      "step": 128
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994829240622524e-05,
+      "loss": 0.965,
+      "step": 129
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999450104867622e-05,
+      "loss": 0.9558,
+      "step": 130
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994162762274848e-05,
+      "loss": 0.9608,
+      "step": 131
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993814381760073e-05,
+      "loss": 0.9694,
+      "step": 132
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993455907483762e-05,
+      "loss": 0.9762,
+      "step": 133
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993087339807973e-05,
+      "loss": 0.9646,
+      "step": 134
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992708679104963e-05,
+      "loss": 0.9546,
+      "step": 135
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992319925757176e-05,
+      "loss": 0.9628,
+      "step": 136
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991921080157254e-05,
+      "loss": 0.9565,
+      "step": 137
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991512142708032e-05,
+      "loss": 0.9775,
+      "step": 138
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991093113822542e-05,
+      "loss": 0.9586,
+      "step": 139
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990663993924e-05,
+      "loss": 0.9512,
+      "step": 140
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999022478344581e-05,
+      "loss": 0.9643,
+      "step": 141
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989775482831583e-05,
+      "loss": 0.9699,
+      "step": 142
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989316092535113e-05,
+      "loss": 0.9395,
+      "step": 143
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988846613020383e-05,
+      "loss": 0.9903,
+      "step": 144
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988367044761568e-05,
+      "loss": 0.9717,
+      "step": 145
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987877388243033e-05,
+      "loss": 0.9521,
+      "step": 146
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987377643959327e-05,
+      "loss": 0.9689,
+      "step": 147
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99868678124152e-05,
+      "loss": 0.9474,
+      "step": 148
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986347894125578e-05,
+      "loss": 0.9519,
+      "step": 149
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985817889615578e-05,
+      "loss": 0.9418,
+      "step": 150
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985277799420504e-05,
+      "loss": 0.9421,
+      "step": 151
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984727624085855e-05,
+      "loss": 0.9577,
+      "step": 152
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984167364167298e-05,
+      "loss": 0.9456,
+      "step": 153
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983597020230702e-05,
+      "loss": 0.9394,
+      "step": 154
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983016592852114e-05,
+      "loss": 0.9483,
+      "step": 155
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998242608261777e-05,
+      "loss": 0.9588,
+      "step": 156
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998182549012408e-05,
+      "loss": 0.9676,
+      "step": 157
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998121481597765e-05,
+      "loss": 0.9825,
+      "step": 158
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998059406079525e-05,
+      "loss": 0.9591,
+      "step": 159
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979963225203852e-05,
+      "loss": 0.9303,
+      "step": 160
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99793223098406e-05,
+      "loss": 0.9417,
+      "step": 161
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997867131535282e-05,
+      "loss": 0.9523,
+      "step": 162
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997801024239801e-05,
+      "loss": 0.9479,
+      "step": 163
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977339091643858e-05,
+      "loss": 0.9537,
+      "step": 164
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997665786376823e-05,
+      "loss": 0.9348,
+      "step": 165
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975966559459167e-05,
+      "loss": 0.943,
+      "step": 166
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975265179414883e-05,
+      "loss": 0.9402,
+      "step": 167
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974553724343774e-05,
+      "loss": 0.9526,
+      "step": 168
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997383219496441e-05,
+      "loss": 0.9427,
+      "step": 169
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973100592005534e-05,
+      "loss": 0.9531,
+      "step": 170
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972358916206073e-05,
+      "loss": 0.9348,
+      "step": 171
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997160716831511e-05,
+      "loss": 0.9371,
+      "step": 172
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997084534909192e-05,
+      "loss": 0.9422,
+      "step": 173
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970073459305936e-05,
+      "loss": 0.9465,
+      "step": 174
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996929149973677e-05,
+      "loss": 0.9536,
+      "step": 175
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968499471174204e-05,
+      "loss": 0.9164,
+      "step": 176
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996769737441818e-05,
+      "loss": 0.9439,
+      "step": 177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966885210278822e-05,
+      "loss": 0.9346,
+      "step": 178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966062979576417e-05,
+      "loss": 0.9318,
+      "step": 179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965230683141415e-05,
+      "loss": 0.95,
+      "step": 180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996438832181444e-05,
+      "loss": 0.9469,
+      "step": 181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963535896446276e-05,
+      "loss": 0.9187,
+      "step": 182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996267340789787e-05,
+      "loss": 0.9533,
+      "step": 183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961800857040343e-05,
+      "loss": 0.9498,
+      "step": 184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960918244754968e-05,
+      "loss": 0.9647,
+      "step": 185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960025571933187e-05,
+      "loss": 0.9491,
+      "step": 186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959122839476594e-05,
+      "loss": 0.9303,
+      "step": 187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958210048296957e-05,
+      "loss": 0.9268,
+      "step": 188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995728719931619e-05,
+      "loss": 0.9404,
+      "step": 189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956354293466374e-05,
+      "loss": 0.941,
+      "step": 190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995541133168975e-05,
+      "loss": 0.9399,
+      "step": 191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99544583149387e-05,
+      "loss": 0.9369,
+      "step": 192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995349524417578e-05,
+      "loss": 0.9146,
+      "step": 193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995252212037369e-05,
+      "loss": 0.9459,
+      "step": 194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951538944515283e-05,
+      "loss": 0.94,
+      "step": 195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995054571759357e-05,
+      "loss": 0.9253,
+      "step": 196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994954244061171e-05,
+      "loss": 0.9243,
+      "step": 197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948529114583015e-05,
+      "loss": 0.9262,
+      "step": 198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947505740530947e-05,
+      "loss": 0.925,
+      "step": 199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994647231948911e-05,
+      "loss": 0.9535,
+      "step": 200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994542885250126e-05,
+      "loss": 0.9277,
+      "step": 201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944375340621306e-05,
+      "loss": 0.9206,
+      "step": 202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943311784913292e-05,
+      "loss": 0.9548,
+      "step": 203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942238186451407e-05,
+      "loss": 0.9348,
+      "step": 204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994115454631999e-05,
+      "loss": 0.9056,
+      "step": 205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940060865613523e-05,
+      "loss": 0.9344,
+      "step": 206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938957145436613e-05,
+      "loss": 0.9284,
+      "step": 207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937843386904032e-05,
+      "loss": 0.9103,
+      "step": 208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936719591140666e-05,
+      "loss": 0.9369,
+      "step": 209
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935585759281555e-05,
+      "loss": 0.9163,
+      "step": 210
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934441892471866e-05,
+      "loss": 0.9384,
+      "step": 211
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933287991866913e-05,
+      "loss": 0.9429,
+      "step": 212
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932124058632133e-05,
+      "loss": 0.9211,
+      "step": 213
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930950093943095e-05,
+      "loss": 0.9285,
+      "step": 214
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992976609898551e-05,
+      "loss": 0.919,
+      "step": 215
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928572074955214e-05,
+      "loss": 0.9218,
+      "step": 216
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992736802305817e-05,
+      "loss": 0.9103,
+      "step": 217
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992615394451047e-05,
+      "loss": 0.9541,
+      "step": 218
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924929840538335e-05,
+      "loss": 0.901,
+      "step": 219
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923695712378117e-05,
+      "loss": 0.9021,
+      "step": 220
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992245156127628e-05,
+      "loss": 0.9313,
+      "step": 221
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992119738848942e-05,
+      "loss": 0.9301,
+      "step": 222
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919933195284255e-05,
+      "loss": 0.956,
+      "step": 223
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918658982937616e-05,
+      "loss": 0.9255,
+      "step": 224
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991737475273646e-05,
+      "loss": 0.9263,
+      "step": 225
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916080505977864e-05,
+      "loss": 0.9026,
+      "step": 226
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991477624396902e-05,
+      "loss": 0.9013,
+      "step": 227
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913461968027228e-05,
+      "loss": 0.9189,
+      "step": 228
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991213767947991e-05,
+      "loss": 0.9027,
+      "step": 229
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910803379664596e-05,
+      "loss": 0.9329,
+      "step": 230
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909459069928937e-05,
+      "loss": 0.9456,
+      "step": 231
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908104751630682e-05,
+      "loss": 0.9079,
+      "step": 232
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906740426137695e-05,
+      "loss": 0.9053,
+      "step": 233
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905366094827942e-05,
+      "loss": 0.9319,
+      "step": 234
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903981759089507e-05,
+      "loss": 0.9408,
+      "step": 235
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902587420320563e-05,
+      "loss": 0.9128,
+      "step": 236
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901183079929395e-05,
+      "loss": 0.9161,
+      "step": 237
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899768739334393e-05,
+      "loss": 0.9041,
+      "step": 238
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898344399964037e-05,
+      "loss": 0.924,
+      "step": 239
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896910063256915e-05,
+      "loss": 0.9,
+      "step": 240
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98954657306617e-05,
+      "loss": 0.9024,
+      "step": 241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894011403637183e-05,
+      "loss": 0.9348,
+      "step": 242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989254708365222e-05,
+      "loss": 0.9127,
+      "step": 243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891072772185786e-05,
+      "loss": 0.9199,
+      "step": 244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889588470726936e-05,
+      "loss": 0.9045,
+      "step": 245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888094180774816e-05,
+      "loss": 0.915,
+      "step": 246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886589903838653e-05,
+      "loss": 0.8792,
+      "step": 247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885075641437775e-05,
+      "loss": 0.9391,
+      "step": 248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988355139510159e-05,
+      "loss": 0.9123,
+      "step": 249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988201716636958e-05,
+      "loss": 0.9002,
+      "step": 250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880472956791325e-05,
+      "loss": 0.9282,
+      "step": 251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878918767926477e-05,
+      "loss": 0.9059,
+      "step": 252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877354601344764e-05,
+      "loss": 0.8949,
+      "step": 253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875780458626e-05,
+      "loss": 0.911,
+      "step": 254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874196341360067e-05,
+      "loss": 0.9051,
+      "step": 255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872602251146925e-05,
+      "loss": 0.9241,
+      "step": 256
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987099818959661e-05,
+      "loss": 0.9152,
+      "step": 257
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869384158329224e-05,
+      "loss": 0.9027,
+      "step": 258
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986776015897494e-05,
+      "loss": 0.9401,
+      "step": 259
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866126193174e-05,
+      "loss": 0.9035,
+      "step": 260
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864482262576704e-05,
+      "loss": 0.8909,
+      "step": 261
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862828368843433e-05,
+      "loss": 0.9259,
+      "step": 262
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861164513644615e-05,
+      "loss": 0.9065,
+      "step": 263
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859490698660752e-05,
+      "loss": 0.8686,
+      "step": 264
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985780692558239e-05,
+      "loss": 0.9197,
+      "step": 265
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856113196110143e-05,
+      "loss": 0.9195,
+      "step": 266
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854409511954688e-05,
+      "loss": 0.9039,
+      "step": 267
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852695874836736e-05,
+      "loss": 0.8872,
+      "step": 268
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985097228648707e-05,
+      "loss": 0.9051,
+      "step": 269
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984923874864651e-05,
+      "loss": 0.877,
+      "step": 270
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847495263065935e-05,
+      "loss": 0.8911,
+      "step": 271
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845741831506265e-05,
+      "loss": 0.9036,
+      "step": 272
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843978455738473e-05,
+      "loss": 0.9117,
+      "step": 273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842205137543563e-05,
+      "loss": 0.8896,
+      "step": 274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840421878712587e-05,
+      "loss": 0.922,
+      "step": 275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838628681046643e-05,
+      "loss": 0.9143,
+      "step": 276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836825546356862e-05,
+      "loss": 0.9038,
+      "step": 277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835012476464407e-05,
+      "loss": 0.8961,
+      "step": 278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833189473200487e-05,
+      "loss": 0.9491,
+      "step": 279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831356538406333e-05,
+      "loss": 0.9009,
+      "step": 280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982951367393321e-05,
+      "loss": 0.9081,
+      "step": 281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827660881642416e-05,
+      "loss": 0.9217,
+      "step": 282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825798163405267e-05,
+      "loss": 0.9025,
+      "step": 283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823925521103116e-05,
+      "loss": 0.9324,
+      "step": 284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982204295662733e-05,
+      "loss": 0.9036,
+      "step": 285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820150471879304e-05,
+      "loss": 0.8836,
+      "step": 286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981824806877045e-05,
+      "loss": 0.884,
+      "step": 287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816335749222188e-05,
+      "loss": 0.9401,
+      "step": 288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814413515165974e-05,
+      "loss": 1.0044,
+      "step": 289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812481368543258e-05,
+      "loss": 0.9514,
+      "step": 290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810539311305516e-05,
+      "loss": 0.8993,
+      "step": 291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980858734541422e-05,
+      "loss": 0.8807,
+      "step": 292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806625472840865e-05,
+      "loss": 0.9201,
+      "step": 293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980465369556694e-05,
+      "loss": 0.9032,
+      "step": 294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802672015583942e-05,
+      "loss": 0.9032,
+      "step": 295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980068043489337e-05,
+      "loss": 0.9046,
+      "step": 296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979867895550672e-05,
+      "loss": 0.8928,
+      "step": 297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796667579445494e-05,
+      "loss": 0.9065,
+      "step": 298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979464630874118e-05,
+      "loss": 0.9088,
+      "step": 299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979261514543526e-05,
+      "loss": 0.9185,
+      "step": 300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790574091579216e-05,
+      "loss": 0.8807,
+      "step": 301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978852314923451e-05,
+      "loss": 0.8571,
+      "step": 302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786462320472605e-05,
+      "loss": 0.8975,
+      "step": 303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784391607374928e-05,
+      "loss": 0.9122,
+      "step": 304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782311012032905e-05,
+      "loss": 0.8855,
+      "step": 305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978022053654794e-05,
+      "loss": 0.8874,
+      "step": 306
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778120183031414e-05,
+      "loss": 0.8886,
+      "step": 307
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977600995360469e-05,
+      "loss": 0.9007,
+      "step": 308
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97738898503991e-05,
+      "loss": 0.8981,
+      "step": 309
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771759875555945e-05,
+      "loss": 0.9226,
+      "step": 310
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976962003122651e-05,
+      "loss": 0.8878,
+      "step": 311
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976747031957203e-05,
+      "loss": 0.9064,
+      "step": 312
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765310742763723e-05,
+      "loss": 0.8728,
+      "step": 313
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763141302982763e-05,
+      "loss": 0.901,
+      "step": 314
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760962002420284e-05,
+      "loss": 0.8696,
+      "step": 315
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975877284327738e-05,
+      "loss": 0.8888,
+      "step": 316
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756573827765105e-05,
+      "loss": 0.8848,
+      "step": 317
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975436495810447e-05,
+      "loss": 0.9122,
+      "step": 318
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975214623652643e-05,
+      "loss": 0.8726,
+      "step": 319
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749917665271903e-05,
+      "loss": 0.8773,
+      "step": 320
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974767924659174e-05,
+      "loss": 0.8925,
+      "step": 321
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974543098274675e-05,
+      "loss": 0.8656,
+      "step": 322
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743172876007682e-05,
+      "loss": 0.8942,
+      "step": 323
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974090492865522e-05,
+      "loss": 0.8884,
+      "step": 324
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738627142980003e-05,
+      "loss": 0.8906,
+      "step": 325
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973633952128259e-05,
+      "loss": 0.9099,
+      "step": 326
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973404206587348e-05,
+      "loss": 0.8778,
+      "step": 327
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973173477907311e-05,
+      "loss": 0.8947,
+      "step": 328
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729417663211837e-05,
+      "loss": 0.8767,
+      "step": 329
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727090720629952e-05,
+      "loss": 0.9044,
+      "step": 330
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972475395367767e-05,
+      "loss": 0.895,
+      "step": 331
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722407364715126e-05,
+      "loss": 0.9064,
+      "step": 332
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972005095611238e-05,
+      "loss": 0.8684,
+      "step": 333
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717684730249403e-05,
+      "loss": 0.8892,
+      "step": 334
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715308689516087e-05,
+      "loss": 0.8857,
+      "step": 335
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712922836312237e-05,
+      "loss": 0.9093,
+      "step": 336
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710527173047565e-05,
+      "loss": 0.8702,
+      "step": 337
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970812170214169e-05,
+      "loss": 0.9115,
+      "step": 338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705706426024145e-05,
+      "loss": 0.8833,
+      "step": 339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970328134713436e-05,
+      "loss": 0.8821,
+      "step": 340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700846467921663e-05,
+      "loss": 0.8749,
+      "step": 341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698401790845288e-05,
+      "loss": 0.894,
+      "step": 342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969594731837436e-05,
+      "loss": 0.9075,
+      "step": 343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693483052987898e-05,
+      "loss": 0.8865,
+      "step": 344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691008997174813e-05,
+      "loss": 0.8706,
+      "step": 345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688525153433904e-05,
+      "loss": 0.8975,
+      "step": 346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686031524273854e-05,
+      "loss": 0.8789,
+      "step": 347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683528112213236e-05,
+      "loss": 0.9039,
+      "step": 348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968101491978049e-05,
+      "loss": 0.8675,
+      "step": 349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678491949513946e-05,
+      "loss": 0.8983,
+      "step": 350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675959203961804e-05,
+      "loss": 0.8695,
+      "step": 351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967341668568215e-05,
+      "loss": 0.8862,
+      "step": 352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670864397242914e-05,
+      "loss": 0.886,
+      "step": 353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668302341221922e-05,
+      "loss": 0.8821,
+      "step": 354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665730520206846e-05,
+      "loss": 0.8862,
+      "step": 355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663148936795234e-05,
+      "loss": 0.8796,
+      "step": 356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660557593594477e-05,
+      "loss": 0.8812,
+      "step": 357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657956493221847e-05,
+      "loss": 0.8889,
+      "step": 358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655345638304447e-05,
+      "loss": 0.9008,
+      "step": 359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965272503147925e-05,
+      "loss": 0.8565,
+      "step": 360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650094675393064e-05,
+      "loss": 0.874,
+      "step": 361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964745457270256e-05,
+      "loss": 0.8946,
+      "step": 362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644804726074236e-05,
+      "loss": 0.8844,
+      "step": 363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642145138184446e-05,
+      "loss": 0.8878,
+      "step": 364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963947581171937e-05,
+      "loss": 0.9071,
+      "step": 365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636796749375033e-05,
+      "loss": 0.8602,
+      "step": 366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963410795385729e-05,
+      "loss": 0.8662,
+      "step": 367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631409427881832e-05,
+      "loss": 0.8629,
+      "step": 368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628701174174165e-05,
+      "loss": 0.8797,
+      "step": 369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625983195469632e-05,
+      "loss": 0.8805,
+      "step": 370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962325549451339e-05,
+      "loss": 0.8946,
+      "step": 371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620518074060423e-05,
+      "loss": 0.8761,
+      "step": 372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617770936875528e-05,
+      "loss": 0.8842,
+      "step": 373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615014085733313e-05,
+      "loss": 0.8873,
+      "step": 374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612247523418206e-05,
+      "loss": 0.8764,
+      "step": 375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609471252724433e-05,
+      "loss": 0.874,
+      "step": 376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960668527645603e-05,
+      "loss": 0.8781,
+      "step": 377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603889597426836e-05,
+      "loss": 0.8867,
+      "step": 378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601084218460494e-05,
+      "loss": 0.8739,
+      "step": 379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959826914239043e-05,
+      "loss": 0.8913,
+      "step": 380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959544437205988e-05,
+      "loss": 0.8673,
+      "step": 381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592609910321864e-05,
+      "loss": 0.8629,
+      "step": 382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958976576003919e-05,
+      "loss": 0.8779,
+      "step": 383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586911924084445e-05,
+      "loss": 0.8812,
+      "step": 384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958404840534002e-05,
+      "loss": 0.8849,
+      "step": 385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581175206698062e-05,
+      "loss": 0.8651,
+      "step": 386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957829233106051e-05,
+      "loss": 0.875,
+      "step": 387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575399781339067e-05,
+      "loss": 0.8588,
+      "step": 388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572497560455207e-05,
+      "loss": 0.8738,
+      "step": 389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569585671340183e-05,
+      "loss": 0.8665,
+      "step": 390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566664116935004e-05,
+      "loss": 0.8888,
+      "step": 391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563732900190443e-05,
+      "loss": 0.9032,
+      "step": 392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560792024067027e-05,
+      "loss": 0.8885,
+      "step": 393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955784149153505e-05,
+      "loss": 0.8946,
+      "step": 394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554881305574545e-05,
+      "loss": 0.9004,
+      "step": 395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955191146917531e-05,
+      "loss": 0.8598,
+      "step": 396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548931985336876e-05,
+      "loss": 0.8679,
+      "step": 397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545942857068526e-05,
+      "loss": 0.8973,
+      "step": 398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954294408738929e-05,
+      "loss": 0.8693,
+      "step": 399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539935679327916e-05,
+      "loss": 0.8866,
+      "step": 400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953691763592291e-05,
+      "loss": 0.871,
+      "step": 401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533889960222488e-05,
+      "loss": 0.8912,
+      "step": 402
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530852655284612e-05,
+      "loss": 0.8641,
+      "step": 403
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952780572417696e-05,
+      "loss": 0.861,
+      "step": 404
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952474916997694e-05,
+      "loss": 0.8861,
+      "step": 405
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521682995771666e-05,
+      "loss": 0.8745,
+      "step": 406
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518607204657985e-05,
+      "loss": 0.8383,
+      "step": 407
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515521799742444e-05,
+      "loss": 0.8803,
+      "step": 408
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512426784141307e-05,
+      "loss": 0.8571,
+      "step": 409
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509322160980543e-05,
+      "loss": 0.8755,
+      "step": 410
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950620793339582e-05,
+      "loss": 0.8634,
+      "step": 411
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950308410453251e-05,
+      "loss": 0.8701,
+      "step": 412
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499950677545693e-05,
+      "loss": 0.8743,
+      "step": 413
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496807655600122e-05,
+      "loss": 0.8915,
+      "step": 414
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493655041870258e-05,
+      "loss": 0.8712,
+      "step": 415
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490492839540243e-05,
+      "loss": 0.8626,
+      "step": 416
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487321051803905e-05,
+      "loss": 0.8947,
+      "step": 417
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484139681864747e-05,
+      "loss": 0.8804,
+      "step": 418
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948094873293596e-05,
+      "loss": 0.8793,
+      "step": 419
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477748208240407e-05,
+      "loss": 0.8614,
+      "step": 420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474538111010615e-05,
+      "loss": 0.8553,
+      "step": 421
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471318444488793e-05,
+      "loss": 0.87,
+      "step": 422
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94680892119268e-05,
+      "loss": 0.8732,
+      "step": 423
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464850416586166e-05,
+      "loss": 0.8709,
+      "step": 424
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946160206173808e-05,
+      "loss": 0.8815,
+      "step": 425
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458344150663377e-05,
+      "loss": 0.8604,
+      "step": 426
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455076686652555e-05,
+      "loss": 0.8619,
+      "step": 427
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945179967300576e-05,
+      "loss": 0.8444,
+      "step": 428
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9448513113032766e-05,
+      "loss": 0.8439,
+      "step": 429
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445217010053015e-05,
+      "loss": 0.8769,
+      "step": 430
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944191136739556e-05,
+      "loss": 0.8571,
+      "step": 431
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943859618839912e-05,
+      "loss": 0.8931,
+      "step": 432
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435271476412015e-05,
+      "loss": 0.8674,
+      "step": 433
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431937234792214e-05,
+      "loss": 0.8733,
+      "step": 434
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94285934669073e-05,
+      "loss": 0.8763,
+      "step": 435
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942524017613449e-05,
+      "loss": 0.8708,
+      "step": 436
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421877365860596e-05,
+      "loss": 0.8631,
+      "step": 437
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941850503948207e-05,
+      "loss": 0.8708,
+      "step": 438
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415123200404962e-05,
+      "loss": 0.8679,
+      "step": 439
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411731852044933e-05,
+      "loss": 0.8643,
+      "step": 440
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940833099782725e-05,
+      "loss": 0.884,
+      "step": 441
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404920641186776e-05,
+      "loss": 0.8918,
+      "step": 442
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940150078556797e-05,
+      "loss": 0.8585,
+      "step": 443
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93980714344249e-05,
+      "loss": 0.8483,
+      "step": 444
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394632591221208e-05,
+      "loss": 0.8729,
+      "step": 445
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391184259430126e-05,
+      "loss": 0.8621,
+      "step": 446
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387726442534476e-05,
+      "loss": 0.8614,
+      "step": 447
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384259144026655e-05,
+      "loss": 0.8877,
+      "step": 448
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380782367408635e-05,
+      "loss": 0.8797,
+      "step": 449
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937729611619197e-05,
+      "loss": 0.8873,
+      "step": 450
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373800393897772e-05,
+      "loss": 0.8684,
+      "step": 451
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937029520405672e-05,
+      "loss": 0.8774,
+      "step": 452
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366780550209066e-05,
+      "loss": 0.8572,
+      "step": 453
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363256435904616e-05,
+      "loss": 0.8763,
+      "step": 454
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935972286470272e-05,
+      "loss": 0.8742,
+      "step": 455
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935617984017229e-05,
+      "loss": 0.8659,
+      "step": 456
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935262736589179e-05,
+      "loss": 0.8368,
+      "step": 457
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349065445449214e-05,
+      "loss": 0.8791,
+      "step": 458
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934549408244211e-05,
+      "loss": 0.8661,
+      "step": 459
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934191328047756e-05,
+      "loss": 0.8649,
+      "step": 460
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338323043172172e-05,
+      "loss": 0.857,
+      "step": 461
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334723374152087e-05,
+      "loss": 0.8577,
+      "step": 462
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331114277052985e-05,
+      "loss": 0.8504,
+      "step": 463
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327495755520047e-05,
+      "loss": 0.8653,
+      "step": 464
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323867813207987e-05,
+      "loss": 0.8837,
+      "step": 465
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320230453781026e-05,
+      "loss": 0.8541,
+      "step": 466
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316583680912907e-05,
+      "loss": 0.8662,
+      "step": 467
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312927498286867e-05,
+      "loss": 0.8682,
+      "step": 468
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930926190959566e-05,
+      "loss": 0.8754,
+      "step": 469
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305586918541527e-05,
+      "loss": 0.8399,
+      "step": 470
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930190252883622e-05,
+      "loss": 0.874,
+      "step": 471
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298208744200972e-05,
+      "loss": 0.8641,
+      "step": 472
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294505568366507e-05,
+      "loss": 0.8772,
+      "step": 473
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290793005073042e-05,
+      "loss": 0.8662,
+      "step": 474
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287071058070262e-05,
+      "loss": 0.8671,
+      "step": 475
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283339731117338e-05,
+      "loss": 0.8649,
+      "step": 476
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927959902798292e-05,
+      "loss": 0.8642,
+      "step": 477
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275848952445114e-05,
+      "loss": 0.856,
+      "step": 478
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272089508291507e-05,
+      "loss": 0.8752,
+      "step": 479
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268320699319134e-05,
+      "loss": 0.8822,
+      "step": 480
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264542529334497e-05,
+      "loss": 0.867,
+      "step": 481
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260755002153555e-05,
+      "loss": 0.8281,
+      "step": 482
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925695812160171e-05,
+      "loss": 0.8415,
+      "step": 483
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925315189151382e-05,
+      "loss": 0.8563,
+      "step": 484
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249336315734168e-05,
+      "loss": 0.8818,
+      "step": 485
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245511398116504e-05,
+      "loss": 0.8727,
+      "step": 486
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924167714252399e-05,
+      "loss": 0.8585,
+      "step": 487
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923783355282923e-05,
+      "loss": 0.8812,
+      "step": 488
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233980632914254e-05,
+      "loss": 0.8939,
+      "step": 489
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230118386670513e-05,
+      "loss": 0.8626,
+      "step": 490
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922624681799888e-05,
+      "loss": 0.8339,
+      "step": 491
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922236593080964e-05,
+      "loss": 0.8871,
+      "step": 492
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218475729022498e-05,
+      "loss": 0.8478,
+      "step": 493
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9214576216566557e-05,
+      "loss": 0.8486,
+      "step": 494
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921066739738033e-05,
+      "loss": 0.858,
+      "step": 495
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9206749275411724e-05,
+      "loss": 0.8472,
+      "step": 496
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202821854618052e-05,
+      "loss": 0.858,
+      "step": 497
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198885138966007e-05,
+      "loss": 0.851,
+      "step": 498
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194939132431682e-05,
+      "loss": 0.8748,
+      "step": 499
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919098383900054e-05,
+      "loss": 0.8688,
+      "step": 500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187019262667434e-05,
+      "loss": 0.8353,
+      "step": 501
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183045407436592e-05,
+      "loss": 0.8717,
+      "step": 502
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179062277321606e-05,
+      "loss": 0.8468,
+      "step": 503
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175069876345445e-05,
+      "loss": 0.8557,
+      "step": 504
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917106820854044e-05,
+      "loss": 0.8566,
+      "step": 505
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167057277948272e-05,
+      "loss": 0.8635,
+      "step": 506
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163037088619986e-05,
+      "loss": 0.8682,
+      "step": 507
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159007644615983e-05,
+      "loss": 0.8586,
+      "step": 508
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154968950006e-05,
+      "loss": 0.8793,
+      "step": 509
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915092100886912e-05,
+      "loss": 0.8513,
+      "step": 510
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9146863825293776e-05,
+      "loss": 0.833,
+      "step": 511
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142797403377717e-05,
+      "loss": 0.8284,
+      "step": 512
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913872174722804e-05,
+      "loss": 0.8843,
+      "step": 513
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134636860961157e-05,
+      "loss": 0.8568,
+      "step": 514
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913054274870281e-05,
+      "loss": 0.8668,
+      "step": 515
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126439414588056e-05,
+      "loss": 0.849,
+      "step": 516
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912232686276126e-05,
+      "loss": 0.8874,
+      "step": 517
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118205097376113e-05,
+      "loss": 0.8567,
+      "step": 518
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114074122595598e-05,
+      "loss": 0.8631,
+      "step": 519
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109933942591998e-05,
+      "loss": 0.848,
+      "step": 520
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91057845615469e-05,
+      "loss": 0.8407,
+      "step": 521
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910162598365119e-05,
+      "loss": 0.8518,
+      "step": 522
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909745821310503e-05,
+      "loss": 0.8622,
+      "step": 523
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909328125411787e-05,
+      "loss": 0.8168,
+      "step": 524
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089095110908444e-05,
+      "loss": 0.8607,
+      "step": 525
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9084899787704762e-05,
+      "loss": 0.8511,
+      "step": 526
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080695288744107e-05,
+      "loss": 0.8483,
+      "step": 527
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907648161827302e-05,
+      "loss": 0.8481,
+      "step": 528
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072258780547317e-05,
+      "loss": 0.8454,
+      "step": 529
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068026779832063e-05,
+      "loss": 0.8588,
+      "step": 530
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063785620401585e-05,
+      "loss": 0.8602,
+      "step": 531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905953530653946e-05,
+      "loss": 0.8613,
+      "step": 532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905527584253851e-05,
+      "loss": 0.866,
+      "step": 533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905100723270079e-05,
+      "loss": 0.8568,
+      "step": 534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904672948133761e-05,
+      "loss": 0.8712,
+      "step": 535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042442592769493e-05,
+      "loss": 0.8591,
+      "step": 536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9038146571326205e-05,
+      "loss": 0.8299,
+      "step": 537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033841421346734e-05,
+      "loss": 0.8628,
+      "step": 538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902952714717928e-05,
+      "loss": 0.8635,
+      "step": 539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025203753181267e-05,
+      "loss": 0.8858,
+      "step": 540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020871243719324e-05,
+      "loss": 0.8762,
+      "step": 541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016529623169294e-05,
+      "loss": 0.8574,
+      "step": 542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012178895916215e-05,
+      "loss": 0.8278,
+      "step": 543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007819066354325e-05,
+      "loss": 0.8453,
+      "step": 544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003450138887056e-05,
+      "loss": 0.8243,
+      "step": 545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899907211792703e-05,
+      "loss": 0.8275,
+      "step": 546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994685007896053e-05,
+      "loss": 0.8703,
+      "step": 547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8990288813225105e-05,
+      "loss": 0.8514,
+      "step": 548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985883538354352e-05,
+      "loss": 0.8521,
+      "step": 549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981469187733123e-05,
+      "loss": 0.8775,
+      "step": 550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977045765819917e-05,
+      "loss": 0.8501,
+      "step": 551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897261327708239e-05,
+      "loss": 0.8511,
+      "step": 552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896817172599737e-05,
+      "loss": 0.8466,
+      "step": 553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963721117050816e-05,
+      "loss": 0.8507,
+      "step": 554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959261454737854e-05,
+      "loss": 0.8612,
+      "step": 555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954792743562745e-05,
+      "loss": 0.8589,
+      "step": 556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895031498803889e-05,
+      "loss": 0.846,
+      "step": 557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894582819268883e-05,
+      "loss": 0.8827,
+      "step": 558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894133236204423e-05,
+      "loss": 0.8662,
+      "step": 559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893682750064588e-05,
+      "loss": 0.835,
+      "step": 560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932313613043694e-05,
+      "loss": 0.8554,
+      "step": 561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927790703796713e-05,
+      "loss": 0.8801,
+      "step": 562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923258777473067e-05,
+      "loss": 0.8469,
+      "step": 563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891871783865001e-05,
+      "loss": 0.8424,
+      "step": 564
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89141678919139e-05,
+      "loss": 0.8617,
+      "step": 565
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909608941860174e-05,
+      "loss": 0.871,
+      "step": 566
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905040993093386e-05,
+      "loss": 0.8564,
+      "step": 567
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890046405022717e-05,
+      "loss": 0.8591,
+      "step": 568
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895878117884238e-05,
+      "loss": 0.8463,
+      "step": 569
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891283200696383e-05,
+      "loss": 0.8473,
+      "step": 570
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8886679303304483e-05,
+      "loss": 0.8762,
+      "step": 571
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888206643035847e-05,
+      "loss": 0.8509,
+      "step": 572
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877444586517356e-05,
+      "loss": 0.8552,
+      "step": 573
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8872813776449206e-05,
+      "loss": 0.8328,
+      "step": 574
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868174004831143e-05,
+      "loss": 0.8574,
+      "step": 575
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863525276349336e-05,
+      "loss": 0.8458,
+      "step": 576
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858867595699017e-05,
+      "loss": 0.8231,
+      "step": 577
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885420096758443e-05,
+      "loss": 0.8458,
+      "step": 578
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849525396718884e-05,
+      "loss": 0.8694,
+      "step": 579
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884484088782471e-05,
+      "loss": 0.8396,
+      "step": 580
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840147445633268e-05,
+      "loss": 0.8267,
+      "step": 581
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883544507488493e-05,
+      "loss": 0.8438,
+      "step": 582
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830733780329105e-05,
+      "loss": 0.8609,
+      "step": 583
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826013566724196e-05,
+      "loss": 0.8576,
+      "step": 584
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882128443883763e-05,
+      "loss": 0.858,
+      "step": 585
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8816546401445825e-05,
+      "loss": 0.839,
+      "step": 586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881179945933421e-05,
+      "loss": 0.8438,
+      "step": 587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880704361729719e-05,
+      "loss": 0.831,
+      "step": 588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880227888013818e-05,
+      "loss": 0.8457,
+      "step": 589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797505252669563e-05,
+      "loss": 0.8488,
+      "step": 590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879272273971271e-05,
+      "loss": 0.8478,
+      "step": 591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787931346097962e-05,
+      "loss": 0.8595,
+      "step": 592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8783131076664635e-05,
+      "loss": 0.8482,
+      "step": 593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778321936260994e-05,
+      "loss": 0.8631,
+      "step": 594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877350392974429e-05,
+      "loss": 0.8411,
+      "step": 595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768677061980703e-05,
+      "loss": 0.8488,
+      "step": 596
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763841337845376e-05,
+      "loss": 0.8469,
+      "step": 597
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758996762222398e-05,
+      "loss": 0.8494,
+      "step": 598
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754143340004797e-05,
+      "loss": 0.8264,
+      "step": 599
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749281076094524e-05,
+      "loss": 0.8773,
+      "step": 600
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874440997540248e-05,
+      "loss": 0.855,
+      "step": 601
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739530042848474e-05,
+      "loss": 0.8274,
+      "step": 602
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734641283361243e-05,
+      "loss": 0.8527,
+      "step": 603
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872974370187844e-05,
+      "loss": 0.8362,
+      "step": 604
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724837303346634e-05,
+      "loss": 0.8424,
+      "step": 605
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719922092721282e-05,
+      "loss": 0.8569,
+      "step": 606
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714998074966753e-05,
+      "loss": 0.8617,
+      "step": 607
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710065255056315e-05,
+      "loss": 0.8484,
+      "step": 608
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705123637972112e-05,
+      "loss": 0.8604,
+      "step": 609
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700173228705188e-05,
+      "loss": 0.8509,
+      "step": 610
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695214032255457e-05,
+      "loss": 0.8605,
+      "step": 611
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690246053631716e-05,
+      "loss": 0.8392,
+      "step": 612
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685269297851625e-05,
+      "loss": 0.8517,
+      "step": 613
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868028376994171e-05,
+      "loss": 0.8646,
+      "step": 614
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675289474937364e-05,
+      "loss": 0.8515,
+      "step": 615
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670286417882825e-05,
+      "loss": 0.8713,
+      "step": 616
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866527460383119e-05,
+      "loss": 0.8544,
+      "step": 617
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.8591,
+      "step": 618
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655224724993202e-05,
+      "loss": 0.8602,
+      "step": 619
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650186670357238e-05,
+      "loss": 0.8528,
+      "step": 620
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645139879024943e-05,
+      "loss": 0.8309,
+      "step": 621
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864008435609357e-05,
+      "loss": 0.843,
+      "step": 622
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863502010666921e-05,
+      "loss": 0.8463,
+      "step": 623
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629947135866756e-05,
+      "loss": 0.8474,
+      "step": 624
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624865448809914e-05,
+      "loss": 0.8399,
+      "step": 625
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619775050631194e-05,
+      "loss": 0.8324,
+      "step": 626
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86146759464719e-05,
+      "loss": 0.8444,
+      "step": 627
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609568141482134e-05,
+      "loss": 0.8378,
+      "step": 628
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860445164082078e-05,
+      "loss": 0.8259,
+      "step": 629
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599326449655515e-05,
+      "loss": 0.8491,
+      "step": 630
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859419257316278e-05,
+      "loss": 0.853,
+      "step": 631
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85890500165278e-05,
+      "loss": 0.8597,
+      "step": 632
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8583898784944553e-05,
+      "loss": 0.8348,
+      "step": 633
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85787388836158e-05,
+      "loss": 0.859,
+      "step": 634
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857357031775304e-05,
+      "loss": 0.8409,
+      "step": 635
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8568393092576524e-05,
+      "loss": 0.8565,
+      "step": 636
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563207213315264e-05,
+      "loss": 0.8366,
+      "step": 637
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558012685206995e-05,
+      "loss": 0.849,
+      "step": 638
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85528095134982e-05,
+      "loss": 0.8656,
+      "step": 639
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547597703444085e-05,
+      "loss": 0.8828,
+      "step": 640
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854237726030858e-05,
+      "loss": 0.848,
+      "step": 641
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537148189364343e-05,
+      "loss": 0.8612,
+      "step": 642
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531910495892737e-05,
+      "loss": 0.837,
+      "step": 643
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526664185183837e-05,
+      "loss": 0.857,
+      "step": 644
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521409262536422e-05,
+      "loss": 0.846,
+      "step": 645
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516145733257974e-05,
+      "loss": 0.8295,
+      "step": 646
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851087360266465e-05,
+      "loss": 0.849,
+      "step": 647
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505592876081318e-05,
+      "loss": 0.84,
+      "step": 648
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850030355884151e-05,
+      "loss": 0.8461,
+      "step": 649
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495005656287445e-05,
+      "loss": 0.8326,
+      "step": 650
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848969917377001e-05,
+      "loss": 0.8498,
+      "step": 651
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848438411664875e-05,
+      "loss": 0.8276,
+      "step": 652
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847906049029189e-05,
+      "loss": 0.8298,
+      "step": 653
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8473728300076284e-05,
+      "loss": 0.8473,
+      "step": 654
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8468387551387456e-05,
+      "loss": 0.8604,
+      "step": 655
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463038249619566e-05,
+      "loss": 0.8343,
+      "step": 656
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457680400175417e-05,
+      "loss": 0.8409,
+      "step": 657
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845231400846643e-05,
+      "loss": 0.8438,
+      "step": 658
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844693907991268e-05,
+      "loss": 0.8463,
+      "step": 659
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441555619942845e-05,
+      "loss": 0.8505,
+      "step": 660
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843616363399422e-05,
+      "loss": 0.8465,
+      "step": 661
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843076312751272e-05,
+      "loss": 0.8351,
+      "step": 662
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425354105952857e-05,
+      "loss": 0.8386,
+      "step": 663
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419936574777755e-05,
+      "loss": 0.8441,
+      "step": 664
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414510539459118e-05,
+      "loss": 0.8363,
+      "step": 665
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840907600547725e-05,
+      "loss": 0.8332,
+      "step": 666
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403632978321037e-05,
+      "loss": 0.8572,
+      "step": 667
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398181463487936e-05,
+      "loss": 0.8634,
+      "step": 668
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8392721466483983e-05,
+      "loss": 0.8394,
+      "step": 669
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838725299282379e-05,
+      "loss": 0.843,
+      "step": 670
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381776048030504e-05,
+      "loss": 0.8558,
+      "step": 671
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8376290637635857e-05,
+      "loss": 0.8347,
+      "step": 672
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837079676718011e-05,
+      "loss": 0.8488,
+      "step": 673
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365294442212085e-05,
+      "loss": 0.8304,
+      "step": 674
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835978366828913e-05,
+      "loss": 0.8624,
+      "step": 675
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354264450977132e-05,
+      "loss": 0.8365,
+      "step": 676
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348736795850508e-05,
+      "loss": 0.8366,
+      "step": 677
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343200708492192e-05,
+      "loss": 0.826,
+      "step": 678
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337656194493637e-05,
+      "loss": 0.8328,
+      "step": 679
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833210325945481e-05,
+      "loss": 0.8566,
+      "step": 680
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832654190898418e-05,
+      "loss": 0.8305,
+      "step": 681
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832097214869871e-05,
+      "loss": 0.8531,
+      "step": 682
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831539398422387e-05,
+      "loss": 0.8571,
+      "step": 683
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830980742119361e-05,
+      "loss": 0.8446,
+      "step": 684
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830421246525036e-05,
+      "loss": 0.8353,
+      "step": 685
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298609122045033e-05,
+      "loss": 0.8444,
+      "step": 686
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292997397237013e-05,
+      "loss": 0.8508,
+      "step": 687
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287377296494143e-05,
+      "loss": 0.8387,
+      "step": 688
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828174882549273e-05,
+      "loss": 0.851,
+      "step": 689
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827611198991754e-05,
+      "loss": 0.8423,
+      "step": 690
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270466795461778e-05,
+      "loss": 0.826,
+      "step": 691
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264813247827092e-05,
+      "loss": 0.8509,
+      "step": 692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825915135272358e-05,
+      "loss": 0.8399,
+      "step": 693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825348111586975e-05,
+      "loss": 0.868,
+      "step": 694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824780254299255e-05,
+      "loss": 0.8448,
+      "step": 695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242115639827354e-05,
+      "loss": 0.8389,
+      "step": 696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236420412117923e-05,
+      "loss": 0.8384,
+      "step": 697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8230716865616452e-05,
+      "loss": 0.8427,
+      "step": 698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8225005006083526e-05,
+      "loss": 0.8341,
+      "step": 699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821928483928813e-05,
+      "loss": 0.8405,
+      "step": 700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821355637100763e-05,
+      "loss": 0.8379,
+      "step": 701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207819607027795e-05,
+      "loss": 0.8551,
+      "step": 702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8202074553142756e-05,
+      "loss": 0.8315,
+      "step": 703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196321215155023e-05,
+      "loss": 0.827,
+      "step": 704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190559598875475e-05,
+      "loss": 0.8587,
+      "step": 705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818478971012334e-05,
+      "loss": 0.8546,
+      "step": 706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8179011554726228e-05,
+      "loss": 0.8427,
+      "step": 707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173225138520064e-05,
+      "loss": 0.8251,
+      "step": 708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167430467349145e-05,
+      "loss": 0.8607,
+      "step": 709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161627547066088e-05,
+      "loss": 0.8692,
+      "step": 710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815581638353185e-05,
+      "loss": 0.8148,
+      "step": 711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149996982615715e-05,
+      "loss": 0.8369,
+      "step": 712
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814416935019528e-05,
+      "loss": 0.8212,
+      "step": 713
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813833349215645e-05,
+      "loss": 0.8229,
+      "step": 714
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813248941439346e-05,
+      "loss": 0.8403,
+      "step": 715
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812663712280883e-05,
+      "loss": 0.8446,
+      "step": 716
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8120776623313377e-05,
+      "loss": 0.8203,
+      "step": 717
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114907921826218e-05,
+      "loss": 0.8252,
+      "step": 718
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8109031024274735e-05,
+      "loss": 0.8421,
+      "step": 719
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103145936594612e-05,
+      "loss": 0.828,
+      "step": 720
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097252664729782e-05,
+      "loss": 0.8595,
+      "step": 721
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091351214632464e-05,
+      "loss": 0.8192,
+      "step": 722
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8085441592263123e-05,
+      "loss": 0.8543,
+      "step": 723
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079523803590482e-05,
+      "loss": 0.8408,
+      "step": 724
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8073597854591515e-05,
+      "loss": 0.8588,
+      "step": 725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067663751251436e-05,
+      "loss": 0.8372,
+      "step": 726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061721499563696e-05,
+      "loss": 0.8577,
+      "step": 727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805577110552997e-05,
+      "loss": 0.8259,
+      "step": 728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804981257516017e-05,
+      "loss": 0.8427,
+      "step": 729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043845914472404e-05,
+      "loss": 0.8416,
+      "step": 730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037871129493018e-05,
+      "loss": 0.8513,
+      "step": 731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803188822625654e-05,
+      "loss": 0.8286,
+      "step": 732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025897210805712e-05,
+      "loss": 0.8342,
+      "step": 733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019898089191466e-05,
+      "loss": 0.844,
+      "step": 734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013890867472923e-05,
+      "loss": 0.8563,
+      "step": 735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007875551717372e-05,
+      "loss": 0.8297,
+      "step": 736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001852148000298e-05,
+      "loss": 0.8332,
+      "step": 737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799582066240534e-05,
+      "loss": 0.8393,
+      "step": 738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989781101024305e-05,
+      "loss": 0.8513,
+      "step": 739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798373346995715e-05,
+      "loss": 0.8423,
+      "step": 740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977677775311994e-05,
+      "loss": 0.8575,
+      "step": 741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971614023205092e-05,
+      "loss": 0.8506,
+      "step": 742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965542219760842e-05,
+      "loss": 0.8216,
+      "step": 743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959462371111764e-05,
+      "loss": 0.8494,
+      "step": 744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953374483398517e-05,
+      "loss": 0.8307,
+      "step": 745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947278562769873e-05,
+      "loss": 0.8356,
+      "step": 746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941174615382713e-05,
+      "loss": 0.8441,
+      "step": 747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793506264740203e-05,
+      "loss": 0.868,
+      "step": 748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792894266500092e-05,
+      "loss": 0.8307,
+      "step": 749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792281467436057e-05,
+      "loss": 0.8432,
+      "step": 750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916678681670256e-05,
+      "loss": 0.84,
+      "step": 751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910534693127336e-05,
+      "loss": 0.8357,
+      "step": 752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904382714937243e-05,
+      "loss": 0.8146,
+      "step": 753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898222753313483e-05,
+      "loss": 0.8211,
+      "step": 754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892054814477617e-05,
+      "loss": 0.8038,
+      "step": 755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885878904659276e-05,
+      "loss": 0.8488,
+      "step": 756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787969503009613e-05,
+      "loss": 0.8387,
+      "step": 757
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78735031970339e-05,
+      "loss": 0.8316,
+      "step": 758
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7867303411726343e-05,
+      "loss": 0.8134,
+      "step": 759
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861095680435245e-05,
+      "loss": 0.8432,
+      "step": 760
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785488000943042e-05,
+      "loss": 0.8191,
+      "step": 761
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848656404989712e-05,
+      "loss": 0.8323,
+      "step": 762
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784242487339895e-05,
+      "loss": 0.8603,
+      "step": 763
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836185420952e-05,
+      "loss": 0.8542,
+      "step": 764
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829938053950707e-05,
+      "loss": 0.8609,
+      "step": 765
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782368277870492e-05,
+      "loss": 0.8449,
+      "step": 766
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7817419601532474e-05,
+      "loss": 0.8237,
+      "step": 767
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811148528759183e-05,
+      "loss": 0.8392,
+      "step": 768
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7804869566718833e-05,
+      "loss": 0.8237,
+      "step": 769
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798582721753186e-05,
+      "loss": 0.8386,
+      "step": 770
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792288000211958e-05,
+      "loss": 0.8264,
+      "step": 771
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778598540845282e-05,
+      "loss": 0.8053,
+      "step": 772
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779674952841407e-05,
+      "loss": 0.8415,
+      "step": 773
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773356639751274e-05,
+      "loss": 0.8362,
+      "step": 774
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7767030475563927e-05,
+      "loss": 0.8086,
+      "step": 775
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760696466668794e-05,
+      "loss": 0.8557,
+      "step": 776
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754354619463236e-05,
+      "loss": 0.8148,
+      "step": 777
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774800494035252e-05,
+      "loss": 0.8307,
+      "step": 778
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741647435749823e-05,
+      "loss": 0.8448,
+      "step": 779
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7735282112076242e-05,
+      "loss": 0.837,
+      "step": 780
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728908975760753e-05,
+      "loss": 0.8437,
+      "step": 781
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772252803324023e-05,
+      "loss": 0.8089,
+      "step": 782
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771613929095943e-05,
+      "loss": 0.8285,
+      "step": 783
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770974275537099e-05,
+      "loss": 0.8211,
+      "step": 784
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7703338432935417e-05,
+      "loss": 0.8301,
+      "step": 785
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7696926330121077e-05,
+      "loss": 0.8486,
+      "step": 786
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690506453404207e-05,
+      "loss": 0.8448,
+      "step": 787
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7684078809268887e-05,
+      "loss": 0.8361,
+      "step": 788
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767764340420704e-05,
+      "loss": 0.834,
+      "step": 789
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767120024471843e-05,
+      "loss": 0.8312,
+      "step": 790
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766474933731066e-05,
+      "loss": 0.8477,
+      "step": 791
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765829068849915e-05,
+      "loss": 0.8574,
+      "step": 792
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651824304807135e-05,
+      "loss": 0.8107,
+      "step": 793
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645350192765677e-05,
+      "loss": 0.8134,
+      "step": 794
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638868358913627e-05,
+      "loss": 0.8112,
+      "step": 795
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763237880979765e-05,
+      "loss": 0.8361,
+      "step": 796
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7625881551972196e-05,
+      "loss": 0.826,
+      "step": 797
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619376591999496e-05,
+      "loss": 0.8301,
+      "step": 798
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612863936449568e-05,
+      "loss": 0.8204,
+      "step": 799
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606343591900202e-05,
+      "loss": 0.8404,
+      "step": 800
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7599815564936947e-05,
+      "loss": 0.8239,
+      "step": 801
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593279862153125e-05,
+      "loss": 0.8375,
+      "step": 802
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758673649014979e-05,
+      "loss": 0.8261,
+      "step": 803
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580185455535766e-05,
+      "loss": 0.8268,
+      "step": 804
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7573626764927592e-05,
+      "loss": 0.8113,
+      "step": 805
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756706042494956e-05,
+      "loss": 0.8314,
+      "step": 806
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560486442233676e-05,
+      "loss": 0.8261,
+      "step": 807
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553904823419667e-05,
+      "loss": 0.8513,
+      "step": 808
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547315575154977e-05,
+      "loss": 0.8414,
+      "step": 809
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754071870409475e-05,
+      "loss": 0.8426,
+      "step": 810
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7534114216901838e-05,
+      "loss": 0.8401,
+      "step": 811
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752750212024677e-05,
+      "loss": 0.8314,
+      "step": 812
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7520882420807772e-05,
+      "loss": 0.833,
+      "step": 813
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514255125270748e-05,
+      "loss": 0.8476,
+      "step": 814
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507620240329278e-05,
+      "loss": 0.8252,
+      "step": 815
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500977772684594e-05,
+      "loss": 0.8249,
+      "step": 816
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749432772904559e-05,
+      "loss": 0.849,
+      "step": 817
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487670116128834e-05,
+      "loss": 0.8274,
+      "step": 818
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481004940658503e-05,
+      "loss": 0.831,
+      "step": 819
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474332209366437e-05,
+      "loss": 0.8312,
+      "step": 820
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467651928992104e-05,
+      "loss": 0.8091,
+      "step": 821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7460964106282587e-05,
+      "loss": 0.8155,
+      "step": 822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7454268747992596e-05,
+      "loss": 0.8368,
+      "step": 823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744756586088445e-05,
+      "loss": 0.8409,
+      "step": 824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744085545172807e-05,
+      "loss": 0.8185,
+      "step": 825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7434137527300975e-05,
+      "loss": 0.8303,
+      "step": 826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742741209438827e-05,
+      "loss": 0.845,
+      "step": 827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742067915978266e-05,
+      "loss": 0.8287,
+      "step": 828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7413938730284404e-05,
+      "loss": 0.8247,
+      "step": 829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407190812701348e-05,
+      "loss": 0.8029,
+      "step": 830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740043541384889e-05,
+      "loss": 0.8304,
+      "step": 831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393672540549992e-05,
+      "loss": 0.8378,
+      "step": 832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386902199635166e-05,
+      "loss": 0.8371,
+      "step": 833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380124397942454e-05,
+      "loss": 0.8197,
+      "step": 834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7373339142317448e-05,
+      "loss": 0.8251,
+      "step": 835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736654643961326e-05,
+      "loss": 0.8335,
+      "step": 836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7359746296690524e-05,
+      "loss": 0.8189,
+      "step": 837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352938720417397e-05,
+      "loss": 0.8221,
+      "step": 838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734612371766953e-05,
+      "loss": 0.816,
+      "step": 839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339301295330087e-05,
+      "loss": 0.8405,
+      "step": 840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332471460289716e-05,
+      "loss": 0.8236,
+      "step": 841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732563421944656e-05,
+      "loss": 0.8354,
+      "step": 842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731878957970624e-05,
+      "loss": 0.8404,
+      "step": 843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311937547981843e-05,
+      "loss": 0.8327,
+      "step": 844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7305078131193933e-05,
+      "loss": 0.8054,
+      "step": 845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729821133627052e-05,
+      "loss": 0.8458,
+      "step": 846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729133717014708e-05,
+      "loss": 0.83,
+      "step": 847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7284455639766523e-05,
+      "loss": 0.8352,
+      "step": 848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72775667520792e-05,
+      "loss": 0.8084,
+      "step": 849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72706705140429e-05,
+      "loss": 0.8194,
+      "step": 850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7263766932622824e-05,
+      "loss": 0.8494,
+      "step": 851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7256856014791598e-05,
+      "loss": 0.8301,
+      "step": 852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249937767529253e-05,
+      "loss": 0.838,
+      "step": 853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243012197823224e-05,
+      "loss": 0.8366,
+      "step": 854
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236079312668347e-05,
+      "loss": 0.8065,
+      "step": 855
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229139119066843e-05,
+      "loss": 0.8344,
+      "step": 856
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722219162402831e-05,
+      "loss": 0.8073,
+      "step": 857
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721523683456972e-05,
+      "loss": 0.8262,
+      "step": 858
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208274757715425e-05,
+      "loss": 0.8502,
+      "step": 859
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201305400497125e-05,
+      "loss": 0.8229,
+      "step": 860
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719432876995388e-05,
+      "loss": 0.8028,
+      "step": 861
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718734487313209e-05,
+      "loss": 0.8101,
+      "step": 862
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180353717085497e-05,
+      "loss": 0.8334,
+      "step": 863
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7173355308875176e-05,
+      "loss": 0.8224,
+      "step": 864
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7166349655569524e-05,
+      "loss": 0.8273,
+      "step": 865
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715933676424426e-05,
+      "loss": 0.8518,
+      "step": 866
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715231664198241e-05,
+      "loss": 0.8273,
+      "step": 867
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145289295874304e-05,
+      "loss": 0.8341,
+      "step": 868
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138254733017565e-05,
+      "loss": 0.8119,
+      "step": 869
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7131212960517113e-05,
+      "loss": 0.8328,
+      "step": 870
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7124163985485138e-05,
+      "loss": 0.838,
+      "step": 871
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7117107815041114e-05,
+      "loss": 0.8219,
+      "step": 872
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711004445631178e-05,
+      "loss": 0.8525,
+      "step": 873
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102973916431137e-05,
+      "loss": 0.8156,
+      "step": 874
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709589620254043e-05,
+      "loss": 0.8228,
+      "step": 875
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088811321788167e-05,
+      "loss": 0.8173,
+      "step": 876
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081719281330072e-05,
+      "loss": 0.8277,
+      "step": 877
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074620088329123e-05,
+      "loss": 0.8083,
+      "step": 878
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067513749955502e-05,
+      "loss": 0.8306,
+      "step": 879
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060400273386623e-05,
+      "loss": 0.8096,
+      "step": 880
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7053279665807105e-05,
+      "loss": 0.8312,
+      "step": 881
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046151934408762e-05,
+      "loss": 0.8227,
+      "step": 882
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703901708639061e-05,
+      "loss": 0.8126,
+      "step": 883
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031875128958856e-05,
+      "loss": 0.8415,
+      "step": 884
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7024726069326883e-05,
+      "loss": 0.8021,
+      "step": 885
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701756991471525e-05,
+      "loss": 0.8365,
+      "step": 886
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7010406672351672e-05,
+      "loss": 0.8123,
+      "step": 887
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003236349471034e-05,
+      "loss": 0.8027,
+      "step": 888
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6996058953315372e-05,
+      "loss": 0.8253,
+      "step": 889
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6988874491133855e-05,
+      "loss": 0.8356,
+      "step": 890
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69816829701828e-05,
+      "loss": 0.8432,
+      "step": 891
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6974484397725653e-05,
+      "loss": 0.8201,
+      "step": 892
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967278781032976e-05,
+      "loss": 0.837,
+      "step": 893
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696006612738245e-05,
+      "loss": 0.8402,
+      "step": 894
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695284644405886e-05,
+      "loss": 0.8301,
+      "step": 895
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694561973835409e-05,
+      "loss": 0.8189,
+      "step": 896
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938386017567123e-05,
+      "loss": 0.8016,
+      "step": 897
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931145289004024e-05,
+      "loss": 0.8339,
+      "step": 898
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923897559977934e-05,
+      "loss": 0.8299,
+      "step": 899
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691664283780906e-05,
+      "loss": 0.8321,
+      "step": 900
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6909381129824695e-05,
+      "loss": 0.8464,
+      "step": 901
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690211244335915e-05,
+      "loss": 0.8356,
+      "step": 902
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894836785753815e-05,
+      "loss": 0.8495,
+      "step": 903
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887554164357113e-05,
+      "loss": 0.8114,
+      "step": 904
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688026458652449e-05,
+      "loss": 0.8151,
+      "step": 905
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687296805961844e-05,
+      "loss": 0.8505,
+      "step": 906
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686566459100845e-05,
+      "loss": 0.8302,
+      "step": 907
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685835418807103e-05,
+      "loss": 0.821,
+      "step": 908
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851036858189696e-05,
+      "loss": 0.8001,
+      "step": 909
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684371260875496e-05,
+      "loss": 0.8319,
+      "step": 910
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683638144716432e-05,
+      "loss": 0.797,
+      "step": 911
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829043380822254e-05,
+      "loss": 0.8283,
+      "step": 912
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821698417140216e-05,
+      "loss": 0.8161,
+      "step": 913
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814346563536626e-05,
+      "loss": 0.8255,
+      "step": 914
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680698782743686e-05,
+      "loss": 0.811,
+      "step": 915
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6799622216273256e-05,
+      "loss": 0.8248,
+      "step": 916
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679224973748508e-05,
+      "loss": 0.8152,
+      "step": 917
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784870398518546e-05,
+      "loss": 0.8191,
+      "step": 918
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777484206826793e-05,
+      "loss": 0.8032,
+      "step": 919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770091169869882e-05,
+      "loss": 0.8214,
+      "step": 920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676269129511478e-05,
+      "loss": 0.7981,
+      "step": 921
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675528459003538e-05,
+      "loss": 0.8226,
+      "step": 922
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747871062112445e-05,
+      "loss": 0.8246,
+      "step": 923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674045071883366e-05,
+      "loss": 0.81,
+      "step": 924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673302356769357e-05,
+      "loss": 0.8253,
+      "step": 925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6725589616193607e-05,
+      "loss": 0.8109,
+      "step": 926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718148871842064e-05,
+      "loss": 0.8303,
+      "step": 927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671070134215411e-05,
+      "loss": 0.8193,
+      "step": 928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703247034651743e-05,
+      "loss": 0.8152,
+      "step": 929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669578595686383e-05,
+      "loss": 0.8186,
+      "step": 930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6688318116326064e-05,
+      "loss": 0.8422,
+      "step": 931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668084352058097e-05,
+      "loss": 0.7982,
+      "step": 932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6673362177177898e-05,
+      "loss": 0.8551,
+      "step": 933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665874093673006e-05,
+      "loss": 0.8156,
+      "step": 934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6658379277629274e-05,
+      "loss": 0.8437,
+      "step": 935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6650877736616467e-05,
+      "loss": 0.8141,
+      "step": 936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643369478211145e-05,
+      "loss": 0.8279,
+      "step": 937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663585450999667e-05,
+      "loss": 0.7998,
+      "step": 938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662833283956315e-05,
+      "loss": 0.8345,
+      "step": 939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6620804474507487e-05,
+      "loss": 0.8396,
+      "step": 940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613269422433336e-05,
+      "loss": 0.844,
+      "step": 941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605727690951104e-05,
+      "loss": 0.8301,
+      "step": 942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659817928767795e-05,
+      "loss": 0.7975,
+      "step": 943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6590624220237764e-05,
+      "loss": 0.8391,
+      "step": 944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583062496261178e-05,
+      "loss": 0.8256,
+      "step": 945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575494123385535e-05,
+      "loss": 0.8251,
+      "step": 946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567919109254897e-05,
+      "loss": 0.8122,
+      "step": 947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560337461520036e-05,
+      "loss": 0.8141,
+      "step": 948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552749187838425e-05,
+      "loss": 0.8312,
+      "step": 949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545154295874226e-05,
+      "loss": 0.8103,
+      "step": 950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653755279329829e-05,
+      "loss": 0.8183,
+      "step": 951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529944687788134e-05,
+      "loss": 0.8131,
+      "step": 952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522329987027956e-05,
+      "loss": 0.8036,
+      "step": 953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6514708698708613e-05,
+      "loss": 0.8204,
+      "step": 954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650708083052761e-05,
+      "loss": 0.8026,
+      "step": 955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64994463901891e-05,
+      "loss": 0.8247,
+      "step": 956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491805385403878e-05,
+      "loss": 0.8091,
+      "step": 957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484157823889364e-05,
+      "loss": 0.8026,
+      "step": 958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476503713369602e-05,
+      "loss": 0.8463,
+      "step": 959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468843061575245e-05,
+      "loss": 0.8188,
+      "step": 960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461175876243565e-05,
+      "loss": 0.8106,
+      "step": 961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645350216511843e-05,
+      "loss": 0.8116,
+      "step": 962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445821935950287e-05,
+      "loss": 0.8331,
+      "step": 963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438135196496174e-05,
+      "loss": 0.7934,
+      "step": 964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643044195451971e-05,
+      "loss": 0.808,
+      "step": 965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422742217791076e-05,
+      "loss": 0.8348,
+      "step": 966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641503599408701e-05,
+      "loss": 0.839,
+      "step": 967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407323291190806e-05,
+      "loss": 0.8301,
+      "step": 968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63996041168923e-05,
+      "loss": 0.812,
+      "step": 969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639187847898787e-05,
+      "loss": 0.8254,
+      "step": 970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384146385280407e-05,
+      "loss": 0.8051,
+      "step": 971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376407843579344e-05,
+      "loss": 0.8059,
+      "step": 972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368662861700608e-05,
+      "loss": 0.8275,
+      "step": 973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636091144746664e-05,
+      "loss": 0.8002,
+      "step": 974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353153608706373e-05,
+      "loss": 0.802,
+      "step": 975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345389353255234e-05,
+      "loss": 0.8084,
+      "step": 976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633761868895513e-05,
+      "loss": 0.8209,
+      "step": 977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329841623654434e-05,
+      "loss": 0.8049,
+      "step": 978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632205816520799e-05,
+      "loss": 0.8052,
+      "step": 979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.63142683214771e-05,
+      "loss": 0.8403,
+      "step": 980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6306472100329515e-05,
+      "loss": 0.8016,
+      "step": 981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6298669509639418e-05,
+      "loss": 0.822,
+      "step": 982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629086055728744e-05,
+      "loss": 0.8224,
+      "step": 983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283045251160627e-05,
+      "loss": 0.8165,
+      "step": 984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275223599152444e-05,
+      "loss": 0.8105,
+      "step": 985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626739560916277e-05,
+      "loss": 0.8356,
+      "step": 986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259561289097877e-05,
+      "loss": 0.8045,
+      "step": 987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251720646870443e-05,
+      "loss": 0.7919,
+      "step": 988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6243873690399518e-05,
+      "loss": 0.8105,
+      "step": 989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236020427610536e-05,
+      "loss": 0.813,
+      "step": 990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.62281608664353e-05,
+      "loss": 0.8264,
+      "step": 991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622029501481197e-05,
+      "loss": 0.7994,
+      "step": 992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621242288068507e-05,
+      "loss": 0.8048,
+      "step": 993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6204544472005458e-05,
+      "loss": 0.8018,
+      "step": 994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196659796730335e-05,
+      "loss": 0.8069,
+      "step": 995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6188768862823233e-05,
+      "loss": 0.7856,
+      "step": 996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180871678253996e-05,
+      "loss": 0.8294,
+      "step": 997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617296825099879e-05,
+      "loss": 0.8307,
+      "step": 998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165058589040092e-05,
+      "loss": 0.8055,
+      "step": 999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6157142700366654e-05,
+      "loss": 0.8119,
+      "step": 1000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6149220592973542e-05,
+      "loss": 0.8191,
+      "step": 1001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141292274862085e-05,
+      "loss": 0.8059,
+      "step": 1002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613335775403989e-05,
+      "loss": 0.8312,
+      "step": 1003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6125417038520835e-05,
+      "loss": 0.8213,
+      "step": 1004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6117470136325054e-05,
+      "loss": 0.8234,
+      "step": 1005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109517055478915e-05,
+      "loss": 0.7897,
+      "step": 1006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610155780401504e-05,
+      "loss": 0.8183,
+      "step": 1007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6093592389972284e-05,
+      "loss": 0.8269,
+      "step": 1008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085620821395722e-05,
+      "loss": 0.803,
+      "step": 1009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077643106336643e-05,
+      "loss": 0.8069,
+      "step": 1010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069659252852546e-05,
+      "loss": 0.8274,
+      "step": 1011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6061669269007128e-05,
+      "loss": 0.8187,
+      "step": 1012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6053673162870285e-05,
+      "loss": 0.831,
+      "step": 1013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604567094251809e-05,
+      "loss": 0.7893,
+      "step": 1014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037662616032795e-05,
+      "loss": 0.8244,
+      "step": 1015
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602964819150281e-05,
+      "loss": 0.8239,
+      "step": 1016
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021627677022714e-05,
+      "loss": 0.8256,
+      "step": 1017
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601360108069324e-05,
+      "loss": 0.8011,
+      "step": 1018
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600556841062125e-05,
+      "loss": 0.8021,
+      "step": 1019
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997529674919755e-05,
+      "loss": 0.789,
+      "step": 1020
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989484881707874e-05,
+      "loss": 0.8075,
+      "step": 1021
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5981434039110864e-05,
+      "loss": 0.8183,
+      "step": 1022
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973377155260084e-05,
+      "loss": 0.7976,
+      "step": 1023
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596531423829299e-05,
+      "loss": 0.8002,
+      "step": 1024
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595724529635314e-05,
+      "loss": 0.818,
+      "step": 1025
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949170337590166e-05,
+      "loss": 0.8058,
+      "step": 1026
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594108937015979e-05,
+      "loss": 0.8059,
+      "step": 1027
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593300240222379e-05,
+      "loss": 0.8343,
+      "step": 1028
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924909441950017e-05,
+      "loss": 0.7758,
+      "step": 1029
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916810497512364e-05,
+      "loss": 0.81,
+      "step": 1030
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908705577090777e-05,
+      "loss": 0.8225,
+      "step": 1031
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5900594688871227e-05,
+      "loss": 0.8085,
+      "step": 1032
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892477841045723e-05,
+      "loss": 0.8231,
+      "step": 1033
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5884355041812284e-05,
+      "loss": 0.8255,
+      "step": 1034
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587622629937495e-05,
+      "loss": 0.8269,
+      "step": 1035
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868091621943754e-05,
+      "loss": 0.8157,
+      "step": 1036
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585995101773473e-05,
+      "loss": 0.8221,
+      "step": 1037
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851804494969893e-05,
+      "loss": 0.8383,
+      "step": 1038
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843652061877245e-05,
+      "loss": 0.8237,
+      "step": 1039
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583549372669074e-05,
+      "loss": 0.799,
+      "step": 1040
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582732949765031e-05,
+      "loss": 0.7963,
+      "step": 1041
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819159383001838e-05,
+      "loss": 0.8355,
+      "step": 1042
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5810983390997142e-05,
+      "loss": 0.8199,
+      "step": 1043
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580280152989398e-05,
+      "loss": 0.826,
+      "step": 1044
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579461380795604e-05,
+      "loss": 0.795,
+      "step": 1045
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5786420233452932e-05,
+      "loss": 0.7989,
+      "step": 1046
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778220814660167e-05,
+      "loss": 0.823,
+      "step": 1047
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5770015559859173e-05,
+      "loss": 0.844,
+      "step": 1048
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5761804477337262e-05,
+      "loss": 0.8232,
+      "step": 1049
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575358757538763e-05,
+      "loss": 0.8226,
+      "step": 1050
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574536486230936e-05,
+      "loss": 0.8145,
+      "step": 1051
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737136346407403e-05,
+      "loss": 0.8112,
+      "step": 1052
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572890203599256e-05,
+      "loss": 0.8047,
+      "step": 1053
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5720661939381497e-05,
+      "loss": 0.8168,
+      "step": 1054
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571241606489672e-05,
+      "loss": 0.8154,
+      "step": 1055
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570416442086656e-05,
+      "loss": 0.8217,
+      "step": 1056
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56959070156252e-05,
+      "loss": 0.8121,
+      "step": 1057
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687643857512615e-05,
+      "loss": 0.8224,
+      "step": 1058
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5679374954874607e-05,
+      "loss": 0.832,
+      "step": 1059
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567110031606277e-05,
+      "loss": 0.8319,
+      "step": 1060
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5662819949434505e-05,
+      "loss": 0.8079,
+      "step": 1061
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654533863352983e-05,
+      "loss": 0.8083,
+      "step": 1062
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5646242066187162e-05,
+      "loss": 0.8241,
+      "step": 1063
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5637944566311763e-05,
+      "loss": 0.8094,
+      "step": 1064
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5629641372107268e-05,
+      "loss": 0.8331,
+      "step": 1065
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5621332491959914e-05,
+      "loss": 0.8172,
+      "step": 1066
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5613017934261675e-05,
+      "loss": 0.7983,
+      "step": 1067
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604697707410257e-05,
+      "loss": 0.8098,
+      "step": 1068
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596371819809104e-05,
+      "loss": 0.786,
+      "step": 1069
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558804027986737e-05,
+      "loss": 0.8082,
+      "step": 1070
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5579703095999913e-05,
+      "loss": 0.7962,
+      "step": 1071
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571360276627298e-05,
+      "loss": 0.8297,
+      "step": 1072
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5563011830175782e-05,
+      "loss": 0.8275,
+      "step": 1073
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5554657765077296e-05,
+      "loss": 0.8188,
+      "step": 1074
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546298089769465e-05,
+      "loss": 0.8203,
+      "step": 1075
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5537932812695563e-05,
+      "loss": 0.8268,
+      "step": 1076
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529561942304524e-05,
+      "loss": 0.8035,
+      "step": 1077
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552118548705094e-05,
+      "loss": 0.7903,
+      "step": 1078
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5512803455395036e-05,
+      "loss": 0.8163,
+      "step": 1079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550441585580267e-05,
+      "loss": 0.8065,
+      "step": 1080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5496022696745327e-05,
+      "loss": 0.8257,
+      "step": 1081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5487623986700105e-05,
+      "loss": 0.7962,
+      "step": 1082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547921973414971e-05,
+      "loss": 0.8096,
+      "step": 1083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5470809947582445e-05,
+      "loss": 0.8082,
+      "step": 1084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5462394635492198e-05,
+      "loss": 0.808,
+      "step": 1085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5453973806378443e-05,
+      "loss": 0.8192,
+      "step": 1086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544554746874623e-05,
+      "loss": 0.8015,
+      "step": 1087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437115631106162e-05,
+      "loss": 0.8146,
+      "step": 1088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428678301974405e-05,
+      "loss": 0.8163,
+      "step": 1089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420235489872666e-05,
+      "loss": 0.7901,
+      "step": 1090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5411787203328197e-05,
+      "loss": 0.7972,
+      "step": 1091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5403333450873773e-05,
+      "loss": 0.7935,
+      "step": 1092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539487424104769e-05,
+      "loss": 0.8073,
+      "step": 1093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386409582393763e-05,
+      "loss": 0.8415,
+      "step": 1094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53779394834613e-05,
+      "loss": 0.7768,
+      "step": 1095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536946395280511e-05,
+      "loss": 0.8041,
+      "step": 1096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360982998985486e-05,
+      "loss": 0.8034,
+      "step": 1097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53524966305682e-05,
+      "loss": 0.804,
+      "step": 1098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534400485612449e-05,
+      "loss": 0.8337,
+      "step": 1099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5335507684231058e-05,
+      "loss": 0.8017,
+      "step": 1100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5327005123470056e-05,
+      "loss": 0.7967,
+      "step": 1101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5318497182429075e-05,
+      "loss": 0.8066,
+      "step": 1102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5309983869701146e-05,
+      "loss": 0.8204,
+      "step": 1103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530146519388472e-05,
+      "loss": 0.813,
+      "step": 1104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5292941163583668e-05,
+      "loss": 0.7854,
+      "step": 1105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284411787407268e-05,
+      "loss": 0.8095,
+      "step": 1106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52758770739702e-05,
+      "loss": 0.8235,
+      "step": 1107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526733703189253e-05,
+      "loss": 0.8238,
+      "step": 1108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258791669799705e-05,
+      "loss": 0.8136,
+      "step": 1109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250240996322556e-05,
+      "loss": 0.7979,
+      "step": 1110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241685020097267e-05,
+      "loss": 0.8014,
+      "step": 1111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523312374976538e-05,
+      "loss": 0.82,
+      "step": 1112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522455719397379e-05,
+      "loss": 0.8006,
+      "step": 1113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5215985361374723e-05,
+      "loss": 0.8141,
+      "step": 1114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207408260625736e-05,
+      "loss": 0.8068,
+      "step": 1115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198825900389714e-05,
+      "loss": 0.8218,
+      "step": 1116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5190238289334842e-05,
+      "loss": 0.812,
+      "step": 1117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518164543613462e-05,
+      "loss": 0.8133,
+      "step": 1118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5173047349467836e-05,
+      "loss": 0.7996,
+      "step": 1119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164444038018564e-05,
+      "loss": 0.8031,
+      "step": 1120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155835510476158e-05,
+      "loss": 0.7876,
+      "step": 1121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147221775535239e-05,
+      "loss": 0.8131,
+      "step": 1122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138602841895682e-05,
+      "loss": 0.8223,
+      "step": 1123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129978718262623e-05,
+      "loss": 0.8102,
+      "step": 1124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121349413346438e-05,
+      "loss": 0.8038,
+      "step": 1125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112714935862724e-05,
+      "loss": 0.8112,
+      "step": 1126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104075294532314e-05,
+      "loss": 0.8084,
+      "step": 1127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5095430498081257e-05,
+      "loss": 0.8224,
+      "step": 1128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086780555240803e-05,
+      "loss": 0.805,
+      "step": 1129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50781254747474e-05,
+      "loss": 0.7975,
+      "step": 1130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5069465265342691e-05,
+      "loss": 0.7726,
+      "step": 1131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060799935773493e-05,
+      "loss": 0.8353,
+      "step": 1132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052129494791796e-05,
+      "loss": 0.8281,
+      "step": 1133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043453951154754e-05,
+      "loss": 0.8039,
+      "step": 1134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5034773313624675e-05,
+      "loss": 0.7927,
+      "step": 1135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502608759096901e-05,
+      "loss": 0.8055,
+      "step": 1136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017396791960346e-05,
+      "loss": 0.7843,
+      "step": 1137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008700925376401e-05,
+      "loss": 0.7996,
+      "step": 1138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7996,
+      "step": 1139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991294024619096e-05,
+      "loss": 0.8047,
+      "step": 1140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982583008026726e-05,
+      "loss": 0.8056,
+      "step": 1141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973866959021024e-05,
+      "loss": 0.7877,
+      "step": 1142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4965145886405209e-05,
+      "loss": 0.8241,
+      "step": 1143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4956419798987573e-05,
+      "loss": 0.8284,
+      "step": 1144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494768870558147e-05,
+      "loss": 0.8159,
+      "step": 1145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938952615005316e-05,
+      "loss": 0.8314,
+      "step": 1146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930211536082566e-05,
+      "loss": 0.7931,
+      "step": 1147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4921465477641718e-05,
+      "loss": 0.8141,
+      "step": 1148
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.49127144485163e-05,
+      "loss": 0.8005,
+      "step": 1149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903958457544865e-05,
+      "loss": 0.8255,
+      "step": 1150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895197513570964e-05,
+      "loss": 0.7976,
+      "step": 1151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488643162544316e-05,
+      "loss": 0.8101,
+      "step": 1152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877660802015008e-05,
+      "loss": 0.7898,
+      "step": 1153
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.486888505214505e-05,
+      "loss": 0.8121,
+      "step": 1154
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4860104384696796e-05,
+      "loss": 0.7887,
+      "step": 1155
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851318808538732e-05,
+      "loss": 0.8121,
+      "step": 1156
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842528332544304e-05,
+      "loss": 0.8005,
+      "step": 1157
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833732965591887e-05,
+      "loss": 0.7926,
+      "step": 1158
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482493271656482e-05,
+      "loss": 0.8246,
+      "step": 1159
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816127594351358e-05,
+      "loss": 0.8167,
+      "step": 1160
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807317607844683e-05,
+      "loss": 0.7883,
+      "step": 1161
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4798502765942888e-05,
+      "loss": 0.7974,
+      "step": 1162
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478968307754898e-05,
+      "loss": 0.784,
+      "step": 1163
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780858551570838e-05,
+      "loss": 0.8171,
+      "step": 1164
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4772029196921251e-05,
+      "loss": 0.8108,
+      "step": 1165
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763195022517874e-05,
+      "loss": 0.7939,
+      "step": 1166
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4754356037283226e-05,
+      "loss": 0.7949,
+      "step": 1167
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745512250144695e-05,
+      "loss": 0.7903,
+      "step": 1168
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736663670034514e-05,
+      "loss": 0.7883,
+      "step": 1169
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4727810305889758e-05,
+      "loss": 0.7823,
+      "step": 1170
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4718952166652328e-05,
+      "loss": 0.831,
+      "step": 1171
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710089261268956e-05,
+      "loss": 0.8162,
+      "step": 1172
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.470122159869118e-05,
+      "loss": 0.7889,
+      "step": 1173
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692349187875352e-05,
+      "loss": 0.8025,
+      "step": 1174
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4683472037782615e-05,
+      "loss": 0.7972,
+      "step": 1175
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46745901573789e-05,
+      "loss": 0.8155,
+      "step": 1176
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665703555634907e-05,
+      "loss": 0.7887,
+      "step": 1177
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4656812241526117e-05,
+      "loss": 0.8054,
+      "step": 1178
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647916224032765e-05,
+      "loss": 0.8051,
+      "step": 1179
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639015512139836e-05,
+      "loss": 0.8032,
+      "step": 1180
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630110114837056e-05,
+      "loss": 0.8384,
+      "step": 1181
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462120004111889e-05,
+      "loss": 0.8111,
+      "step": 1182
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612285299984512e-05,
+      "loss": 0.8022,
+      "step": 1183
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603365900437823e-05,
+      "loss": 0.8192,
+      "step": 1184
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4594441851487425e-05,
+      "loss": 0.8163,
+      "step": 1185
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585513162146613e-05,
+      "loss": 0.807,
+      "step": 1186
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4576579841433371e-05,
+      "loss": 0.8125,
+      "step": 1187
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456764189837037e-05,
+      "loss": 0.8232,
+      "step": 1188
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558699341984928e-05,
+      "loss": 0.7991,
+      "step": 1189
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454975218130904e-05,
+      "loss": 0.8025,
+      "step": 1190
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454080042537935e-05,
+      "loss": 0.8075,
+      "step": 1191
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531844083237134e-05,
+      "loss": 0.7978,
+      "step": 1192
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522883163928307e-05,
+      "loss": 0.7733,
+      "step": 1193
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513917676503409e-05,
+      "loss": 0.822,
+      "step": 1194
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504947630017588e-05,
+      "loss": 0.8108,
+      "step": 1195
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44959730335306e-05,
+      "loss": 0.8137,
+      "step": 1196
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486993896106788e-05,
+      "loss": 0.7995,
+      "step": 1197
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478010226815103e-05,
+      "loss": 0.8083,
+      "step": 1198
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469022034729047e-05,
+      "loss": 0.8033,
+      "step": 1199
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446002932892671e-05,
+      "loss": 0.8311,
+      "step": 1200
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445103211849073e-05,
+      "loss": 0.8174,
+      "step": 1201
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4442030412508292e-05,
+      "loss": 0.8197,
+      "step": 1202
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433024220071135e-05,
+      "loss": 0.8026,
+      "step": 1203
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424013550275522e-05,
+      "loss": 0.8035,
+      "step": 1204
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4414998412222229e-05,
+      "loss": 0.8032,
+      "step": 1205
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405978815016561e-05,
+      "loss": 0.8326,
+      "step": 1206
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4396954767768323e-05,
+      "loss": 0.7743,
+      "step": 1207
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387926279591799e-05,
+      "loss": 0.8105,
+      "step": 1208
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4378893359605777e-05,
+      "loss": 0.8203,
+      "step": 1209
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369856016933516e-05,
+      "loss": 0.8013,
+      "step": 1210
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4360814260702736e-05,
+      "loss": 0.798,
+      "step": 1211
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435176810004562e-05,
+      "loss": 0.8147,
+      "step": 1212
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342717544098804e-05,
+      "loss": 0.8121,
+      "step": 1213
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4333662602003348e-05,
+      "loss": 0.7968,
+      "step": 1214
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4324603282904759e-05,
+      "loss": 0.7843,
+      "step": 1215
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4315539595952956e-05,
+      "loss": 0.8005,
+      "step": 1216
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306471550302268e-05,
+      "loss": 0.7991,
+      "step": 1217
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297399155111432e-05,
+      "loss": 0.812,
+      "step": 1218
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288322419543576e-05,
+      "loss": 0.8087,
+      "step": 1219
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4279241352766211e-05,
+      "loss": 0.7917,
+      "step": 1220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4270155963951217e-05,
+      "loss": 0.7967,
+      "step": 1221
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261066262274852e-05,
+      "loss": 0.8183,
+      "step": 1222
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425197225691772e-05,
+      "loss": 0.7957,
+      "step": 1223
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4242873957064775e-05,
+      "loss": 0.7858,
+      "step": 1224
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423377137190531e-05,
+      "loss": 0.8142,
+      "step": 1225
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224664510632943e-05,
+      "loss": 0.8267,
+      "step": 1226
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215553382445606e-05,
+      "loss": 0.8206,
+      "step": 1227
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4206437996545555e-05,
+      "loss": 0.8072,
+      "step": 1228
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197318362139334e-05,
+      "loss": 0.7913,
+      "step": 1229
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188194488437782e-05,
+      "loss": 0.8045,
+      "step": 1230
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179066384656021e-05,
+      "loss": 0.8189,
+      "step": 1231
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169934060013449e-05,
+      "loss": 0.8143,
+      "step": 1232
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4160797523733712e-05,
+      "loss": 0.8044,
+      "step": 1233
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4151656785044725e-05,
+      "loss": 0.812,
+      "step": 1234
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4142511853178643e-05,
+      "loss": 0.8141,
+      "step": 1235
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133362737371852e-05,
+      "loss": 0.7909,
+      "step": 1236
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.412420944686497e-05,
+      "loss": 0.7942,
+      "step": 1237
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411505199090283e-05,
+      "loss": 0.8139,
+      "step": 1238
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105890378734471e-05,
+      "loss": 0.7926,
+      "step": 1239
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4096724619613123e-05,
+      "loss": 0.7931,
+      "step": 1240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4087554722796216e-05,
+      "loss": 0.8177,
+      "step": 1241
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078380697545355e-05,
+      "loss": 0.8319,
+      "step": 1242
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406920255312631e-05,
+      "loss": 0.8031,
+      "step": 1243
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4060020298809023e-05,
+      "loss": 0.8162,
+      "step": 1244
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4050833943867573e-05,
+      "loss": 0.7789,
+      "step": 1245
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041643497580181e-05,
+      "loss": 0.7916,
+      "step": 1246
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4032448969229218e-05,
+      "loss": 0.78,
+      "step": 1247
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023250368101157e-05,
+      "loss": 0.7919,
+      "step": 1248
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40140477034866e-05,
+      "loss": 0.7927,
+      "step": 1249
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004840984680241e-05,
+      "loss": 0.8068,
+      "step": 1250
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399563022098088e-05,
+      "loss": 0.8169,
+      "step": 1251
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398641542169139e-05,
+      "loss": 0.8245,
+      "step": 1252
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977196596118732e-05,
+      "loss": 0.8092,
+      "step": 1253
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967973753573924e-05,
+      "loss": 0.7862,
+      "step": 1254
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395874690337205e-05,
+      "loss": 0.7891,
+      "step": 1255
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394951605483223e-05,
+      "loss": 0.8316,
+      "step": 1256
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940281217277636e-05,
+      "loss": 0.7826,
+      "step": 1257
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393104240003546e-05,
+      "loss": 0.7979,
+      "step": 1258
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921799612436917e-05,
+      "loss": 0.7923,
+      "step": 1259
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391255286381723e-05,
+      "loss": 0.8372,
+      "step": 1260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903302163515621e-05,
+      "loss": 0.782,
+      "step": 1261
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894047520875306e-05,
+      "loss": 0.7916,
+      "step": 1262
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884788945243489e-05,
+      "loss": 0.7935,
+      "step": 1263
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3875526445971333e-05,
+      "loss": 0.8132,
+      "step": 1264
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866260032413969e-05,
+      "loss": 0.8113,
+      "step": 1265
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385698971393049e-05,
+      "loss": 0.7935,
+      "step": 1266
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384771549988392e-05,
+      "loss": 0.8159,
+      "step": 1267
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838437399641226e-05,
+      "loss": 0.8077,
+      "step": 1268
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3829155422573299e-05,
+      "loss": 0.8273,
+      "step": 1269
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819869578054944e-05,
+      "loss": 0.8077,
+      "step": 1270
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810579875464869e-05,
+      "loss": 0.7933,
+      "step": 1271
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801286324185683e-05,
+      "loss": 0.8036,
+      "step": 1272
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3791988933603885e-05,
+      "loss": 0.8097,
+      "step": 1273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3782687713109844e-05,
+      "loss": 0.7662,
+      "step": 1274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3773382672097801e-05,
+      "loss": 0.7967,
+      "step": 1275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764073819965864e-05,
+      "loss": 0.8028,
+      "step": 1276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3754761166115969e-05,
+      "loss": 0.7926,
+      "step": 1277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745444719953908e-05,
+      "loss": 0.7791,
+      "step": 1278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.373612449088931e-05,
+      "loss": 0.8037,
+      "step": 1279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726800488335603e-05,
+      "loss": 0.7952,
+      "step": 1280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371747272171004e-05,
+      "loss": 0.8046,
+      "step": 1281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708141200433685e-05,
+      "loss": 0.7965,
+      "step": 1282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698805933931369e-05,
+      "loss": 0.7908,
+      "step": 1283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689466931631725e-05,
+      "loss": 0.8074,
+      "step": 1284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680124202967155e-05,
+      "loss": 0.7774,
+      "step": 1285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3670777757373819e-05,
+      "loss": 0.8013,
+      "step": 1286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366142760429164e-05,
+      "loss": 0.8072,
+      "step": 1287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.365207375316428e-05,
+      "loss": 0.7754,
+      "step": 1288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642716213439138e-05,
+      "loss": 0.7952,
+      "step": 1289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633354994567333e-05,
+      "loss": 0.8079,
+      "step": 1290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3623990106003708e-05,
+      "loss": 0.7974,
+      "step": 1291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614621557206813e-05,
+      "loss": 0.7874,
+      "step": 1292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3605249357638881e-05,
+      "loss": 0.7888,
+      "step": 1293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3595873516765847e-05,
+      "loss": 0.7941,
+      "step": 1294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586494044057326e-05,
+      "loss": 0.7894,
+      "step": 1295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3577110948986581e-05,
+      "loss": 0.8152,
+      "step": 1296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567724241030553e-05,
+      "loss": 0.7876,
+      "step": 1297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3558333929669826e-05,
+      "loss": 0.7924,
+      "step": 1298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3548940024388619e-05,
+      "loss": 0.7769,
+      "step": 1299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.353954253467479e-05,
+      "loss": 0.8101,
+      "step": 1300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530141470019812e-05,
+      "loss": 0.8084,
+      "step": 1301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520736839918764e-05,
+      "loss": 0.7672,
+      "step": 1302
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511328653870333e-05,
+      "loss": 0.7827,
+      "step": 1303
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501916921376796e-05,
+      "loss": 0.8127,
+      "step": 1304
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349250165194401e-05,
+      "loss": 0.7862,
+      "step": 1305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3483082855081406e-05,
+      "loss": 0.7728,
+      "step": 1306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473660540301983e-05,
+      "loss": 0.8097,
+      "step": 1307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.346423471712228e-05,
+      "loss": 0.8182,
+      "step": 1308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454805395062386e-05,
+      "loss": 0.7988,
+      "step": 1309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344537258364593e-05,
+      "loss": 0.7839,
+      "step": 1310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343593629240006e-05,
+      "loss": 0.7919,
+      "step": 1311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426496530855435e-05,
+      "loss": 0.7997,
+      "step": 1312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417053308546226e-05,
+      "loss": 0.8015,
+      "step": 1313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3407606635010096e-05,
+      "loss": 0.8024,
+      "step": 1314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398156519788188e-05,
+      "loss": 0.7912,
+      "step": 1315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388702972425133e-05,
+      "loss": 0.8046,
+      "step": 1316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3379246002469021e-05,
+      "loss": 0.8079,
+      "step": 1317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3369785619471397e-05,
+      "loss": 0.7855,
+      "step": 1318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3360321832987261e-05,
+      "loss": 0.7854,
+      "step": 1319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350854652575049e-05,
+      "loss": 0.7873,
+      "step": 1320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341384087796612e-05,
+      "loss": 0.7726,
+      "step": 1321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331910148217233e-05,
+      "loss": 0.8018,
+      "step": 1322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3322432843405602e-05,
+      "loss": 0.7852,
+      "step": 1323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312952182933802e-05,
+      "loss": 0.7856,
+      "step": 1324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330346817637731e-05,
+      "loss": 0.8118,
+      "step": 1325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3293980833314983e-05,
+      "loss": 0.7954,
+      "step": 1326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284490163329047e-05,
+      "loss": 0.7952,
+      "step": 1327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327499617600508e-05,
+      "loss": 0.8108,
+      "step": 1328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265498880932026e-05,
+      "loss": 0.7879,
+      "step": 1329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3255998287702155e-05,
+      "loss": 0.7968,
+      "step": 1330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246494405911076e-05,
+      "loss": 0.8297,
+      "step": 1331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236987245157727e-05,
+      "loss": 0.7847,
+      "step": 1332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227476815044337e-05,
+      "loss": 0.815,
+      "step": 1333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217963125176457e-05,
+      "loss": 0.7905,
+      "step": 1334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3208446185162913e-05,
+      "loss": 0.8011,
+      "step": 1335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198926004615836e-05,
+      "loss": 0.791,
+      "step": 1336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3189402593150607e-05,
+      "loss": 0.7968,
+      "step": 1337
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179875960385883e-05,
+      "loss": 0.7864,
+      "step": 1338
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3170346115943575e-05,
+      "loss": 0.791,
+      "step": 1339
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3160813069448828e-05,
+      "loss": 0.8038,
+      "step": 1340
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151276830530032e-05,
+      "loss": 0.7893,
+      "step": 1341
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141737408818798e-05,
+      "loss": 0.795,
+      "step": 1342
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132194813949944e-05,
+      "loss": 0.7852,
+      "step": 1343
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3122649055561505e-05,
+      "loss": 0.7921,
+      "step": 1344
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311310014329471e-05,
+      "loss": 0.7809,
+      "step": 1345
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103548086793957e-05,
+      "loss": 0.7917,
+      "step": 1346
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3093992895706842e-05,
+      "loss": 0.797,
+      "step": 1347
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084434579684114e-05,
+      "loss": 0.8041,
+      "step": 1348
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3074873148379674e-05,
+      "loss": 0.7762,
+      "step": 1349
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065308611450584e-05,
+      "loss": 0.7971,
+      "step": 1350
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3055740978557037e-05,
+      "loss": 0.7899,
+      "step": 1351
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046170259362342e-05,
+      "loss": 0.7731,
+      "step": 1352
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303659646353294e-05,
+      "loss": 0.7819,
+      "step": 1353
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3027019600738373e-05,
+      "loss": 0.811,
+      "step": 1354
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017439680651279e-05,
+      "loss": 0.7733,
+      "step": 1355
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007856712947384e-05,
+      "loss": 0.7735,
+      "step": 1356
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29982707073055e-05,
+      "loss": 0.7975,
+      "step": 1357
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988681673407501e-05,
+      "loss": 0.8062,
+      "step": 1358
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979089620938314e-05,
+      "loss": 0.7814,
+      "step": 1359
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2969494559585922e-05,
+      "loss": 0.7817,
+      "step": 1360
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2959896499041352e-05,
+      "loss": 0.7983,
+      "step": 1361
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950295448998647e-05,
+      "loss": 0.7875,
+      "step": 1362
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940691419154879e-05,
+      "loss": 0.7908,
+      "step": 1363
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931084419210128e-05,
+      "loss": 0.7861,
+      "step": 1364
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921474458867467e-05,
+      "loss": 0.7832,
+      "step": 1365
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291186154783297e-05,
+      "loss": 0.7964,
+      "step": 1366
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902245695815685e-05,
+      "loss": 0.7883,
+      "step": 1367
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289262691252763e-05,
+      "loss": 0.7903,
+      "step": 1368
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883005207683784e-05,
+      "loss": 0.8042,
+      "step": 1369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873380591002085e-05,
+      "loss": 0.8056,
+      "step": 1370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286375307220339e-05,
+      "loss": 0.7829,
+      "step": 1371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854122661011515e-05,
+      "loss": 0.7774,
+      "step": 1372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844489367153179e-05,
+      "loss": 0.8102,
+      "step": 1373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834853200358016e-05,
+      "loss": 0.7912,
+      "step": 1374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825214170358563e-05,
+      "loss": 0.8245,
+      "step": 1375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2815572286890257e-05,
+      "loss": 0.7908,
+      "step": 1376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805927559691398e-05,
+      "loss": 0.7984,
+      "step": 1377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796279998503174e-05,
+      "loss": 0.8047,
+      "step": 1378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.278662961306963e-05,
+      "loss": 0.7812,
+      "step": 1379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776976413137668e-05,
+      "loss": 0.774,
+      "step": 1380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767320408457021e-05,
+      "loss": 0.7888,
+      "step": 1381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275766160878027e-05,
+      "loss": 0.7853,
+      "step": 1382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748000023862806e-05,
+      "loss": 0.7942,
+      "step": 1383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738335663462837e-05,
+      "loss": 0.7867,
+      "step": 1384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728668537341383e-05,
+      "loss": 0.7811,
+      "step": 1385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2718998655262244e-05,
+      "loss": 0.7993,
+      "step": 1386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2709326026992013e-05,
+      "loss": 0.7726,
+      "step": 1387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699650662300052e-05,
+      "loss": 0.7433,
+      "step": 1388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689972570958488e-05,
+      "loss": 0.7899,
+      "step": 1389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2680291762742203e-05,
+      "loss": 0.7994,
+      "step": 1390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2670608247428819e-05,
+      "loss": 0.8078,
+      "step": 1391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.26609220347987e-05,
+      "loss": 0.7878,
+      "step": 1392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651233134634922e-05,
+      "loss": 0.796,
+      "step": 1393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641541556723289e-05,
+      "loss": 0.7756,
+      "step": 1394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2631847310852305e-05,
+      "loss": 0.7986,
+      "step": 1395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622150406813164e-05,
+      "loss": 0.7972,
+      "step": 1396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2612450854399747e-05,
+      "loss": 0.7857,
+      "step": 1397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2602748663408613e-05,
+      "loss": 0.8126,
+      "step": 1398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593043843638978e-05,
+      "loss": 0.7892,
+      "step": 1399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583336404892725e-05,
+      "loss": 0.7827,
+      "step": 1400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573626356974376e-05,
+      "loss": 0.7981,
+      "step": 1401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2563913709691087e-05,
+      "loss": 0.7787,
+      "step": 1402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554198472852635e-05,
+      "loss": 0.7776,
+      "step": 1403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544480656271422e-05,
+      "loss": 0.7895,
+      "step": 1404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2534760269762458e-05,
+      "loss": 0.7557,
+      "step": 1405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2525037323143333e-05,
+      "loss": 0.8081,
+      "step": 1406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515311826234235e-05,
+      "loss": 0.7975,
+      "step": 1407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505583788857924e-05,
+      "loss": 0.8281,
+      "step": 1408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249585322083973e-05,
+      "loss": 0.7762,
+      "step": 1409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486120132007527e-05,
+      "loss": 0.7849,
+      "step": 1410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476384532191754e-05,
+      "loss": 0.8067,
+      "step": 1411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2466646431225369e-05,
+      "loss": 0.8007,
+      "step": 1412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2456905838943865e-05,
+      "loss": 0.7824,
+      "step": 1413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244716276518525e-05,
+      "loss": 0.7995,
+      "step": 1414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437417219790034e-05,
+      "loss": 0.7827,
+      "step": 1415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427669212601228e-05,
+      "loss": 0.8199,
+      "step": 1416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2417918753464329e-05,
+      "loss": 0.8041,
+      "step": 1417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240816585222731e-05,
+      "loss": 0.8017,
+      "step": 1418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398410518740607e-05,
+      "loss": 0.8104,
+      "step": 1419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388652762857118e-05,
+      "loss": 0.8075,
+      "step": 1420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378892594432189e-05,
+      "loss": 0.8083,
+      "step": 1421
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236913002332359e-05,
+      "loss": 0.7894,
+      "step": 1422
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359365059391533e-05,
+      "loss": 0.8022,
+      "step": 1423
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349597712498638e-05,
+      "loss": 0.7845,
+      "step": 1424
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2339827992509933e-05,
+      "loss": 0.7973,
+      "step": 1425
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2330055909292851e-05,
+      "loss": 0.7873,
+      "step": 1426
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23202814727172e-05,
+      "loss": 0.7709,
+      "step": 1427
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310504692655165e-05,
+      "loss": 0.7886,
+      "step": 1428
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2300725578981308e-05,
+      "loss": 0.7893,
+      "step": 1429
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2290944141572545e-05,
+      "loss": 0.8031,
+      "step": 1430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228116039030813e-05,
+      "loss": 0.8016,
+      "step": 1431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2271374335069667e-05,
+      "loss": 0.7855,
+      "step": 1432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261585985741078e-05,
+      "loss": 0.7953,
+      "step": 1433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2251795352208602e-05,
+      "loss": 0.7909,
+      "step": 1434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2242002444360788e-05,
+      "loss": 0.7841,
+      "step": 1435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232207272088487e-05,
+      "loss": 0.7903,
+      "step": 1436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222409845284828e-05,
+      "loss": 0.7927,
+      "step": 1437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212610173845221e-05,
+      "loss": 0.7905,
+      "step": 1438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202808267667348e-05,
+      "loss": 0.7883,
+      "step": 1439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2193004136651133e-05,
+      "loss": 0.7915,
+      "step": 1440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2183197790698764e-05,
+      "loss": 0.8142,
+      "step": 1441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173389239714658e-05,
+      "loss": 0.7987,
+      "step": 1442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2163578493605461e-05,
+      "loss": 0.7785,
+      "step": 1443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2153765562280033e-05,
+      "loss": 0.7774,
+      "step": 1444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143950455649449e-05,
+      "loss": 0.7927,
+      "step": 1445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213413318362697e-05,
+      "loss": 0.7949,
+      "step": 1446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124313756128052e-05,
+      "loss": 0.7846,
+      "step": 1447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114492183070324e-05,
+      "loss": 0.7847,
+      "step": 1448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2104668474373584e-05,
+      "loss": 0.78,
+      "step": 1449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094842639959787e-05,
+      "loss": 0.804,
+      "step": 1450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085014689753036e-05,
+      "loss": 0.7606,
+      "step": 1451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075184633679572e-05,
+      "loss": 0.7806,
+      "step": 1452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065352481667752e-05,
+      "loss": 0.8032,
+      "step": 1453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055518243648063e-05,
+      "loss": 0.7907,
+      "step": 1454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045681929553096e-05,
+      "loss": 0.8014,
+      "step": 1455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2035843549317536e-05,
+      "loss": 0.811,
+      "step": 1456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026003112878158e-05,
+      "loss": 0.7829,
+      "step": 1457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2016160630173808e-05,
+      "loss": 0.7889,
+      "step": 1458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2006316111145401e-05,
+      "loss": 0.7892,
+      "step": 1459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1996469565735916e-05,
+      "loss": 0.7974,
+      "step": 1460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986621003890369e-05,
+      "loss": 0.7797,
+      "step": 1461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976770435555815e-05,
+      "loss": 0.78,
+      "step": 1462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1966917870681338e-05,
+      "loss": 0.7663,
+      "step": 1463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957063319218042e-05,
+      "loss": 0.8045,
+      "step": 1464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1947206791119028e-05,
+      "loss": 0.8239,
+      "step": 1465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.19373482963394e-05,
+      "loss": 0.7873,
+      "step": 1466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1927487844836244e-05,
+      "loss": 0.7902,
+      "step": 1467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1917625446568625e-05,
+      "loss": 0.7896,
+      "step": 1468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907761111497579e-05,
+      "loss": 0.7845,
+      "step": 1469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897894849586091e-05,
+      "loss": 0.7825,
+      "step": 1470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888026670799092e-05,
+      "loss": 0.7702,
+      "step": 1471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878156585103455e-05,
+      "loss": 0.7679,
+      "step": 1472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868284602467972e-05,
+      "loss": 0.781,
+      "step": 1473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858410732863357e-05,
+      "loss": 0.7786,
+      "step": 1474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1848534986262226e-05,
+      "loss": 0.7652,
+      "step": 1475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838657372639094e-05,
+      "loss": 0.7855,
+      "step": 1476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1828777901970357e-05,
+      "loss": 0.7831,
+      "step": 1477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818896584234288e-05,
+      "loss": 0.7782,
+      "step": 1478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1809013429411027e-05,
+      "loss": 0.775,
+      "step": 1479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799128447482578e-05,
+      "loss": 0.8011,
+      "step": 1480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178924164843277e-05,
+      "loss": 0.7877,
+      "step": 1481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1779353042247285e-05,
+      "loss": 0.7791,
+      "step": 1482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769462638913624e-05,
+      "loss": 0.788,
+      "step": 1483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759570448421102e-05,
+      "loss": 0.7777,
+      "step": 1484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749676480760839e-05,
+      "loss": 0.7885,
+      "step": 1485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739780745925756e-05,
+      "loss": 0.7883,
+      "step": 1486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729883253910548e-05,
+      "loss": 0.8052,
+      "step": 1487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1719984014711694e-05,
+      "loss": 0.7911,
+      "step": 1488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1710083038327436e-05,
+      "loss": 0.7977,
+      "step": 1489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1700180334757769e-05,
+      "loss": 0.7872,
+      "step": 1490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1690275914004428e-05,
+      "loss": 0.7714,
+      "step": 1491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168036978607089e-05,
+      "loss": 0.7966,
+      "step": 1492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670461960962354e-05,
+      "loss": 0.7938,
+      "step": 1493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1660552448685731e-05,
+      "loss": 0.7751,
+      "step": 1494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165064125924964e-05,
+      "loss": 0.7971,
+      "step": 1495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640728402664388e-05,
+      "loss": 0.7891,
+      "step": 1496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630813888941965e-05,
+      "loss": 0.769,
+      "step": 1497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620897728096048e-05,
+      "loss": 0.7896,
+      "step": 1498
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1610979930141965e-05,
+      "loss": 0.8026,
+      "step": 1499
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601060505096697e-05,
+      "loss": 0.8017,
+      "step": 1500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.159113946297888e-05,
+      "loss": 0.781,
+      "step": 1501
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1581216813808768e-05,
+      "loss": 0.7881,
+      "step": 1502
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1571292567608247e-05,
+      "loss": 0.7686,
+      "step": 1503
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156136673440082e-05,
+      "loss": 0.759,
+      "step": 1504
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155143932421158e-05,
+      "loss": 0.7944,
+      "step": 1505
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1541510347067224e-05,
+      "loss": 0.7905,
+      "step": 1506
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153157981299603e-05,
+      "loss": 0.8192,
+      "step": 1507
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521647732027844e-05,
+      "loss": 0.7785,
+      "step": 1508
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1511714114194072e-05,
+      "loss": 0.7771,
+      "step": 1509
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501778969527685e-05,
+      "loss": 0.779,
+      "step": 1510
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1491842308063187e-05,
+      "loss": 0.7794,
+      "step": 1511
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481904139836609e-05,
+      "loss": 0.8042,
+      "step": 1512
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471964474885513e-05,
+      "loss": 0.7708,
+      "step": 1513
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1462023323248975e-05,
+      "loss": 0.7795,
+      "step": 1514
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452080694967559e-05,
+      "loss": 0.7866,
+      "step": 1515
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1442136600083331e-05,
+      "loss": 0.773,
+      "step": 1516
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432191048639838e-05,
+      "loss": 0.8108,
+      "step": 1517
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1422244050682097e-05,
+      "loss": 0.7792,
+      "step": 1518
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412295616256577e-05,
+      "loss": 0.7599,
+      "step": 1519
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402345755411217e-05,
+      "loss": 0.8036,
+      "step": 1520
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139239447819538e-05,
+      "loss": 0.7996,
+      "step": 1521
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138244179465986e-05,
+      "loss": 0.7872,
+      "step": 1522
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.137248771485688e-05,
+      "loss": 0.7681,
+      "step": 1523
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362532248840075e-05,
+      "loss": 0.79,
+      "step": 1524
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352575406664467e-05,
+      "loss": 0.7962,
+      "step": 1525
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1342617198386484e-05,
+      "loss": 0.7818,
+      "step": 1526
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1332657634063916e-05,
+      "loss": 0.8038,
+      "step": 1527
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1322696723755935e-05,
+      "loss": 0.7875,
+      "step": 1528
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1312734477523071e-05,
+      "loss": 0.7604,
+      "step": 1529
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302770905427203e-05,
+      "loss": 0.7736,
+      "step": 1530
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1292806017531543e-05,
+      "loss": 0.8054,
+      "step": 1531
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1282839823900645e-05,
+      "loss": 0.7878,
+      "step": 1532
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1272872334600362e-05,
+      "loss": 0.7891,
+      "step": 1533
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1262903559697879e-05,
+      "loss": 0.7864,
+      "step": 1534
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252933509261658e-05,
+      "loss": 0.7837,
+      "step": 1535
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242962193361466e-05,
+      "loss": 0.7847,
+      "step": 1536
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232989622068336e-05,
+      "loss": 0.7604,
+      "step": 1537
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223015805454573e-05,
+      "loss": 0.7785,
+      "step": 1538
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121304075359375e-05,
+      "loss": 0.7675,
+      "step": 1539
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120306447656067e-05,
+      "loss": 0.7886,
+      "step": 1540
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193086984431383e-05,
+      "loss": 0.7807,
+      "step": 1541
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183108287283167e-05,
+      "loss": 0.7881,
+      "step": 1542
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173128395194512e-05,
+      "loss": 0.7919,
+      "step": 1543
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163147318245122e-05,
+      "loss": 0.7786,
+      "step": 1544
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153165066515893e-05,
+      "loss": 0.7876,
+      "step": 1545
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143181650088908e-05,
+      "loss": 0.7946,
+      "step": 1546
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133197079047422e-05,
+      "loss": 0.776,
+      "step": 1547
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123211363475863e-05,
+      "loss": 0.7766,
+      "step": 1548
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113224513459819e-05,
+      "loss": 0.7865,
+      "step": 1549
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110323653908601e-05,
+      "loss": 0.7693,
+      "step": 1550
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093247450442304e-05,
+      "loss": 0.7886,
+      "step": 1551
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083257257617688e-05,
+      "loss": 0.793,
+      "step": 1552
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073265970702264e-05,
+      "loss": 0.7895,
+      "step": 1553
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106327359978724e-05,
+      "loss": 0.7733,
+      "step": 1554
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1053280154964923e-05,
+      "loss": 0.7723,
+      "step": 1555
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043285646328696e-05,
+      "loss": 0.7902,
+      "step": 1556
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033290083973031e-05,
+      "loss": 0.7629,
+      "step": 1557
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023293477993446e-05,
+      "loss": 0.7845,
+      "step": 1558
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101329583848653e-05,
+      "loss": 0.7887,
+      "step": 1559
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1003297175549898e-05,
+      "loss": 0.7735,
+      "step": 1560
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993297499282218e-05,
+      "loss": 0.7727,
+      "step": 1561
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0983296819783167e-05,
+      "loss": 0.7786,
+      "step": 1562
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0973295147153442e-05,
+      "loss": 0.7654,
+      "step": 1563
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.096329249149474e-05,
+      "loss": 0.7976,
+      "step": 1564
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953288862909756e-05,
+      "loss": 0.7775,
+      "step": 1565
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943284271502158e-05,
+      "loss": 0.7857,
+      "step": 1566
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0933278727376593e-05,
+      "loss": 0.7884,
+      "step": 1567
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923272240638677e-05,
+      "loss": 0.783,
+      "step": 1568
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0913264821394962e-05,
+      "loss": 0.77,
+      "step": 1569
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903256479752957e-05,
+      "loss": 0.7843,
+      "step": 1570
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089324722582109e-05,
+      "loss": 0.7646,
+      "step": 1571
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088323706970872e-05,
+      "loss": 0.7911,
+      "step": 1572
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873226021526113e-05,
+      "loss": 0.7724,
+      "step": 1573
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086321409138444e-05,
+      "loss": 0.7878,
+      "step": 1574
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853201289395755e-05,
+      "loss": 0.7785,
+      "step": 1575
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843187625672997e-05,
+      "loss": 0.7786,
+      "step": 1576
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833173110329978e-05,
+      "loss": 0.7687,
+      "step": 1577
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0823157753481367e-05,
+      "loss": 0.7643,
+      "step": 1578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0813141565242682e-05,
+      "loss": 0.7783,
+      "step": 1579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803124555730282e-05,
+      "loss": 0.7822,
+      "step": 1580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793106735061358e-05,
+      "loss": 0.8176,
+      "step": 1581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783088113353917e-05,
+      "loss": 0.7825,
+      "step": 1582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0773068700726774e-05,
+      "loss": 0.7854,
+      "step": 1583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0763048507299546e-05,
+      "loss": 0.7819,
+      "step": 1584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753027543192638e-05,
+      "loss": 0.7755,
+      "step": 1585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0743005818527233e-05,
+      "loss": 0.7946,
+      "step": 1586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0732983343425282e-05,
+      "loss": 0.7939,
+      "step": 1587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722960128009492e-05,
+      "loss": 0.7775,
+      "step": 1588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0712936182403321e-05,
+      "loss": 0.7639,
+      "step": 1589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0702911516730964e-05,
+      "loss": 0.7918,
+      "step": 1590
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0692886141117343e-05,
+      "loss": 0.7922,
+      "step": 1591
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0682860065688089e-05,
+      "loss": 0.7936,
+      "step": 1592
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0672833300569559e-05,
+      "loss": 0.7613,
+      "step": 1593
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0662805855888784e-05,
+      "loss": 0.7684,
+      "step": 1594
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.06527777417735e-05,
+      "loss": 0.7547,
+      "step": 1595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064274896835211e-05,
+      "loss": 0.7533,
+      "step": 1596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632719545753678e-05,
+      "loss": 0.7889,
+      "step": 1597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622689484107936e-05,
+      "loss": 0.8081,
+      "step": 1598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0612658793545254e-05,
+      "loss": 0.7562,
+      "step": 1599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060262748419664e-05,
+      "loss": 0.7759,
+      "step": 1600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592595566193723e-05,
+      "loss": 0.7749,
+      "step": 1601
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582563049668746e-05,
+      "loss": 0.7932,
+      "step": 1602
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572529944754566e-05,
+      "loss": 0.791,
+      "step": 1603
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562496261584625e-05,
+      "loss": 0.7658,
+      "step": 1604
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552462010292955e-05,
+      "loss": 0.7625,
+      "step": 1605
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542427201014154e-05,
+      "loss": 0.785,
+      "step": 1606
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0532391843883394e-05,
+      "loss": 0.7668,
+      "step": 1607
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522355949036386e-05,
+      "loss": 0.7857,
+      "step": 1608
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512319526609405e-05,
+      "loss": 0.7717,
+      "step": 1609
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502282586739237e-05,
+      "loss": 0.7761,
+      "step": 1610
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492245139563202e-05,
+      "loss": 0.7727,
+      "step": 1611
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482207195219134e-05,
+      "loss": 0.7612,
+      "step": 1612
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0472168763845362e-05,
+      "loss": 0.7644,
+      "step": 1613
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462129855580718e-05,
+      "loss": 0.78,
+      "step": 1614
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.04520904805645e-05,
+      "loss": 0.7774,
+      "step": 1615
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442050648936487e-05,
+      "loss": 0.7703,
+      "step": 1616
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0432010370836923e-05,
+      "loss": 0.7654,
+      "step": 1617
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0421969656406495e-05,
+      "loss": 0.7394,
+      "step": 1618
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411928515786332e-05,
+      "loss": 0.7812,
+      "step": 1619
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0401886959117998e-05,
+      "loss": 0.7858,
+      "step": 1620
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0391844996543474e-05,
+      "loss": 0.7994,
+      "step": 1621
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381802638205148e-05,
+      "loss": 0.7937,
+      "step": 1622
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0371759894245812e-05,
+      "loss": 0.7831,
+      "step": 1623
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036171677480865e-05,
+      "loss": 0.7701,
+      "step": 1624
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0351673290037216e-05,
+      "loss": 0.7657,
+      "step": 1625
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341629450075448e-05,
+      "loss": 0.766,
+      "step": 1626
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331585265067623e-05,
+      "loss": 0.7663,
+      "step": 1627
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321540745158383e-05,
+      "loss": 0.798,
+      "step": 1628
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311495900492699e-05,
+      "loss": 0.7813,
+      "step": 1629
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0301450741215873e-05,
+      "loss": 0.7834,
+      "step": 1630
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291405277473531e-05,
+      "loss": 0.7923,
+      "step": 1631
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0281359519411595e-05,
+      "loss": 0.7699,
+      "step": 1632
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0271313477176289e-05,
+      "loss": 0.7777,
+      "step": 1633
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261267160914128e-05,
+      "loss": 0.7855,
+      "step": 1634
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251220580771897e-05,
+      "loss": 0.7832,
+      "step": 1635
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241173746896654e-05,
+      "loss": 0.7792,
+      "step": 1636
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0231126669435709e-05,
+      "loss": 0.7593,
+      "step": 1637
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.022107935853662e-05,
+      "loss": 0.7792,
+      "step": 1638
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021103182434718e-05,
+      "loss": 0.7871,
+      "step": 1639
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200984077015408e-05,
+      "loss": 0.7617,
+      "step": 1640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0190936126689533e-05,
+      "loss": 0.8047,
+      "step": 1641
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0180887983517999e-05,
+      "loss": 0.8111,
+      "step": 1642
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017083965764944e-05,
+      "loss": 0.7605,
+      "step": 1643
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0160791159232673e-05,
+      "loss": 0.7995,
+      "step": 1644
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0150742498416691e-05,
+      "loss": 0.7817,
+      "step": 1645
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0140693685350646e-05,
+      "loss": 0.7633,
+      "step": 1646
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130644730183856e-05,
+      "loss": 0.7655,
+      "step": 1647
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012059564306577e-05,
+      "loss": 0.7818,
+      "step": 1648
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0110546434145976e-05,
+      "loss": 0.7952,
+      "step": 1649
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100497113574182e-05,
+      "loss": 0.7778,
+      "step": 1650
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090447691500213e-05,
+      "loss": 0.7667,
+      "step": 1651
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0080398178073993e-05,
+      "loss": 0.7697,
+      "step": 1652
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070348583445544e-05,
+      "loss": 0.7583,
+      "step": 1653
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060298917764959e-05,
+      "loss": 0.7696,
+      "step": 1654
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0050249191182412e-05,
+      "loss": 0.7687,
+      "step": 1655
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0040199413848141e-05,
+      "loss": 0.7606,
+      "step": 1656
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030149595912423e-05,
+      "loss": 0.7705,
+      "step": 1657
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020099747525587e-05,
+      "loss": 0.7764,
+      "step": 1658
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0010049878837989e-05,
+      "loss": 0.7965,
+      "step": 1659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.7815,
+      "step": 1660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989950121162014e-06,
+      "loss": 0.7917,
+      "step": 1661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.979900252474415e-06,
+      "loss": 0.797,
+      "step": 1662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969850404087578e-06,
+      "loss": 0.7772,
+      "step": 1663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.959800586151859e-06,
+      "loss": 0.7671,
+      "step": 1664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949750808817588e-06,
+      "loss": 0.7842,
+      "step": 1665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93970108223504e-06,
+      "loss": 0.7775,
+      "step": 1666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.929651416554461e-06,
+      "loss": 0.7795,
+      "step": 1667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.91960182192601e-06,
+      "loss": 0.7743,
+      "step": 1668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.909552308499792e-06,
+      "loss": 0.778,
+      "step": 1669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899502886425823e-06,
+      "loss": 0.785,
+      "step": 1670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.889453565854029e-06,
+      "loss": 0.7617,
+      "step": 1671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879404356934234e-06,
+      "loss": 0.7907,
+      "step": 1672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.869355269816146e-06,
+      "loss": 0.7598,
+      "step": 1673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859306314649355e-06,
+      "loss": 0.7646,
+      "step": 1674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.849257501583312e-06,
+      "loss": 0.7768,
+      "step": 1675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.839208840767328e-06,
+      "loss": 0.7928,
+      "step": 1676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829160342350561e-06,
+      "loss": 0.7884,
+      "step": 1677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.819112016482e-06,
+      "loss": 0.7683,
+      "step": 1678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809063873310472e-06,
+      "loss": 0.741,
+      "step": 1679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799015922984598e-06,
+      "loss": 0.7643,
+      "step": 1680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.788968175652825e-06,
+      "loss": 0.76,
+      "step": 1681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778920641463384e-06,
+      "loss": 0.7656,
+      "step": 1682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.768873330564293e-06,
+      "loss": 0.7872,
+      "step": 1683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75882625310335e-06,
+      "loss": 0.7635,
+      "step": 1684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748779419228107e-06,
+      "loss": 0.7954,
+      "step": 1685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738732839085875e-06,
+      "loss": 0.7718,
+      "step": 1686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728686522823713e-06,
+      "loss": 0.7658,
+      "step": 1687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.718640480588408e-06,
+      "loss": 0.7828,
+      "step": 1688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70859472252647e-06,
+      "loss": 0.7842,
+      "step": 1689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.698549258784127e-06,
+      "loss": 0.7853,
+      "step": 1690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688504099507303e-06,
+      "loss": 0.7633,
+      "step": 1691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678459254841622e-06,
+      "loss": 0.797,
+      "step": 1692
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66841473493238e-06,
+      "loss": 0.7643,
+      "step": 1693
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658370549924559e-06,
+      "loss": 0.7768,
+      "step": 1694
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.648326709962786e-06,
+      "loss": 0.762,
+      "step": 1695
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638283225191353e-06,
+      "loss": 0.797,
+      "step": 1696
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.628240105754191e-06,
+      "loss": 0.7694,
+      "step": 1697
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618197361794855e-06,
+      "loss": 0.786,
+      "step": 1698
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.608155003456529e-06,
+      "loss": 0.7799,
+      "step": 1699
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.598113040882004e-06,
+      "loss": 0.7717,
+      "step": 1700
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588071484213668e-06,
+      "loss": 0.7652,
+      "step": 1701
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.578030343593507e-06,
+      "loss": 0.7798,
+      "step": 1702
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.567989629163078e-06,
+      "loss": 0.7835,
+      "step": 1703
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557949351063516e-06,
+      "loss": 0.7742,
+      "step": 1704
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547909519435505e-06,
+      "loss": 0.7744,
+      "step": 1705
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537870144419287e-06,
+      "loss": 0.7785,
+      "step": 1706
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52783123615464e-06,
+      "loss": 0.7743,
+      "step": 1707
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517792804780867e-06,
+      "loss": 0.7668,
+      "step": 1708
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507754860436801e-06,
+      "loss": 0.7661,
+      "step": 1709
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497717413260767e-06,
+      "loss": 0.7712,
+      "step": 1710
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487680473390598e-06,
+      "loss": 0.7457,
+      "step": 1711
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477644050963615e-06,
+      "loss": 0.7858,
+      "step": 1712
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46760815611661e-06,
+      "loss": 0.7879,
+      "step": 1713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457572798985846e-06,
+      "loss": 0.7989,
+      "step": 1714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.447537989707047e-06,
+      "loss": 0.7798,
+      "step": 1715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437503738415375e-06,
+      "loss": 0.7846,
+      "step": 1716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.427470055245437e-06,
+      "loss": 0.7785,
+      "step": 1717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417436950331257e-06,
+      "loss": 0.7958,
+      "step": 1718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407404433806284e-06,
+      "loss": 0.7824,
+      "step": 1719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397372515803365e-06,
+      "loss": 0.7657,
+      "step": 1720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.387341206454748e-06,
+      "loss": 0.7877,
+      "step": 1721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.377310515892066e-06,
+      "loss": 0.7626,
+      "step": 1722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367280454246324e-06,
+      "loss": 0.799,
+      "step": 1723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.357251031647892e-06,
+      "loss": 0.7682,
+      "step": 1724
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347222258226501e-06,
+      "loss": 0.7677,
+      "step": 1725
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337194144111216e-06,
+      "loss": 0.7604,
+      "step": 1726
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327166699430444e-06,
+      "loss": 0.7663,
+      "step": 1727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.317139934311911e-06,
+      "loss": 0.764,
+      "step": 1728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.307113858882664e-06,
+      "loss": 0.7594,
+      "step": 1729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297088483269039e-06,
+      "loss": 0.761,
+      "step": 1730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.28706381759668e-06,
+      "loss": 0.7649,
+      "step": 1731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277039871990512e-06,
+      "loss": 0.7589,
+      "step": 1732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.267016656574721e-06,
+      "loss": 0.7813,
+      "step": 1733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25699418147277e-06,
+      "loss": 0.7916,
+      "step": 1734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246972456807364e-06,
+      "loss": 0.781,
+      "step": 1735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236951492700455e-06,
+      "loss": 0.7909,
+      "step": 1736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22693129927323e-06,
+      "loss": 0.7647,
+      "step": 1737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.216911886646086e-06,
+      "loss": 0.7843,
+      "step": 1738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206893264938644e-06,
+      "loss": 0.7834,
+      "step": 1739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19687544426972e-06,
+      "loss": 0.7687,
+      "step": 1740
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.18685843475732e-06,
+      "loss": 0.7751,
+      "step": 1741
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.176842246518638e-06,
+      "loss": 0.77,
+      "step": 1742
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166826889670026e-06,
+      "loss": 0.7906,
+      "step": 1743
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156812374327007e-06,
+      "loss": 0.7766,
+      "step": 1744
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146798710604249e-06,
+      "loss": 0.7478,
+      "step": 1745
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.136785908615562e-06,
+      "loss": 0.7622,
+      "step": 1746
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126773978473889e-06,
+      "loss": 0.7699,
+      "step": 1747
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.116762930291282e-06,
+      "loss": 0.7771,
+      "step": 1748
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.106752774178911e-06,
+      "loss": 0.7515,
+      "step": 1749
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096743520247046e-06,
+      "loss": 0.789,
+      "step": 1750
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08673517860504e-06,
+      "loss": 0.7726,
+      "step": 1751
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076727759361325e-06,
+      "loss": 0.7731,
+      "step": 1752
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.066721272623407e-06,
+      "loss": 0.7775,
+      "step": 1753
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.056715728497847e-06,
+      "loss": 0.7664,
+      "step": 1754
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046711137090249e-06,
+      "loss": 0.7778,
+      "step": 1755
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.036707508505263e-06,
+      "loss": 0.7895,
+      "step": 1756
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026704852846562e-06,
+      "loss": 0.7789,
+      "step": 1757
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016703180216835e-06,
+      "loss": 0.7834,
+      "step": 1758
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006702500717786e-06,
+      "loss": 0.7764,
+      "step": 1759
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.996702824450104e-06,
+      "loss": 0.7801,
+      "step": 1760
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.986704161513473e-06,
+      "loss": 0.7464,
+      "step": 1761
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976706522006556e-06,
+      "loss": 0.7852,
+      "step": 1762
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96670991602697e-06,
+      "loss": 0.7793,
+      "step": 1763
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956714353671302e-06,
+      "loss": 0.7866,
+      "step": 1764
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.946719845035082e-06,
+      "loss": 0.767,
+      "step": 1765
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936726400212762e-06,
+      "loss": 0.7959,
+      "step": 1766
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926734029297743e-06,
+      "loss": 0.7877,
+      "step": 1767
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916742742382317e-06,
+      "loss": 0.7444,
+      "step": 1768
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9067525495577e-06,
+      "loss": 0.7859,
+      "step": 1769
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896763460913993e-06,
+      "loss": 0.7437,
+      "step": 1770
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.886775486540183e-06,
+      "loss": 0.7474,
+      "step": 1771
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876788636524138e-06,
+      "loss": 0.7712,
+      "step": 1772
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866802920952583e-06,
+      "loss": 0.7752,
+      "step": 1773
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856818349911097e-06,
+      "loss": 0.7822,
+      "step": 1774
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.846834933484109e-06,
+      "loss": 0.7692,
+      "step": 1775
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836852681754878e-06,
+      "loss": 0.7663,
+      "step": 1776
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.826871604805488e-06,
+      "loss": 0.7658,
+      "step": 1777
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816891712716834e-06,
+      "loss": 0.7498,
+      "step": 1778
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806913015568622e-06,
+      "loss": 0.7645,
+      "step": 1779
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.796935523439334e-06,
+      "loss": 0.7609,
+      "step": 1780
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786959246406252e-06,
+      "loss": 0.7954,
+      "step": 1781
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776984194545428e-06,
+      "loss": 0.7707,
+      "step": 1782
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76701037793167e-06,
+      "loss": 0.7747,
+      "step": 1783
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757037806638538e-06,
+      "loss": 0.7885,
+      "step": 1784
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747066490738345e-06,
+      "loss": 0.7645,
+      "step": 1785
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.737096440302124e-06,
+      "loss": 0.7811,
+      "step": 1786
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72712766539964e-06,
+      "loss": 0.7733,
+      "step": 1787
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717160176099358e-06,
+      "loss": 0.7797,
+      "step": 1788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707193982468456e-06,
+      "loss": 0.7774,
+      "step": 1789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6972290945728e-06,
+      "loss": 0.7617,
+      "step": 1790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68726552247693e-06,
+      "loss": 0.758,
+      "step": 1791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67730327624407e-06,
+      "loss": 0.7649,
+      "step": 1792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66734236593609e-06,
+      "loss": 0.7626,
+      "step": 1793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.657382801613523e-06,
+      "loss": 0.7593,
+      "step": 1794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.647424593335536e-06,
+      "loss": 0.7596,
+      "step": 1795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637467751159929e-06,
+      "loss": 0.7667,
+      "step": 1796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.627512285143123e-06,
+      "loss": 0.7573,
+      "step": 1797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.617558205340145e-06,
+      "loss": 0.7889,
+      "step": 1798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.607605521804626e-06,
+      "loss": 0.7457,
+      "step": 1799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597654244588786e-06,
+      "loss": 0.7865,
+      "step": 1800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587704383743423e-06,
+      "loss": 0.7725,
+      "step": 1801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.577755949317905e-06,
+      "loss": 0.7716,
+      "step": 1802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567808951360164e-06,
+      "loss": 0.7483,
+      "step": 1803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.557863399916674e-06,
+      "loss": 0.7673,
+      "step": 1804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.547919305032446e-06,
+      "loss": 0.7713,
+      "step": 1805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.53797667675103e-06,
+      "loss": 0.7638,
+      "step": 1806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52803552511449e-06,
+      "loss": 0.7729,
+      "step": 1807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518095860163395e-06,
+      "loss": 0.7484,
+      "step": 1808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.508157691936819e-06,
+      "loss": 0.7809,
+      "step": 1809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498221030472317e-06,
+      "loss": 0.7799,
+      "step": 1810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.48828588580593e-06,
+      "loss": 0.7605,
+      "step": 1811
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47835226797216e-06,
+      "loss": 0.7742,
+      "step": 1812
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.468420187003973e-06,
+      "loss": 0.7475,
+      "step": 1813
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.458489652932778e-06,
+      "loss": 0.795,
+      "step": 1814
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448560675788423e-06,
+      "loss": 0.7676,
+      "step": 1815
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438633265599182e-06,
+      "loss": 0.7745,
+      "step": 1816
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428707432391758e-06,
+      "loss": 0.7863,
+      "step": 1817
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418783186191237e-06,
+      "loss": 0.7861,
+      "step": 1818
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.408860537021127e-06,
+      "loss": 0.7711,
+      "step": 1819
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398939494903304e-06,
+      "loss": 0.7617,
+      "step": 1820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.389020069858038e-06,
+      "loss": 0.7716,
+      "step": 1821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379102271903956e-06,
+      "loss": 0.7669,
+      "step": 1822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369186111058037e-06,
+      "loss": 0.7679,
+      "step": 1823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359271597335617e-06,
+      "loss": 0.753,
+      "step": 1824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.349358740750364e-06,
+      "loss": 0.7609,
+      "step": 1825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33944755131427e-06,
+      "loss": 0.7708,
+      "step": 1826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329538039037646e-06,
+      "loss": 0.7688,
+      "step": 1827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31963021392911e-06,
+      "loss": 0.7619,
+      "step": 1828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309724085995577e-06,
+      "loss": 0.7748,
+      "step": 1829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299819665242236e-06,
+      "loss": 0.7664,
+      "step": 1830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289916961672565e-06,
+      "loss": 0.7704,
+      "step": 1831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.280015985288309e-06,
+      "loss": 0.7556,
+      "step": 1832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.270116746089455e-06,
+      "loss": 0.7849,
+      "step": 1833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260219254074247e-06,
+      "loss": 0.7874,
+      "step": 1834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250323519239165e-06,
+      "loss": 0.7768,
+      "step": 1835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240429551578901e-06,
+      "loss": 0.7596,
+      "step": 1836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.230537361086378e-06,
+      "loss": 0.7669,
+      "step": 1837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.220646957752716e-06,
+      "loss": 0.7494,
+      "step": 1838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210758351567231e-06,
+      "loss": 0.7719,
+      "step": 1839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200871552517427e-06,
+      "loss": 0.8036,
+      "step": 1840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.190986570588973e-06,
+      "loss": 0.778,
+      "step": 1841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181103415765717e-06,
+      "loss": 0.765,
+      "step": 1842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17122209802965e-06,
+      "loss": 0.7802,
+      "step": 1843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161342627360911e-06,
+      "loss": 0.7882,
+      "step": 1844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.151465013737777e-06,
+      "loss": 0.7605,
+      "step": 1845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.141589267136644e-06,
+      "loss": 0.7537,
+      "step": 1846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131715397532032e-06,
+      "loss": 0.7599,
+      "step": 1847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121843414896548e-06,
+      "loss": 0.7664,
+      "step": 1848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111973329200909e-06,
+      "loss": 0.7537,
+      "step": 1849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102105150413912e-06,
+      "loss": 0.7735,
+      "step": 1850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092238888502422e-06,
+      "loss": 0.7728,
+      "step": 1851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.082374553431373e-06,
+      "loss": 0.7434,
+      "step": 1852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.072512155163758e-06,
+      "loss": 0.7685,
+      "step": 1853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.062651703660607e-06,
+      "loss": 0.783,
+      "step": 1854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052793208880977e-06,
+      "loss": 0.7747,
+      "step": 1855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042936680781961e-06,
+      "loss": 0.7524,
+      "step": 1856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.033082129318665e-06,
+      "loss": 0.7592,
+      "step": 1857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023229564444188e-06,
+      "loss": 0.7566,
+      "step": 1858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013378996109635e-06,
+      "loss": 0.7719,
+      "step": 1859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.003530434264088e-06,
+      "loss": 0.7627,
+      "step": 1860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9936838888546e-06,
+      "loss": 0.7597,
+      "step": 1861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.983839369826194e-06,
+      "loss": 0.7713,
+      "step": 1862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973996887121845e-06,
+      "loss": 0.7835,
+      "step": 1863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.964156450682464e-06,
+      "loss": 0.7586,
+      "step": 1864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954318070446902e-06,
+      "loss": 0.7748,
+      "step": 1865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944481756351937e-06,
+      "loss": 0.7701,
+      "step": 1866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.934647518332253e-06,
+      "loss": 0.774,
+      "step": 1867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924815366320433e-06,
+      "loss": 0.7664,
+      "step": 1868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914985310246966e-06,
+      "loss": 0.7731,
+      "step": 1869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905157360040215e-06,
+      "loss": 0.7735,
+      "step": 1870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.895331525626418e-06,
+      "loss": 0.7662,
+      "step": 1871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885507816929681e-06,
+      "loss": 0.7785,
+      "step": 1872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875686243871951e-06,
+      "loss": 0.7712,
+      "step": 1873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865866816373032e-06,
+      "loss": 0.7679,
+      "step": 1874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856049544350553e-06,
+      "loss": 0.7758,
+      "step": 1875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.846234437719967e-06,
+      "loss": 0.7652,
+      "step": 1876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836421506394539e-06,
+      "loss": 0.7711,
+      "step": 1877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826610760285343e-06,
+      "loss": 0.7618,
+      "step": 1878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.816802209301241e-06,
+      "loss": 0.7634,
+      "step": 1879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.806995863348872e-06,
+      "loss": 0.7653,
+      "step": 1880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797191732332658e-06,
+      "loss": 0.7691,
+      "step": 1881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.78738982615478e-06,
+      "loss": 0.7593,
+      "step": 1882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777590154715175e-06,
+      "loss": 0.7666,
+      "step": 1883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.767792727911514e-06,
+      "loss": 0.7815,
+      "step": 1884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.757997555639213e-06,
+      "loss": 0.748,
+      "step": 1885
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.748204647791402e-06,
+      "loss": 0.7502,
+      "step": 1886
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738414014258926e-06,
+      "loss": 0.7629,
+      "step": 1887
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728625664930336e-06,
+      "loss": 0.7567,
+      "step": 1888
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71883960969187e-06,
+      "loss": 0.7716,
+      "step": 1889
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.709055858427457e-06,
+      "loss": 0.7592,
+      "step": 1890
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699274421018692e-06,
+      "loss": 0.7519,
+      "step": 1891
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68949530734484e-06,
+      "loss": 0.7747,
+      "step": 1892
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679718527282807e-06,
+      "loss": 0.7529,
+      "step": 1893
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66994409070715e-06,
+      "loss": 0.7673,
+      "step": 1894
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660172007490068e-06,
+      "loss": 0.7726,
+      "step": 1895
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650402287501366e-06,
+      "loss": 0.7543,
+      "step": 1896
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640634940608469e-06,
+      "loss": 0.7863,
+      "step": 1897
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630869976676414e-06,
+      "loss": 0.758,
+      "step": 1898
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6211074055678156e-06,
+      "loss": 0.7557,
+      "step": 1899
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611347237142884e-06,
+      "loss": 0.7655,
+      "step": 1900
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.601589481259395e-06,
+      "loss": 0.7557,
+      "step": 1901
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591834147772692e-06,
+      "loss": 0.7441,
+      "step": 1902
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.582081246535673e-06,
+      "loss": 0.7553,
+      "step": 1903
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5723307873987775e-06,
+      "loss": 0.7627,
+      "step": 1904
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562582780209972e-06,
+      "loss": 0.7502,
+      "step": 1905
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5528372348147535e-06,
+      "loss": 0.776,
+      "step": 1906
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.543094161056139e-06,
+      "loss": 0.7568,
+      "step": 1907
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533353568774634e-06,
+      "loss": 0.7519,
+      "step": 1908
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523615467808249e-06,
+      "loss": 0.7632,
+      "step": 1909
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513879867992475e-06,
+      "loss": 0.7806,
+      "step": 1910
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504146779160275e-06,
+      "loss": 0.7807,
+      "step": 1911
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494416211142077e-06,
+      "loss": 0.7631,
+      "step": 1912
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4846881737657685e-06,
+      "loss": 0.7802,
+      "step": 1913
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.47496267685667e-06,
+      "loss": 0.7742,
+      "step": 1914
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.465239730237544e-06,
+      "loss": 0.761,
+      "step": 1915
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455519343728577e-06,
+      "loss": 0.7793,
+      "step": 1916
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.44580152714737e-06,
+      "loss": 0.76,
+      "step": 1917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4360862903089196e-06,
+      "loss": 0.7718,
+      "step": 1918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.426373643025627e-06,
+      "loss": 0.7652,
+      "step": 1919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.416663595107278e-06,
+      "loss": 0.7646,
+      "step": 1920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406956156361025e-06,
+      "loss": 0.7523,
+      "step": 1921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.397251336591391e-06,
+      "loss": 0.7465,
+      "step": 1922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3875491456002565e-06,
+      "loss": 0.7678,
+      "step": 1923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377849593186839e-06,
+      "loss": 0.7787,
+      "step": 1924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.368152689147697e-06,
+      "loss": 0.7765,
+      "step": 1925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3584584432767105e-06,
+      "loss": 0.7729,
+      "step": 1926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348766865365078e-06,
+      "loss": 0.7573,
+      "step": 1927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.339077965201304e-06,
+      "loss": 0.7769,
+      "step": 1928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329391752571184e-06,
+      "loss": 0.7723,
+      "step": 1929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3197082372578035e-06,
+      "loss": 0.762,
+      "step": 1930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310027429041515e-06,
+      "loss": 0.7655,
+      "step": 1931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300349337699952e-06,
+      "loss": 0.7627,
+      "step": 1932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29067397300799e-06,
+      "loss": 0.7344,
+      "step": 1933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.281001344737757e-06,
+      "loss": 0.7757,
+      "step": 1934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271331462658621e-06,
+      "loss": 0.7735,
+      "step": 1935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.261664336537164e-06,
+      "loss": 0.7854,
+      "step": 1936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251999976137197e-06,
+      "loss": 0.7761,
+      "step": 1937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242338391219734e-06,
+      "loss": 0.7794,
+      "step": 1938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232679591542979e-06,
+      "loss": 0.7602,
+      "step": 1939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223023586862334e-06,
+      "loss": 0.7681,
+      "step": 1940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.213370386930369e-06,
+      "loss": 0.7493,
+      "step": 1941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20372000149683e-06,
+      "loss": 0.7514,
+      "step": 1942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194072440308606e-06,
+      "loss": 0.7939,
+      "step": 1943
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.184427713109747e-06,
+      "loss": 0.7753,
+      "step": 1944
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.174785829641437e-06,
+      "loss": 0.759,
+      "step": 1945
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165146799641987e-06,
+      "loss": 0.7804,
+      "step": 1946
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155510632846823e-06,
+      "loss": 0.7601,
+      "step": 1947
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145877338988487e-06,
+      "loss": 0.7644,
+      "step": 1948
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.136246927796611e-06,
+      "loss": 0.7506,
+      "step": 1949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.126619408997919e-06,
+      "loss": 0.8009,
+      "step": 1950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116994792316216e-06,
+      "loss": 0.7704,
+      "step": 1951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.107373087472371e-06,
+      "loss": 0.741,
+      "step": 1952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097754304184316e-06,
+      "loss": 0.7576,
+      "step": 1953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0881384521670325e-06,
+      "loss": 0.7702,
+      "step": 1954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078525541132537e-06,
+      "loss": 0.7648,
+      "step": 1955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068915580789878e-06,
+      "loss": 0.7663,
+      "step": 1956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059308580845126e-06,
+      "loss": 0.7649,
+      "step": 1957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0497045510013575e-06,
+      "loss": 0.7656,
+      "step": 1958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040103500958651e-06,
+      "loss": 0.7494,
+      "step": 1959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.030505440414079e-06,
+      "loss": 0.7655,
+      "step": 1960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.02091037906169e-06,
+      "loss": 0.7619,
+      "step": 1961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011318326592502e-06,
+      "loss": 0.7686,
+      "step": 1962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001729292694501e-06,
+      "loss": 0.7419,
+      "step": 1963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992143287052616e-06,
+      "loss": 0.7654,
+      "step": 1964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.982560319348723e-06,
+      "loss": 0.7265,
+      "step": 1965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97298039926163e-06,
+      "loss": 0.7662,
+      "step": 1966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963403536467065e-06,
+      "loss": 0.7756,
+      "step": 1967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953829740637662e-06,
+      "loss": 0.7673,
+      "step": 1968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944259021442967e-06,
+      "loss": 0.7481,
+      "step": 1969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934691388549418e-06,
+      "loss": 0.753,
+      "step": 1970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.925126851620327e-06,
+      "loss": 0.7522,
+      "step": 1971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915565420315891e-06,
+      "loss": 0.7768,
+      "step": 1972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906007104293161e-06,
+      "loss": 0.7897,
+      "step": 1973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896451913206044e-06,
+      "loss": 0.7711,
+      "step": 1974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886899856705292e-06,
+      "loss": 0.77,
+      "step": 1975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.877350944438495e-06,
+      "loss": 0.7604,
+      "step": 1976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867805186050056e-06,
+      "loss": 0.7759,
+      "step": 1977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.858262591181206e-06,
+      "loss": 0.7556,
+      "step": 1978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848723169469971e-06,
+      "loss": 0.764,
+      "step": 1979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.839186930551177e-06,
+      "loss": 0.7511,
+      "step": 1980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.829653884056431e-06,
+      "loss": 0.7691,
+      "step": 1981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820124039614119e-06,
+      "loss": 0.7544,
+      "step": 1982
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.810597406849398e-06,
+      "loss": 0.7469,
+      "step": 1983
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801073995384167e-06,
+      "loss": 0.7664,
+      "step": 1984
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.791553814837089e-06,
+      "loss": 0.7778,
+      "step": 1985
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7820368748235474e-06,
+      "loss": 0.7471,
+      "step": 1986
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772523184955665e-06,
+      "loss": 0.7638,
+      "step": 1987
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7630127548422775e-06,
+      "loss": 0.7662,
+      "step": 1988
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753505594088923e-06,
+      "loss": 0.7672,
+      "step": 1989
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.744001712297845e-06,
+      "loss": 0.728,
+      "step": 1990
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734501119067978e-06,
+      "loss": 0.7556,
+      "step": 1991
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725003823994925e-06,
+      "loss": 0.7657,
+      "step": 1992
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715509836670959e-06,
+      "loss": 0.7582,
+      "step": 1993
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7060191666850185e-06,
+      "loss": 0.7645,
+      "step": 1994
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.696531823622692e-06,
+      "loss": 0.7469,
+      "step": 1995
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687047817066201e-06,
+      "loss": 0.7534,
+      "step": 1996
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6775671565944e-06,
+      "loss": 0.7663,
+      "step": 1997
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66808985178277e-06,
+      "loss": 0.7587,
+      "step": 1998
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6586159122033914e-06,
+      "loss": 0.7552,
+      "step": 1999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.649145347424954e-06,
+      "loss": 0.757,
+      "step": 2000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639678167012737e-06,
+      "loss": 0.782,
+      "step": 2001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.630214380528602e-06,
+      "loss": 0.7618,
+      "step": 2002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62075399753098e-06,
+      "loss": 0.7725,
+      "step": 2003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.61129702757487e-06,
+      "loss": 0.7385,
+      "step": 2004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601843480211816e-06,
+      "loss": 0.7699,
+      "step": 2005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592393364989909e-06,
+      "loss": 0.7761,
+      "step": 2006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582946691453775e-06,
+      "loss": 0.7665,
+      "step": 2007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573503469144566e-06,
+      "loss": 0.7574,
+      "step": 2008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.564063707599942e-06,
+      "loss": 0.7765,
+      "step": 2009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.554627416354071e-06,
+      "loss": 0.7683,
+      "step": 2010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545194604937615e-06,
+      "loss": 0.7726,
+      "step": 2011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.535765282877723e-06,
+      "loss": 0.7609,
+      "step": 2012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526339459698021e-06,
+      "loss": 0.7496,
+      "step": 2013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.516917144918593e-06,
+      "loss": 0.7586,
+      "step": 2014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.507498348055989e-06,
+      "loss": 0.747,
+      "step": 2015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4980830786232065e-06,
+      "loss": 0.7448,
+      "step": 2016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4886713461296725e-06,
+      "loss": 0.802,
+      "step": 2017
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479263160081242e-06,
+      "loss": 0.7724,
+      "step": 2018
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.469858529980192e-06,
+      "loss": 0.7716,
+      "step": 2019
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460457465325213e-06,
+      "loss": 0.7457,
+      "step": 2020
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.451059975611383e-06,
+      "loss": 0.7508,
+      "step": 2021
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.441666070330177e-06,
+      "loss": 0.7718,
+      "step": 2022
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.432275758969451e-06,
+      "loss": 0.7698,
+      "step": 2023
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.422889051013421e-06,
+      "loss": 0.7734,
+      "step": 2024
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413505955942677e-06,
+      "loss": 0.7578,
+      "step": 2025
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.404126483234152e-06,
+      "loss": 0.7671,
+      "step": 2026
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3947506423611194e-06,
+      "loss": 0.7319,
+      "step": 2027
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385378442793189e-06,
+      "loss": 0.7548,
+      "step": 2028
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3760098939962935e-06,
+      "loss": 0.7778,
+      "step": 2029
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.36664500543267e-06,
+      "loss": 0.7583,
+      "step": 2030
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357283786560866e-06,
+      "loss": 0.7575,
+      "step": 2031
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.347926246835723e-06,
+      "loss": 0.7503,
+      "step": 2032
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.338572395708362e-06,
+      "loss": 0.7739,
+      "step": 2033
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.329222242626183e-06,
+      "loss": 0.7641,
+      "step": 2034
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.319875797032848e-06,
+      "loss": 0.7567,
+      "step": 2035
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.310533068368279e-06,
+      "loss": 0.7537,
+      "step": 2036
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3011940660686344e-06,
+      "loss": 0.762,
+      "step": 2037
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2918587995663195e-06,
+      "loss": 0.7679,
+      "step": 2038
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2825272782899585e-06,
+      "loss": 0.7572,
+      "step": 2039
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273199511664398e-06,
+      "loss": 0.7442,
+      "step": 2040
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263875509110694e-06,
+      "loss": 0.7598,
+      "step": 2041
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254555280046095e-06,
+      "loss": 0.7607,
+      "step": 2042
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245238833884038e-06,
+      "loss": 0.7375,
+      "step": 2043
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235926180034142e-06,
+      "loss": 0.7635,
+      "step": 2044
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2266173279022e-06,
+      "loss": 0.7579,
+      "step": 2045
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.217312286890158e-06,
+      "loss": 0.7441,
+      "step": 2046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208011066396116e-06,
+      "loss": 0.7424,
+      "step": 2047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.198713675814318e-06,
+      "loss": 0.7705,
+      "step": 2048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.189420124535133e-06,
+      "loss": 0.7628,
+      "step": 2049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.180130421945059e-06,
+      "loss": 0.7687,
+      "step": 2050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170844577426703e-06,
+      "loss": 0.7362,
+      "step": 2051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161562600358775e-06,
+      "loss": 0.7874,
+      "step": 2052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.15228450011608e-06,
+      "loss": 0.7459,
+      "step": 2053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143010286069515e-06,
+      "loss": 0.764,
+      "step": 2054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133739967586035e-06,
+      "loss": 0.7486,
+      "step": 2055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.124473554028673e-06,
+      "loss": 0.7802,
+      "step": 2056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115211054756515e-06,
+      "loss": 0.7648,
+      "step": 2057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105952479124696e-06,
+      "loss": 0.7588,
+      "step": 2058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096697836484382e-06,
+      "loss": 0.7433,
+      "step": 2059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087447136182772e-06,
+      "loss": 0.7366,
+      "step": 2060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078200387563085e-06,
+      "loss": 0.7495,
+      "step": 2061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.06895759996454e-06,
+      "loss": 0.7453,
+      "step": 2062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059718782722366e-06,
+      "loss": 0.7292,
+      "step": 2063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050483945167771e-06,
+      "loss": 0.748,
+      "step": 2064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041253096627952e-06,
+      "loss": 0.7672,
+      "step": 2065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0320262464260785e-06,
+      "loss": 0.7582,
+      "step": 2066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0228034038812725e-06,
+      "loss": 0.7982,
+      "step": 2067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.013584578308615e-06,
+      "loss": 0.7577,
+      "step": 2068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004369779019124e-06,
+      "loss": 0.7682,
+      "step": 2069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9951590153197624e-06,
+      "loss": 0.7728,
+      "step": 2070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985952296513403e-06,
+      "loss": 0.7748,
+      "step": 2071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9767496318988434e-06,
+      "loss": 0.7495,
+      "step": 2072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.967551030770786e-06,
+      "loss": 0.7668,
+      "step": 2073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9583565024198196e-06,
+      "loss": 0.7649,
+      "step": 2074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.949166056132431e-06,
+      "loss": 0.7824,
+      "step": 2075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9399797011909795e-06,
+      "loss": 0.7779,
+      "step": 2076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9307974468736885e-06,
+      "loss": 0.7538,
+      "step": 2077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921619302454645e-06,
+      "loss": 0.7496,
+      "step": 2078
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912445277203787e-06,
+      "loss": 0.7571,
+      "step": 2079
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.903275380386882e-06,
+      "loss": 0.767,
+      "step": 2080
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894109621265536e-06,
+      "loss": 0.7811,
+      "step": 2081
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8849480090971735e-06,
+      "loss": 0.7553,
+      "step": 2082
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875790553135032e-06,
+      "loss": 0.7596,
+      "step": 2083
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866637262628151e-06,
+      "loss": 0.7795,
+      "step": 2084
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8574881468213605e-06,
+      "loss": 0.7569,
+      "step": 2085
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8483432149552785e-06,
+      "loss": 0.7763,
+      "step": 2086
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.83920247626629e-06,
+      "loss": 0.77,
+      "step": 2087
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830065939986553e-06,
+      "loss": 0.7439,
+      "step": 2088
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.820933615343975e-06,
+      "loss": 0.7869,
+      "step": 2089
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811805511562218e-06,
+      "loss": 0.7556,
+      "step": 2090
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802681637860667e-06,
+      "loss": 0.7777,
+      "step": 2091
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79356200345445e-06,
+      "loss": 0.7273,
+      "step": 2092
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7844466175543975e-06,
+      "loss": 0.7556,
+      "step": 2093
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7753354893670625e-06,
+      "loss": 0.7653,
+      "step": 2094
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766228628094692e-06,
+      "loss": 0.7405,
+      "step": 2095
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.757126042935224e-06,
+      "loss": 0.7602,
+      "step": 2096
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748027743082283e-06,
+      "loss": 0.7435,
+      "step": 2097
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.73893373772515e-06,
+      "loss": 0.7718,
+      "step": 2098
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.729844036048784e-06,
+      "loss": 0.7622,
+      "step": 2099
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.720758647233793e-06,
+      "loss": 0.7318,
+      "step": 2100
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711677580456425e-06,
+      "loss": 0.7718,
+      "step": 2101
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702600844888566e-06,
+      "loss": 0.7682,
+      "step": 2102
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.693528449697734e-06,
+      "loss": 0.7364,
+      "step": 2103
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6844604040470505e-06,
+      "loss": 0.7547,
+      "step": 2104
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675396717095245e-06,
+      "loss": 0.7666,
+      "step": 2105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.666337397996655e-06,
+      "loss": 0.7527,
+      "step": 2106
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.657282455901201e-06,
+      "loss": 0.7556,
+      "step": 2107
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648231899954381e-06,
+      "loss": 0.7704,
+      "step": 2108
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639185739297269e-06,
+      "loss": 0.7538,
+      "step": 2109
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630143983066489e-06,
+      "loss": 0.7766,
+      "step": 2110
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.621106640394226e-06,
+      "loss": 0.7444,
+      "step": 2111
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612073720408204e-06,
+      "loss": 0.7601,
+      "step": 2112
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.603045232231681e-06,
+      "loss": 0.7417,
+      "step": 2113
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594021184983437e-06,
+      "loss": 0.7409,
+      "step": 2114
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5850015877777734e-06,
+      "loss": 0.7494,
+      "step": 2115
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.575986449724482e-06,
+      "loss": 0.7661,
+      "step": 2116
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56697577992887e-06,
+      "loss": 0.7485,
+      "step": 2117
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557969587491712e-06,
+      "loss": 0.7438,
+      "step": 2118
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548967881509275e-06,
+      "loss": 0.747,
+      "step": 2119
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539970671073293e-06,
+      "loss": 0.7477,
+      "step": 2120
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530977965270954e-06,
+      "loss": 0.7673,
+      "step": 2121
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521989773184901e-06,
+      "loss": 0.7294,
+      "step": 2122
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5130061038932125e-06,
+      "loss": 0.7429,
+      "step": 2123
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504026966469404e-06,
+      "loss": 0.7556,
+      "step": 2124
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495052369982415e-06,
+      "loss": 0.7603,
+      "step": 2125
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486082323496591e-06,
+      "loss": 0.7399,
+      "step": 2126
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477116836071692e-06,
+      "loss": 0.7695,
+      "step": 2127
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468155916762869e-06,
+      "loss": 0.7525,
+      "step": 2128
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459199574620657e-06,
+      "loss": 0.7651,
+      "step": 2129
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450247818690965e-06,
+      "loss": 0.7503,
+      "step": 2130
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4413006580150776e-06,
+      "loss": 0.7762,
+      "step": 2131
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432358101629635e-06,
+      "loss": 0.7623,
+      "step": 2132
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.423420158566629e-06,
+      "loss": 0.7756,
+      "step": 2133
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.414486837853389e-06,
+      "loss": 0.7553,
+      "step": 2134
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.405558148512581e-06,
+      "loss": 0.7416,
+      "step": 2135
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396634099562181e-06,
+      "loss": 0.7484,
+      "step": 2136
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.387714700015492e-06,
+      "loss": 0.7443,
+      "step": 2137
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3787999588811134e-06,
+      "loss": 0.7725,
+      "step": 2138
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3698898851629425e-06,
+      "loss": 0.7233,
+      "step": 2139
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360984487860167e-06,
+      "loss": 0.758,
+      "step": 2140
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.352083775967237e-06,
+      "loss": 0.7638,
+      "step": 2141
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.343187758473889e-06,
+      "loss": 0.7744,
+      "step": 2142
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.334296444365097e-06,
+      "loss": 0.7417,
+      "step": 2143
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325409842621105e-06,
+      "loss": 0.7483,
+      "step": 2144
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.316527962217386e-06,
+      "loss": 0.7688,
+      "step": 2145
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3076508121246476e-06,
+      "loss": 0.7703,
+      "step": 2146
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298778401308824e-06,
+      "loss": 0.7514,
+      "step": 2147
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.28991073873105e-06,
+      "loss": 0.7508,
+      "step": 2148
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.281047833347676e-06,
+      "loss": 0.75,
+      "step": 2149
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.272189694110245e-06,
+      "loss": 0.7679,
+      "step": 2150
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263336329965485e-06,
+      "loss": 0.7666,
+      "step": 2151
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.254487749855304e-06,
+      "loss": 0.7566,
+      "step": 2152
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.245643962716776e-06,
+      "loss": 0.7624,
+      "step": 2153
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.236804977482133e-06,
+      "loss": 0.7627,
+      "step": 2154
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.227970803078754e-06,
+      "loss": 0.7555,
+      "step": 2155
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2191414484291655e-06,
+      "loss": 0.763,
+      "step": 2156
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2103169224510265e-06,
+      "loss": 0.7329,
+      "step": 2157
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201497234057111e-06,
+      "loss": 0.7515,
+      "step": 2158
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192682392155318e-06,
+      "loss": 0.7576,
+      "step": 2159
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183872405648647e-06,
+      "loss": 0.7703,
+      "step": 2160
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.175067283435184e-06,
+      "loss": 0.7446,
+      "step": 2161
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1662670344081145e-06,
+      "loss": 0.7595,
+      "step": 2162
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157471667455699e-06,
+      "loss": 0.7547,
+      "step": 2163
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.148681191461266e-06,
+      "loss": 0.7368,
+      "step": 2164
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.139895615303207e-06,
+      "loss": 0.7585,
+      "step": 2165
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1311149478549536e-06,
+      "loss": 0.77,
+      "step": 2166
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1223391979849975e-06,
+      "loss": 0.7335,
+      "step": 2167
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113568374556846e-06,
+      "loss": 0.7571,
+      "step": 2168
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1048024864290405e-06,
+      "loss": 0.7371,
+      "step": 2169
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096041542455137e-06,
+      "loss": 0.7238,
+      "step": 2170
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087285551483698e-06,
+      "loss": 0.7547,
+      "step": 2171
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.078534522358285e-06,
+      "loss": 0.7665,
+      "step": 2172
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069788463917438e-06,
+      "loss": 0.7472,
+      "step": 2173
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061047384994688e-06,
+      "loss": 0.7666,
+      "step": 2174
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.05231129441853e-06,
+      "loss": 0.7537,
+      "step": 2175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043580201012428e-06,
+      "loss": 0.7516,
+      "step": 2176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034854113594791e-06,
+      "loss": 0.7765,
+      "step": 2177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02613304097898e-06,
+      "loss": 0.7557,
+      "step": 2178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017416991973281e-06,
+      "loss": 0.7358,
+      "step": 2179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00870597538091e-06,
+      "loss": 0.7618,
+      "step": 2180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.7781,
+      "step": 2181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991299074623603e-06,
+      "loss": 0.7561,
+      "step": 2182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982603208039656e-06,
+      "loss": 0.7635,
+      "step": 2183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.973912409030991e-06,
+      "loss": 0.7623,
+      "step": 2184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965226686375329e-06,
+      "loss": 0.7576,
+      "step": 2185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956546048845249e-06,
+      "loss": 0.7453,
+      "step": 2186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947870505208206e-06,
+      "loss": 0.7261,
+      "step": 2187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93920006422651e-06,
+      "loss": 0.7622,
+      "step": 2188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93053473465731e-06,
+      "loss": 0.7718,
+      "step": 2189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921874525252602e-06,
+      "loss": 0.7601,
+      "step": 2190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9132194447591995e-06,
+      "loss": 0.747,
+      "step": 2191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904569501918748e-06,
+      "loss": 0.7342,
+      "step": 2192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895924705467689e-06,
+      "loss": 0.7591,
+      "step": 2193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88728506413728e-06,
+      "loss": 0.7464,
+      "step": 2194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878650586653566e-06,
+      "loss": 0.7382,
+      "step": 2195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870021281737377e-06,
+      "loss": 0.7218,
+      "step": 2196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861397158104322e-06,
+      "loss": 0.7436,
+      "step": 2197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852778224464766e-06,
+      "loss": 0.7439,
+      "step": 2198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8441644895238445e-06,
+      "loss": 0.733,
+      "step": 2199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835555961981437e-06,
+      "loss": 0.7663,
+      "step": 2200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.826952650532165e-06,
+      "loss": 0.7533,
+      "step": 2201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81835456386538e-06,
+      "loss": 0.7726,
+      "step": 2202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.809761710665161e-06,
+      "loss": 0.7553,
+      "step": 2203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801174099610293e-06,
+      "loss": 0.7437,
+      "step": 2204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7925917393742686e-06,
+      "loss": 0.7825,
+      "step": 2205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784014638625282e-06,
+      "loss": 0.742,
+      "step": 2206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775442806026214e-06,
+      "loss": 0.7297,
+      "step": 2207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7668762502346214e-06,
+      "loss": 0.771,
+      "step": 2208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758314979902734e-06,
+      "loss": 0.7725,
+      "step": 2209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749759003677448e-06,
+      "loss": 0.7669,
+      "step": 2210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.741208330200296e-06,
+      "loss": 0.7508,
+      "step": 2211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732662968107474e-06,
+      "loss": 0.7268,
+      "step": 2212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724122926029803e-06,
+      "loss": 0.7426,
+      "step": 2213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7155882125927336e-06,
+      "loss": 0.7466,
+      "step": 2214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.707058836416333e-06,
+      "loss": 0.7647,
+      "step": 2215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698534806115284e-06,
+      "loss": 0.7584,
+      "step": 2216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.69001613029886e-06,
+      "loss": 0.7711,
+      "step": 2217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6815028175709285e-06,
+      "loss": 0.7497,
+      "step": 2218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6729948765299475e-06,
+      "loss": 0.7826,
+      "step": 2219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.664492315768943e-06,
+      "loss": 0.753,
+      "step": 2220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65599514387551e-06,
+      "loss": 0.7418,
+      "step": 2221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.647503369431805e-06,
+      "loss": 0.7369,
+      "step": 2222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639017001014517e-06,
+      "loss": 0.7652,
+      "step": 2223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630536047194894e-06,
+      "loss": 0.7571,
+      "step": 2224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6220605165387025e-06,
+      "loss": 0.7504,
+      "step": 2225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613590417606239e-06,
+      "loss": 0.7654,
+      "step": 2226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.605125758952309e-06,
+      "loss": 0.7446,
+      "step": 2227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596666549126231e-06,
+      "loss": 0.753,
+      "step": 2228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.58821279667181e-06,
+      "loss": 0.7413,
+      "step": 2229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.57976451012734e-06,
+      "loss": 0.7609,
+      "step": 2230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571321698025601e-06,
+      "loss": 0.7434,
+      "step": 2231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5628843688938425e-06,
+      "loss": 0.7472,
+      "step": 2232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554452531253774e-06,
+      "loss": 0.7613,
+      "step": 2233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.546026193621558e-06,
+      "loss": 0.7572,
+      "step": 2234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.537605364507806e-06,
+      "loss": 0.7297,
+      "step": 2235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529190052417559e-06,
+      "loss": 0.7453,
+      "step": 2236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5207802658502905e-06,
+      "loss": 0.7586,
+      "step": 2237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.512376013299895e-06,
+      "loss": 0.7504,
+      "step": 2238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503977303254673e-06,
+      "loss": 0.7431,
+      "step": 2239
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.495584144197331e-06,
+      "loss": 0.769,
+      "step": 2240
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487196544604968e-06,
+      "loss": 0.7188,
+      "step": 2241
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.478814512949066e-06,
+      "loss": 0.7437,
+      "step": 2242
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.47043805769548e-06,
+      "loss": 0.7499,
+      "step": 2243
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462067187304442e-06,
+      "loss": 0.7459,
+      "step": 2244
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453701910230537e-06,
+      "loss": 0.7704,
+      "step": 2245
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445342234922704e-06,
+      "loss": 0.7408,
+      "step": 2246
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.436988169824225e-06,
+      "loss": 0.7516,
+      "step": 2247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428639723372706e-06,
+      "loss": 0.7674,
+      "step": 2248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.420296904000091e-06,
+      "loss": 0.7431,
+      "step": 2249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4119597201326325e-06,
+      "loss": 0.7517,
+      "step": 2250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403628180190896e-06,
+      "loss": 0.7529,
+      "step": 2251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395302292589741e-06,
+      "loss": 0.7731,
+      "step": 2252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38698206573833e-06,
+      "loss": 0.7225,
+      "step": 2253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.378667508040092e-06,
+      "loss": 0.758,
+      "step": 2254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370358627892736e-06,
+      "loss": 0.761,
+      "step": 2255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.36205543368824e-06,
+      "loss": 0.7413,
+      "step": 2256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.353757933812841e-06,
+      "loss": 0.7714,
+      "step": 2257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.345466136647019e-06,
+      "loss": 0.7664,
+      "step": 2258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337180050565497e-06,
+      "loss": 0.763,
+      "step": 2259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3288996839372335e-06,
+      "loss": 0.7518,
+      "step": 2260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.320625045125398e-06,
+      "loss": 0.7257,
+      "step": 2261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312356142487389e-06,
+      "loss": 0.7668,
+      "step": 2262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304092984374803e-06,
+      "loss": 0.7647,
+      "step": 2263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295835579133437e-06,
+      "loss": 0.7728,
+      "step": 2264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287583935103281e-06,
+      "loss": 0.7719,
+      "step": 2265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2793380606185046e-06,
+      "loss": 0.7391,
+      "step": 2266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271097964007443e-06,
+      "loss": 0.7612,
+      "step": 2267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.262863653592601e-06,
+      "loss": 0.7411,
+      "step": 2268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.25463513769064e-06,
+      "loss": 0.77,
+      "step": 2269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246412424612372e-06,
+      "loss": 0.7411,
+      "step": 2270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2381955226627414e-06,
+      "loss": 0.7543,
+      "step": 2271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229984440140827e-06,
+      "loss": 0.7471,
+      "step": 2272
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.221779185339836e-06,
+      "loss": 0.7792,
+      "step": 2273
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213579766547073e-06,
+      "loss": 0.7433,
+      "step": 2274
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205386192043962e-06,
+      "loss": 0.7687,
+      "step": 2275
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.197198470106022e-06,
+      "loss": 0.7653,
+      "step": 2276
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189016609002859e-06,
+      "loss": 0.7497,
+      "step": 2277
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.180840616998164e-06,
+      "loss": 0.7417,
+      "step": 2278
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172670502349693e-06,
+      "loss": 0.756,
+      "step": 2279
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.164506273309264e-06,
+      "loss": 0.7382,
+      "step": 2280
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.15634793812276e-06,
+      "loss": 0.729,
+      "step": 2281
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148195505030108e-06,
+      "loss": 0.7827,
+      "step": 2282
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140048982265272e-06,
+      "loss": 0.7625,
+      "step": 2283
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131908378056247e-06,
+      "loss": 0.754,
+      "step": 2284
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1237737006250545e-06,
+      "loss": 0.7646,
+      "step": 2285
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115644958187719e-06,
+      "loss": 0.7476,
+      "step": 2286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107522158954281e-06,
+      "loss": 0.7354,
+      "step": 2287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.099405311128775e-06,
+      "loss": 0.7486,
+      "step": 2288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091294422909226e-06,
+      "loss": 0.7319,
+      "step": 2289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0831895024876356e-06,
+      "loss": 0.7538,
+      "step": 2290
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075090558049986e-06,
+      "loss": 0.7701,
+      "step": 2291
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066997597776216e-06,
+      "loss": 0.7551,
+      "step": 2292
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058910629840216e-06,
+      "loss": 0.7328,
+      "step": 2293
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050829662409837e-06,
+      "loss": 0.7715,
+      "step": 2294
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042754703646864e-06,
+      "loss": 0.7263,
+      "step": 2295
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.03468576170701e-06,
+      "loss": 0.7668,
+      "step": 2296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0266228447399154e-06,
+      "loss": 0.7574,
+      "step": 2297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018565960889137e-06,
+      "loss": 0.765,
+      "step": 2298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010515118292127e-06,
+      "loss": 0.7587,
+      "step": 2299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.002470325080249e-06,
+      "loss": 0.7529,
+      "step": 2300
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.99443158937875e-06,
+      "loss": 0.7521,
+      "step": 2301
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9863989193067586e-06,
+      "loss": 0.777,
+      "step": 2302
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.978372322977286e-06,
+      "loss": 0.7424,
+      "step": 2303
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970351808497191e-06,
+      "loss": 0.7355,
+      "step": 2304
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962337383967211e-06,
+      "loss": 0.7443,
+      "step": 2305
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954329057481913e-06,
+      "loss": 0.7432,
+      "step": 2306
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.946326837129717e-06,
+      "loss": 0.759,
+      "step": 2307
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.938330730992875e-06,
+      "loss": 0.769,
+      "step": 2308
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930340747147458e-06,
+      "loss": 0.7713,
+      "step": 2309
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.922356893663364e-06,
+      "loss": 0.7208,
+      "step": 2310
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914379178604282e-06,
+      "loss": 0.7476,
+      "step": 2311
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.906407610027717e-06,
+      "loss": 0.7567,
+      "step": 2312
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898442195984961e-06,
+      "loss": 0.7355,
+      "step": 2313
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8904829445210864e-06,
+      "loss": 0.7364,
+      "step": 2314
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.882529863674947e-06,
+      "loss": 0.7672,
+      "step": 2315
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874582961479165e-06,
+      "loss": 0.7383,
+      "step": 2316
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866642245960114e-06,
+      "loss": 0.7459,
+      "step": 2317
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8587077251379215e-06,
+      "loss": 0.7377,
+      "step": 2318
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850779407026464e-06,
+      "loss": 0.7637,
+      "step": 2319
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.842857299633348e-06,
+      "loss": 0.7481,
+      "step": 2320
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8349414109599115e-06,
+      "loss": 0.7604,
+      "step": 2321
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8270317490012085e-06,
+      "loss": 0.7458,
+      "step": 2322
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.819128321746008e-06,
+      "loss": 0.7542,
+      "step": 2323
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8112311371767718e-06,
+      "loss": 0.7346,
+      "step": 2324
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8033402032696677e-06,
+      "loss": 0.7435,
+      "step": 2325
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.795455527994544e-06,
+      "loss": 0.7419,
+      "step": 2326
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.787577119314931e-06,
+      "loss": 0.7418,
+      "step": 2327
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779704985188033e-06,
+      "loss": 0.742,
+      "step": 2328
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771839133564704e-06,
+      "loss": 0.7561,
+      "step": 2329
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7639795723894703e-06,
+      "loss": 0.732,
+      "step": 2330
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.756126309600486e-06,
+      "loss": 0.7611,
+      "step": 2331
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7482793531295593e-06,
+      "loss": 0.7522,
+      "step": 2332
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740438710902122e-06,
+      "loss": 0.7404,
+      "step": 2333
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.732604390837231e-06,
+      "loss": 0.7631,
+      "step": 2334
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7247764008475584e-06,
+      "loss": 0.7544,
+      "step": 2335
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7169547488393765e-06,
+      "loss": 0.7507,
+      "step": 2336
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7091394427125626e-06,
+      "loss": 0.7583,
+      "step": 2337
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.701330490360583e-06,
+      "loss": 0.7276,
+      "step": 2338
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6935278996704884e-06,
+      "loss": 0.7553,
+      "step": 2339
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6857316785228993e-06,
+      "loss": 0.7437,
+      "step": 2340
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6779418347920125e-06,
+      "loss": 0.7892,
+      "step": 2341
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6701583763455727e-06,
+      "loss": 0.7641,
+      "step": 2342
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6623813110448757e-06,
+      "loss": 0.7269,
+      "step": 2343
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6546106467447685e-06,
+      "loss": 0.7417,
+      "step": 2344
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.64684639129363e-06,
+      "loss": 0.7429,
+      "step": 2345
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639088552533364e-06,
+      "loss": 0.7541,
+      "step": 2346
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6313371382993945e-06,
+      "loss": 0.7417,
+      "step": 2347
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623592156420661e-06,
+      "loss": 0.7457,
+      "step": 2348
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6158536147195956e-06,
+      "loss": 0.7301,
+      "step": 2349
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6081215210121344e-06,
+      "loss": 0.7651,
+      "step": 2350
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6003958831077014e-06,
+      "loss": 0.7465,
+      "step": 2351
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5926767088091953e-06,
+      "loss": 0.7504,
+      "step": 2352
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58496400591299e-06,
+      "loss": 0.7418,
+      "step": 2353
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.577257782208926e-06,
+      "loss": 0.7586,
+      "step": 2354
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5695580454802938e-06,
+      "loss": 0.7624,
+      "step": 2355
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5618648035038297e-06,
+      "loss": 0.7211,
+      "step": 2356
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.554178064049718e-06,
+      "loss": 0.7229,
+      "step": 2357
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5464978348815727e-06,
+      "loss": 0.7634,
+      "step": 2358
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5388241237564337e-06,
+      "loss": 0.7612,
+      "step": 2359
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.531156938424759e-06,
+      "loss": 0.7538,
+      "step": 2360
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5234962866304044e-06,
+      "loss": 0.7284,
+      "step": 2361
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.51584217611064e-06,
+      "loss": 0.7641,
+      "step": 2362
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5081946145961243e-06,
+      "loss": 0.7265,
+      "step": 2363
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5005536098109004e-06,
+      "loss": 0.7453,
+      "step": 2364
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4929191694723907e-06,
+      "loss": 0.7523,
+      "step": 2365
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4852913012913903e-06,
+      "loss": 0.7585,
+      "step": 2366
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4776700129720485e-06,
+      "loss": 0.7498,
+      "step": 2367
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.470055312211872e-06,
+      "loss": 0.7412,
+      "step": 2368
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.462447206701717e-06,
+      "loss": 0.7504,
+      "step": 2369
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4548457041257765e-06,
+      "loss": 0.7439,
+      "step": 2370
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4472508121615778e-06,
+      "loss": 0.7492,
+      "step": 2371
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4396625384799654e-06,
+      "loss": 0.7505,
+      "step": 2372
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4320808907451087e-06,
+      "loss": 0.7554,
+      "step": 2373
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.424505876614469e-06,
+      "loss": 0.7571,
+      "step": 2374
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416937503738824e-06,
+      "loss": 0.7497,
+      "step": 2375
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.409375779762235e-06,
+      "loss": 0.7605,
+      "step": 2376
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4018207123220503e-06,
+      "loss": 0.7433,
+      "step": 2377
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394272309048895e-06,
+      "loss": 0.7618,
+      "step": 2378
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386730577566667e-06,
+      "loss": 0.7415,
+      "step": 2379
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.379195525492519e-06,
+      "loss": 0.7469,
+      "step": 2380
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3716671604368557e-06,
+      "loss": 0.7461,
+      "step": 2381
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3641454900033354e-06,
+      "loss": 0.7324,
+      "step": 2382
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356630521788854e-06,
+      "loss": 0.7455,
+      "step": 2383
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3491222633835363e-06,
+      "loss": 0.7705,
+      "step": 2384
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3416207223707276e-06,
+      "loss": 0.7389,
+      "step": 2385
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334125906326997e-06,
+      "loss": 0.7431,
+      "step": 2386
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3266378228221063e-06,
+      "loss": 0.7258,
+      "step": 2387
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3191564794190324e-06,
+      "loss": 0.7385,
+      "step": 2388
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.311681883673937e-06,
+      "loss": 0.7586,
+      "step": 2389
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30421404313617e-06,
+      "loss": 0.7235,
+      "step": 2390
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.29675296534826e-06,
+      "loss": 0.756,
+      "step": 2391
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2892986578458985e-06,
+      "loss": 0.7543,
+      "step": 2392
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281851128157939e-06,
+      "loss": 0.7464,
+      "step": 2393
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.274410383806398e-06,
+      "loss": 0.7699,
+      "step": 2394
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2669764323064322e-06,
+      "loss": 0.7528,
+      "step": 2395
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.259549281166341e-06,
+      "loss": 0.721,
+      "step": 2396
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2521289378875533e-06,
+      "loss": 0.7497,
+      "step": 2397
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244715409964625e-06,
+      "loss": 0.7605,
+      "step": 2398
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2373087048852203e-06,
+      "loss": 0.7397,
+      "step": 2399
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2299088301301206e-06,
+      "loss": 0.7355,
+      "step": 2400
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2225157931732077e-06,
+      "loss": 0.7711,
+      "step": 2401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2151296014814527e-06,
+      "loss": 0.7584,
+      "step": 2402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2077502625149183e-06,
+      "loss": 0.7369,
+      "step": 2403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2003777837267457e-06,
+      "loss": 0.7481,
+      "step": 2404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1930121725631425e-06,
+      "loss": 0.7366,
+      "step": 2405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1856534364633786e-06,
+      "loss": 0.7632,
+      "step": 2406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1783015828597873e-06,
+      "loss": 0.7327,
+      "step": 2407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1709566191777487e-06,
+      "loss": 0.7438,
+      "step": 2408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.163618552835681e-06,
+      "loss": 0.7123,
+      "step": 2409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1562873912450397e-06,
+      "loss": 0.757,
+      "step": 2410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1489631418103063e-06,
+      "loss": 0.7618,
+      "step": 2411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141645811928974e-06,
+      "loss": 0.7498,
+      "step": 2412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1343354089915545e-06,
+      "loss": 0.7651,
+      "step": 2413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1270319403815618e-06,
+      "loss": 0.7351,
+      "step": 2414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1197354134755065e-06,
+      "loss": 0.7385,
+      "step": 2415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.11244583564289e-06,
+      "loss": 0.733,
+      "step": 2416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1051632142461864e-06,
+      "loss": 0.7299,
+      "step": 2417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097887556640855e-06,
+      "loss": 0.7487,
+      "step": 2418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0906188701753127e-06,
+      "loss": 0.7527,
+      "step": 2419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0833571621909397e-06,
+      "loss": 0.7599,
+      "step": 2420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0761024400220695e-06,
+      "loss": 0.7762,
+      "step": 2421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0688547109959775e-06,
+      "loss": 0.7623,
+      "step": 2422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0616139824328806e-06,
+      "loss": 0.7309,
+      "step": 2423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.054380261645913e-06,
+      "loss": 0.7547,
+      "step": 2424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0471535559411434e-06,
+      "loss": 0.7573,
+      "step": 2425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039933872617551e-06,
+      "loss": 0.7309,
+      "step": 2426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032721218967023e-06,
+      "loss": 0.734,
+      "step": 2427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.025515602274346e-06,
+      "loss": 0.7395,
+      "step": 2428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0183170298172015e-06,
+      "loss": 0.7485,
+      "step": 2429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0111255088661505e-06,
+      "loss": 0.7504,
+      "step": 2430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0039410466846353e-06,
+      "loss": 0.7535,
+      "step": 2431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99676365052897e-06,
+      "loss": 0.7605,
+      "step": 2432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9895933276483323e-06,
+      "loss": 0.7557,
+      "step": 2433
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982430085284753e-06,
+      "loss": 0.7443,
+      "step": 2434
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.975273930673116e-06,
+      "loss": 0.7423,
+      "step": 2435
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.968124871041146e-06,
+      "loss": 0.751,
+      "step": 2436
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9609829136093914e-06,
+      "loss": 0.7399,
+      "step": 2437
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9538480655912415e-06,
+      "loss": 0.7753,
+      "step": 2438
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9467203341928986e-06,
+      "loss": 0.7625,
+      "step": 2439
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.939599726613377e-06,
+      "loss": 0.747,
+      "step": 2440
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9324862500445008e-06,
+      "loss": 0.7456,
+      "step": 2441
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9253799116708802e-06,
+      "loss": 0.7433,
+      "step": 2442
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9182807186699304e-06,
+      "loss": 0.7432,
+      "step": 2443
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.911188678211837e-06,
+      "loss": 0.7517,
+      "step": 2444
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.904103797459571e-06,
+      "loss": 0.7528,
+      "step": 2445
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.897026083568866e-06,
+      "loss": 0.7185,
+      "step": 2446
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.889955543688221e-06,
+      "loss": 0.7463,
+      "step": 2447
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.88289218495889e-06,
+      "loss": 0.7419,
+      "step": 2448
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.875836014514867e-06,
+      "loss": 0.7398,
+      "step": 2449
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.868787039482891e-06,
+      "loss": 0.7644,
+      "step": 2450
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.861745266982435e-06,
+      "loss": 0.7469,
+      "step": 2451
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.854710704125696e-06,
+      "loss": 0.7517,
+      "step": 2452
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.847683358017589e-06,
+      "loss": 0.732,
+      "step": 2453
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8406632357557405e-06,
+      "loss": 0.7323,
+      "step": 2454
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.83365034443048e-06,
+      "loss": 0.7186,
+      "step": 2455
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826644691124828e-06,
+      "loss": 0.7471,
+      "step": 2456
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8196462829145068e-06,
+      "loss": 0.7397,
+      "step": 2457
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812655126867914e-06,
+      "loss": 0.7619,
+      "step": 2458
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.805671230046122e-06,
+      "loss": 0.735,
+      "step": 2459
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7986945995028745e-06,
+      "loss": 0.7436,
+      "step": 2460
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7917252422845774e-06,
+      "loss": 0.7456,
+      "step": 2461
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.784763165430282e-06,
+      "loss": 0.7376,
+      "step": 2462
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7778083759716945e-06,
+      "loss": 0.7365,
+      "step": 2463
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7708608809331592e-06,
+      "loss": 0.7551,
+      "step": 2464
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7639206873316516e-06,
+      "loss": 0.7657,
+      "step": 2465
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.756987802176777e-06,
+      "loss": 0.7426,
+      "step": 2466
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7500622324707504e-06,
+      "loss": 0.7462,
+      "step": 2467
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.743143985208407e-06,
+      "loss": 0.7507,
+      "step": 2468
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.73623306737718e-06,
+      "loss": 0.7512,
+      "step": 2469
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7293294859571027e-06,
+      "loss": 0.7532,
+      "step": 2470
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7224332479208006e-06,
+      "loss": 0.7373,
+      "step": 2471
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7155443602334796e-06,
+      "loss": 0.7367,
+      "step": 2472
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708662829852925e-06,
+      "loss": 0.7373,
+      "step": 2473
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701788663729483e-06,
+      "loss": 0.7446,
+      "step": 2474
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.694921868806072e-06,
+      "loss": 0.7587,
+      "step": 2475
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6880624520181586e-06,
+      "loss": 0.7466,
+      "step": 2476
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6812104202937616e-06,
+      "loss": 0.7357,
+      "step": 2477
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6743657805534396e-06,
+      "loss": 0.7754,
+      "step": 2478
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6675285397102856e-06,
+      "loss": 0.7362,
+      "step": 2479
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6606987046699185e-06,
+      "loss": 0.7586,
+      "step": 2480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653876282330473e-06,
+      "loss": 0.7367,
+      "step": 2481
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.647061279582607e-06,
+      "loss": 0.7163,
+      "step": 2482
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6402537033094764e-06,
+      "loss": 0.7525,
+      "step": 2483
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.633453560386743e-06,
+      "loss": 0.7351,
+      "step": 2484
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6266608576825535e-06,
+      "loss": 0.7433,
+      "step": 2485
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61987560205755e-06,
+      "loss": 0.7677,
+      "step": 2486
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6130978003648388e-06,
+      "loss": 0.7604,
+      "step": 2487
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.606327459450009e-06,
+      "loss": 0.733,
+      "step": 2488
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.599564586151112e-06,
+      "loss": 0.7555,
+      "step": 2489
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.592809187298655e-06,
+      "loss": 0.739,
+      "step": 2490
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5860612697155964e-06,
+      "loss": 0.7792,
+      "step": 2491
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5793208402173442e-06,
+      "loss": 0.7444,
+      "step": 2492
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572587905611733e-06,
+      "loss": 0.7504,
+      "step": 2493
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.565862472699031e-06,
+      "loss": 0.7531,
+      "step": 2494
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5591445482719344e-06,
+      "loss": 0.7337,
+      "step": 2495
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.552434139115553e-06,
+      "loss": 0.7555,
+      "step": 2496
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.545731252007405e-06,
+      "loss": 0.7418,
+      "step": 2497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5390358937174164e-06,
+      "loss": 0.7777,
+      "step": 2498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5323480710078997e-06,
+      "loss": 0.7505,
+      "step": 2499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5256677906335638e-06,
+      "loss": 0.7446,
+      "step": 2500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518995059341499e-06,
+      "loss": 0.7644,
+      "step": 2501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5123298838711685e-06,
+      "loss": 0.7382,
+      "step": 2502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505672270954406e-06,
+      "loss": 0.7271,
+      "step": 2503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4990222273154096e-06,
+      "loss": 0.7152,
+      "step": 2504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4923797596707277e-06,
+      "loss": 0.748,
+      "step": 2505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4857448747292536e-06,
+      "loss": 0.7586,
+      "step": 2506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4791175791922316e-06,
+      "loss": 0.7372,
+      "step": 2507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.472497879753235e-06,
+      "loss": 0.7561,
+      "step": 2508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.465885783098166e-06,
+      "loss": 0.7515,
+      "step": 2509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4592812959052503e-06,
+      "loss": 0.7483,
+      "step": 2510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4526844248450264e-06,
+      "loss": 0.7342,
+      "step": 2511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4460951765803354e-06,
+      "loss": 0.7369,
+      "step": 2512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4395135577663277e-06,
+      "loss": 0.7639,
+      "step": 2513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4329395750504414e-06,
+      "loss": 0.7343,
+      "step": 2514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.426373235072408e-06,
+      "loss": 0.7311,
+      "step": 2515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4198145444642362e-06,
+      "loss": 0.7524,
+      "step": 2516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.413263509850211e-06,
+      "loss": 0.7505,
+      "step": 2517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4067201378468808e-06,
+      "loss": 0.7686,
+      "step": 2518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4001844350630556e-06,
+      "loss": 0.7561,
+      "step": 2519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3936564080998016e-06,
+      "loss": 0.7375,
+      "step": 2520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3871360635504347e-06,
+      "loss": 0.7685,
+      "step": 2521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3806234080005064e-06,
+      "loss": 0.7479,
+      "step": 2522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3741184480278067e-06,
+      "loss": 0.7355,
+      "step": 2523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3676211902023515e-06,
+      "loss": 0.7509,
+      "step": 2524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3611316410863737e-06,
+      "loss": 0.7541,
+      "step": 2525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354649807234326e-06,
+      "loss": 0.7419,
+      "step": 2526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.348175695192866e-06,
+      "loss": 0.7246,
+      "step": 2527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3417093115008527e-06,
+      "loss": 0.717,
+      "step": 2528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.335250662689341e-06,
+      "loss": 0.7737,
+      "step": 2529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.328799755281572e-06,
+      "loss": 0.7418,
+      "step": 2530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322356595792964e-06,
+      "loss": 0.741,
+      "step": 2531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315921190731116e-06,
+      "loss": 0.7515,
+      "step": 2532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3094935465957934e-06,
+      "loss": 0.7584,
+      "step": 2533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3030736698789226e-06,
+      "loss": 0.7464,
+      "step": 2534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.296661567064584e-06,
+      "loss": 0.7529,
+      "step": 2535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.290257244629013e-06,
+      "loss": 0.7759,
+      "step": 2536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2838607090405707e-06,
+      "loss": 0.7319,
+      "step": 2537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277471966759771e-06,
+      "loss": 0.7567,
+      "step": 2538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2710910242392468e-06,
+      "loss": 0.754,
+      "step": 2539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2647178879237562e-06,
+      "loss": 0.7632,
+      "step": 2540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2583525642501734e-06,
+      "loss": 0.7571,
+      "step": 2541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2519950596474836e-06,
+      "loss": 0.7568,
+      "step": 2542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2456453805367695e-06,
+      "loss": 0.7539,
+      "step": 2543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2393035333312097e-06,
+      "loss": 0.7451,
+      "step": 2544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2329695244360784e-06,
+      "loss": 0.719,
+      "step": 2545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2266433602487293e-06,
+      "loss": 0.7266,
+      "step": 2546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2203250471585955e-06,
+      "loss": 0.7716,
+      "step": 2547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2140145915471778e-06,
+      "loss": 0.7634,
+      "step": 2548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2077119997880456e-06,
+      "loss": 0.7393,
+      "step": 2549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.201417278246817e-06,
+      "loss": 0.7388,
+      "step": 2550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1951304332811672e-06,
+      "loss": 0.7422,
+      "step": 2551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1888514712408184e-06,
+      "loss": 0.7316,
+      "step": 2552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1825803984675252e-06,
+      "loss": 0.7402,
+      "step": 2553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.176317221295081e-06,
+      "loss": 0.7622,
+      "step": 2554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1700619460492944e-06,
+      "loss": 0.7503,
+      "step": 2555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1638145790480035e-06,
+      "loss": 0.778,
+      "step": 2556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1575751266010513e-06,
+      "loss": 0.7695,
+      "step": 2557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.151343595010292e-06,
+      "loss": 0.7465,
+      "step": 2558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1451199905695787e-06,
+      "loss": 0.7572,
+      "step": 2559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.138904319564755e-06,
+      "loss": 0.7592,
+      "step": 2560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1326965882736605e-06,
+      "loss": 0.7422,
+      "step": 2561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1264968029661024e-06,
+      "loss": 0.7305,
+      "step": 2562
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1203049699038714e-06,
+      "loss": 0.7357,
+      "step": 2563
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1141210953407256e-06,
+      "loss": 0.7319,
+      "step": 2564
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107945185522383e-06,
+      "loss": 0.7566,
+      "step": 2565
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1017772466865195e-06,
+      "loss": 0.7303,
+      "step": 2566
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09561728506276e-06,
+      "loss": 0.7494,
+      "step": 2567
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.089465306872669e-06,
+      "loss": 0.7295,
+      "step": 2568
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0833213183297475e-06,
+      "loss": 0.7411,
+      "step": 2569
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0771853256394326e-06,
+      "loss": 0.7575,
+      "step": 2570
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0710573349990825e-06,
+      "loss": 0.7374,
+      "step": 2571
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064937352597972e-06,
+      "loss": 0.7417,
+      "step": 2572
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05882538461729e-06,
+      "loss": 0.7582,
+      "step": 2573
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0527214372301317e-06,
+      "loss": 0.7528,
+      "step": 2574
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0466255166014847e-06,
+      "loss": 0.7291,
+      "step": 2575
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.040537628888237e-06,
+      "loss": 0.7491,
+      "step": 2576
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034457780239161e-06,
+      "loss": 0.7478,
+      "step": 2577
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0283859767949077e-06,
+      "loss": 0.7497,
+      "step": 2578
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0223222246880083e-06,
+      "loss": 0.7544,
+      "step": 2579
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.016266530042852e-06,
+      "loss": 0.764,
+      "step": 2580
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.010218898975701e-06,
+      "loss": 0.7446,
+      "step": 2581
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0041793375946626e-06,
+      "loss": 0.7325,
+      "step": 2582
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9981478519997035e-06,
+      "loss": 0.7317,
+      "step": 2583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9921244482826275e-06,
+      "loss": 0.7502,
+      "step": 2584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9861091325270785e-06,
+      "loss": 0.7121,
+      "step": 2585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.980101910808534e-06,
+      "loss": 0.7402,
+      "step": 2586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9741027891942887e-06,
+      "loss": 0.7333,
+      "step": 2587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968111773743461e-06,
+      "loss": 0.7338,
+      "step": 2588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.962128870506984e-06,
+      "loss": 0.7483,
+      "step": 2589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956154085527595e-06,
+      "loss": 0.774,
+      "step": 2590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9501874248398324e-06,
+      "loss": 0.7332,
+      "step": 2591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9442288944700304e-06,
+      "loss": 0.708,
+      "step": 2592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938278500436308e-06,
+      "loss": 0.7664,
+      "step": 2593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9323362487485665e-06,
+      "loss": 0.7569,
+      "step": 2594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.926402145408488e-06,
+      "loss": 0.7541,
+      "step": 2595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920476196409521e-06,
+      "loss": 0.7307,
+      "step": 2596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.914558407736881e-06,
+      "loss": 0.7733,
+      "step": 2597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9086487853675383e-06,
+      "loss": 0.7469,
+      "step": 2598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9027473352702208e-06,
+      "loss": 0.735,
+      "step": 2599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8968540634053922e-06,
+      "loss": 0.7533,
+      "step": 2600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8909689757252658e-06,
+      "loss": 0.7281,
+      "step": 2601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8850920781737847e-06,
+      "loss": 0.7255,
+      "step": 2602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8792233766866208e-06,
+      "loss": 0.7277,
+      "step": 2603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8733628771911695e-06,
+      "loss": 0.7298,
+      "step": 2604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.867510585606541e-06,
+      "loss": 0.746,
+      "step": 2605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8616665078435536e-06,
+      "loss": 0.7488,
+      "step": 2606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8558306498047275e-06,
+      "loss": 0.7466,
+      "step": 2607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8500030173842886e-06,
+      "loss": 0.7387,
+      "step": 2608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8441836164681504e-06,
+      "loss": 0.7474,
+      "step": 2609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838372452933912e-06,
+      "loss": 0.7314,
+      "step": 2610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8325695326508585e-06,
+      "loss": 0.729,
+      "step": 2611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8267748614799373e-06,
+      "loss": 0.7519,
+      "step": 2612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8209884452737758e-06,
+      "loss": 0.7652,
+      "step": 2613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8152102898766587e-06,
+      "loss": 0.7359,
+      "step": 2614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8094404011245281e-06,
+      "loss": 0.7136,
+      "step": 2615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8036787848449767e-06,
+      "loss": 0.7402,
+      "step": 2616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.797925446857245e-06,
+      "loss": 0.7535,
+      "step": 2617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.792180392972208e-06,
+      "loss": 0.7479,
+      "step": 2618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7864436289923716e-06,
+      "loss": 0.757,
+      "step": 2619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7807151607118745e-06,
+      "loss": 0.7339,
+      "step": 2620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7749949939164758e-06,
+      "loss": 0.7307,
+      "step": 2621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7692831343835492e-06,
+      "loss": 0.7462,
+      "step": 2622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7635795878820782e-06,
+      "loss": 0.7445,
+      "step": 2623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7578843601726513e-06,
+      "loss": 0.7653,
+      "step": 2624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7521974570074496e-06,
+      "loss": 0.7603,
+      "step": 2625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7465188841302527e-06,
+      "loss": 0.7182,
+      "step": 2626
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.740848647276423e-06,
+      "loss": 0.7641,
+      "step": 2627
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.735186752172907e-06,
+      "loss": 0.7517,
+      "step": 2628
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.729533204538224e-06,
+      "loss": 0.7274,
+      "step": 2629
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.723888010082463e-06,
+      "loss": 0.7542,
+      "step": 2630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7182511745072728e-06,
+      "loss": 0.7477,
+      "step": 2631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7126227035058608e-06,
+      "loss": 0.7268,
+      "step": 2632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7070026027629904e-06,
+      "loss": 0.7436,
+      "step": 2633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7013908779549693e-06,
+      "loss": 0.751,
+      "step": 2634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6957875347496422e-06,
+      "loss": 0.7524,
+      "step": 2635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6901925788063933e-06,
+      "loss": 0.7452,
+      "step": 2636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6846060157761334e-06,
+      "loss": 0.7531,
+      "step": 2637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6790278513012924e-06,
+      "loss": 0.7341,
+      "step": 2638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.673458091015825e-06,
+      "loss": 0.7233,
+      "step": 2639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6678967405451919e-06,
+      "loss": 0.7546,
+      "step": 2640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6623438055063634e-06,
+      "loss": 0.7329,
+      "step": 2641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6567992915078113e-06,
+      "loss": 0.746,
+      "step": 2642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6512632041494957e-06,
+      "loss": 0.7344,
+      "step": 2643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.64573554902287e-06,
+      "loss": 0.7355,
+      "step": 2644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6402163317108722e-06,
+      "loss": 0.7555,
+      "step": 2645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6347055577879168e-06,
+      "loss": 0.7435,
+      "step": 2646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6292032328198892e-06,
+      "loss": 0.7387,
+      "step": 2647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6237093623641442e-06,
+      "loss": 0.7465,
+      "step": 2648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6182239519694986e-06,
+      "loss": 0.7396,
+      "step": 2649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6127470071762152e-06,
+      "loss": 0.7355,
+      "step": 2650
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6072785335160167e-06,
+      "loss": 0.7485,
+      "step": 2651
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6018185365120663e-06,
+      "loss": 0.7714,
+      "step": 2652
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.596367021678965e-06,
+      "loss": 0.7356,
+      "step": 2653
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5909239945227494e-06,
+      "loss": 0.7506,
+      "step": 2654
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5854894605408843e-06,
+      "loss": 0.7439,
+      "step": 2655
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.580063425222249e-06,
+      "loss": 0.7479,
+      "step": 2656
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5746458940471443e-06,
+      "loss": 0.7379,
+      "step": 2657
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.569236872487283e-06,
+      "loss": 0.7351,
+      "step": 2658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.563836366005782e-06,
+      "loss": 0.7561,
+      "step": 2659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.558444380057157e-06,
+      "loss": 0.7382,
+      "step": 2660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5530609200873193e-06,
+      "loss": 0.7419,
+      "step": 2661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5476859915335695e-06,
+      "loss": 0.7251,
+      "step": 2662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5423195998245877e-06,
+      "loss": 0.7318,
+      "step": 2663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5369617503804358e-06,
+      "loss": 0.7352,
+      "step": 2664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.531612448612545e-06,
+      "loss": 0.7439,
+      "step": 2665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526271699923717e-06,
+      "loss": 0.7479,
+      "step": 2666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5209395097081149e-06,
+      "loss": 0.7486,
+      "step": 2667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5156158833512523e-06,
+      "loss": 0.7372,
+      "step": 2668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5103008262299944e-06,
+      "loss": 0.7225,
+      "step": 2669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5049943437125568e-06,
+      "loss": 0.7439,
+      "step": 2670
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4996964411584914e-06,
+      "loss": 0.7412,
+      "step": 2671
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.494407123918684e-06,
+      "loss": 0.7181,
+      "step": 2672
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.489126397335351e-06,
+      "loss": 0.7544,
+      "step": 2673
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4838542667420307e-06,
+      "loss": 0.7453,
+      "step": 2674
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4785907374635789e-06,
+      "loss": 0.7554,
+      "step": 2675
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4733358148161635e-06,
+      "loss": 0.7343,
+      "step": 2676
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468089504107264e-06,
+      "loss": 0.7477,
+      "step": 2677
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.462851810635658e-06,
+      "loss": 0.7393,
+      "step": 2678
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.45762273969142e-06,
+      "loss": 0.7595,
+      "step": 2679
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4524022965559182e-06,
+      "loss": 0.7299,
+      "step": 2680
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4471904865018049e-06,
+      "loss": 0.7498,
+      "step": 2681
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4419873147930086e-06,
+      "loss": 0.7608,
+      "step": 2682
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4367927866847398e-06,
+      "loss": 0.7467,
+      "step": 2683
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4316069074234772e-06,
+      "loss": 0.7629,
+      "step": 2684
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4264296822469636e-06,
+      "loss": 0.7564,
+      "step": 2685
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4212611163842017e-06,
+      "loss": 0.7435,
+      "step": 2686
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4161012150554486e-06,
+      "loss": 0.7416,
+      "step": 2687
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.410949983472205e-06,
+      "loss": 0.725,
+      "step": 2688
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4058074268372224e-06,
+      "loss": 0.7491,
+      "step": 2689
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4006735503444868e-06,
+      "loss": 0.7388,
+      "step": 2690
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3955483591792197e-06,
+      "loss": 0.7359,
+      "step": 2691
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390431858517869e-06,
+      "loss": 0.7339,
+      "step": 2692
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.385324053528102e-06,
+      "loss": 0.7257,
+      "step": 2693
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.380224949368809e-06,
+      "loss": 0.756,
+      "step": 2694
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3751345511900882e-06,
+      "loss": 0.7334,
+      "step": 2695
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3700528641332456e-06,
+      "loss": 0.743,
+      "step": 2696
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3649798933307923e-06,
+      "loss": 0.7376,
+      "step": 2697
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.359915643906431e-06,
+      "loss": 0.7405,
+      "step": 2698
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3548601209750623e-06,
+      "loss": 0.7348,
+      "step": 2699
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3498133296427628e-06,
+      "loss": 0.7378,
+      "step": 2700
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3447752750068e-06,
+      "loss": 0.736,
+      "step": 2701
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.7518,
+      "step": 2702
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3347253961688134e-06,
+      "loss": 0.7283,
+      "step": 2703
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3297135821171748e-06,
+      "loss": 0.7312,
+      "step": 2704
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.324710525062638e-06,
+      "loss": 0.7363,
+      "step": 2705
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3197162300582933e-06,
+      "loss": 0.7422,
+      "step": 2706
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3147307021483801e-06,
+      "loss": 0.7369,
+      "step": 2707
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3097539463682873e-06,
+      "loss": 0.7423,
+      "step": 2708
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3047859677445451e-06,
+      "loss": 0.7293,
+      "step": 2709
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2998267712948154e-06,
+      "loss": 0.7399,
+      "step": 2710
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2948763620278893e-06,
+      "loss": 0.7539,
+      "step": 2711
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2899347449436893e-06,
+      "loss": 0.7289,
+      "step": 2712
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.285001925033248e-06,
+      "loss": 0.7246,
+      "step": 2713
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2800779072787194e-06,
+      "loss": 0.7392,
+      "step": 2714
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275162696653367e-06,
+      "loss": 0.7642,
+      "step": 2715
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.270256298121557e-06,
+      "loss": 0.7396,
+      "step": 2716
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2653587166387593e-06,
+      "loss": 0.7564,
+      "step": 2717
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.26046995715153e-06,
+      "loss": 0.7112,
+      "step": 2718
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2555900245975262e-06,
+      "loss": 0.7217,
+      "step": 2719
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2507189239054796e-06,
+      "loss": 0.7401,
+      "step": 2720
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2458566599952082e-06,
+      "loss": 0.7224,
+      "step": 2721
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2410032377776028e-06,
+      "loss": 0.7357,
+      "step": 2722
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.236158662154624e-06,
+      "loss": 0.7479,
+      "step": 2723
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2313229380193004e-06,
+      "loss": 0.7424,
+      "step": 2724
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.226496070255714e-06,
+      "loss": 0.7469,
+      "step": 2725
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2216780637390058e-06,
+      "loss": 0.7466,
+      "step": 2726
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.216868923335368e-06,
+      "loss": 0.7396,
+      "step": 2727
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2120686539020377e-06,
+      "loss": 0.7348,
+      "step": 2728
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2072772602872896e-06,
+      "loss": 0.7487,
+      "step": 2729
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2024947473304382e-06,
+      "loss": 0.7398,
+      "step": 2730
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1977211198618232e-06,
+      "loss": 0.7559,
+      "step": 2731
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1929563827028123e-06,
+      "loss": 0.7288,
+      "step": 2732
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1882005406657947e-06,
+      "loss": 0.749,
+      "step": 2733
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1834535985541762e-06,
+      "loss": 0.749,
+      "step": 2734
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1787155611623712e-06,
+      "loss": 0.7571,
+      "step": 2735
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1739864332758044e-06,
+      "loss": 0.7515,
+      "step": 2736
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.169266219670898e-06,
+      "loss": 0.7364,
+      "step": 2737
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1645549251150712e-06,
+      "loss": 0.7282,
+      "step": 2738
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.159852554366735e-06,
+      "loss": 0.7535,
+      "step": 2739
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1551591121752891e-06,
+      "loss": 0.7393,
+      "step": 2740
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1504746032811153e-06,
+      "loss": 0.7598,
+      "step": 2741
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1457990324155732e-06,
+      "loss": 0.751,
+      "step": 2742
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1411324043009887e-06,
+      "loss": 0.7454,
+      "step": 2743
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1364747236506646e-06,
+      "loss": 0.7243,
+      "step": 2744
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1318259951688603e-06,
+      "loss": 0.7451,
+      "step": 2745
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1271862235507958e-06,
+      "loss": 0.7348,
+      "step": 2746
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1225554134826455e-06,
+      "loss": 0.7623,
+      "step": 2747
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1179335696415305e-06,
+      "loss": 0.7628,
+      "step": 2748
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1133206966955213e-06,
+      "loss": 0.7151,
+      "step": 2749
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1087167993036185e-06,
+      "loss": 0.7504,
+      "step": 2750
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1041218821157652e-06,
+      "loss": 0.7479,
+      "step": 2751
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0995359497728319e-06,
+      "loss": 0.7534,
+      "step": 2752
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0949590069066129e-06,
+      "loss": 0.753,
+      "step": 2753
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0903910581398269e-06,
+      "loss": 0.7353,
+      "step": 2754
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0858321080861067e-06,
+      "loss": 0.7361,
+      "step": 2755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0812821613499935e-06,
+      "loss": 0.7285,
+      "step": 2756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.076741222526938e-06,
+      "loss": 0.7205,
+      "step": 2757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0722092962032927e-06,
+      "loss": 0.7674,
+      "step": 2758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.067686386956307e-06,
+      "loss": 0.7226,
+      "step": 2759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0631724993541236e-06,
+      "loss": 0.7334,
+      "step": 2760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0586676379557748e-06,
+      "loss": 0.7214,
+      "step": 2761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054171807311174e-06,
+      "loss": 0.7366,
+      "step": 2762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0496850119611114e-06,
+      "loss": 0.7275,
+      "step": 2763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0452072564372573e-06,
+      "loss": 0.7475,
+      "step": 2764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0407385452621476e-06,
+      "loss": 0.7375,
+      "step": 2765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0362788829491843e-06,
+      "loss": 0.7341,
+      "step": 2766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.031828274002633e-06,
+      "loss": 0.7502,
+      "step": 2767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0273867229176092e-06,
+      "loss": 0.7364,
+      "step": 2768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0229542341800868e-06,
+      "loss": 0.7286,
+      "step": 2769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0185308122668802e-06,
+      "loss": 0.726,
+      "step": 2770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0141164616456511e-06,
+      "loss": 0.7471,
+      "step": 2771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0097111867748966e-06,
+      "loss": 0.7524,
+      "step": 2772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0053149921039495e-06,
+      "loss": 0.7503,
+      "step": 2773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0009278820729706e-06,
+      "loss": 0.7306,
+      "step": 2774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.965498611129453e-07,
+      "loss": 0.7559,
+      "step": 2775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.921809336456767e-07,
+      "loss": 0.7347,
+      "step": 2776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.87821104083786e-07,
+      "loss": 0.7418,
+      "step": 2777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.834703768307063e-07,
+      "loss": 0.7406,
+      "step": 2778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.79128756280675e-07,
+      "loss": 0.7303,
+      "step": 2779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.747962468187356e-07,
+      "loss": 0.7402,
+      "step": 2780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.704728528207242e-07,
+      "loss": 0.7439,
+      "step": 2781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66158578653269e-07,
+      "loss": 0.7239,
+      "step": 2782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.618534286737958e-07,
+      "loss": 0.7569,
+      "step": 2783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.575574072305083e-07,
+      "loss": 0.7548,
+      "step": 2784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.532705186623925e-07,
+      "loss": 0.7259,
+      "step": 2785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.489927672992083e-07,
+      "loss": 0.7505,
+      "step": 2786
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.44724157461493e-07,
+      "loss": 0.7311,
+      "step": 2787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.404646934605399e-07,
+      "loss": 0.727,
+      "step": 2788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.362143795984147e-07,
+      "loss": 0.7225,
+      "step": 2789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.319732201679388e-07,
+      "loss": 0.7298,
+      "step": 2790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.277412194526858e-07,
+      "loss": 0.7357,
+      "step": 2791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.235183817269821e-07,
+      "loss": 0.7438,
+      "step": 2792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.193047112558951e-07,
+      "loss": 0.7373,
+      "step": 2793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.15100212295239e-07,
+      "loss": 0.7294,
+      "step": 2794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.109048890915573e-07,
+      "loss": 0.7475,
+      "step": 2795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.067187458821325e-07,
+      "loss": 0.7514,
+      "step": 2796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.025417868949726e-07,
+      "loss": 0.7527,
+      "step": 2797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.983740163488108e-07,
+      "loss": 0.753,
+      "step": 2798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.942154384530987e-07,
+      "loss": 0.7285,
+      "step": 2799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.900660574080056e-07,
+      "loss": 0.7449,
+      "step": 2800
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 3223,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 3.6817668464395577e+22,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}