diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7030 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 7.972103502423704,
+  "eval_steps": 1000000,
+  "global_step": 583837,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 10.3012,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 10.1562,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 10.056,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 9.9445,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5e-06,
+      "loss": 9.8008,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 9.6288,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.5e-06,
+      "loss": 9.4293,
+      "step": 3500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 9.2098,
+      "step": 4000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.5e-06,
+      "loss": 8.9745,
+      "step": 4500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5e-06,
+      "loss": 8.7257,
+      "step": 5000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 8.4764,
+      "step": 5500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6e-06,
+      "loss": 8.2289,
+      "step": 6000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 7.9941,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7e-06,
+      "loss": 7.7749,
+      "step": 7000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 7.5881,
+      "step": 7500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 7.4356,
+      "step": 8000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.5e-06,
+      "loss": 7.3222,
+      "step": 8500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9e-06,
+      "loss": 7.2418,
+      "step": 9000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.5e-06,
+      "loss": 7.1829,
+      "step": 9500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 7.1432,
+      "step": 10000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.999316332809189e-06,
+      "loss": 7.0945,
+      "step": 10500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.998632665618377e-06,
+      "loss": 7.0529,
+      "step": 11000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.997948998427567e-06,
+      "loss": 7.0211,
+      "step": 11500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.997265331236755e-06,
+      "loss": 6.9849,
+      "step": 12000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.996581664045943e-06,
+      "loss": 6.9597,
+      "step": 12500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.995897996855133e-06,
+      "loss": 6.9323,
+      "step": 13000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.99521432966432e-06,
+      "loss": 6.9077,
+      "step": 13500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.994530662473509e-06,
+      "loss": 6.8872,
+      "step": 14000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.993846995282698e-06,
+      "loss": 6.8663,
+      "step": 14500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.993163328091886e-06,
+      "loss": 6.8463,
+      "step": 15000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.992479660901074e-06,
+      "loss": 6.8294,
+      "step": 15500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.991795993710262e-06,
+      "loss": 6.8188,
+      "step": 16000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.99111232651945e-06,
+      "loss": 6.7988,
+      "step": 16500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.990428659328639e-06,
+      "loss": 6.7872,
+      "step": 17000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.989744992137828e-06,
+      "loss": 6.7705,
+      "step": 17500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.989061324947016e-06,
+      "loss": 6.7631,
+      "step": 18000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.988377657756204e-06,
+      "loss": 6.7477,
+      "step": 18500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.987693990565394e-06,
+      "loss": 6.7346,
+      "step": 19000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.987010323374582e-06,
+      "loss": 6.7284,
+      "step": 19500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.98632665618377e-06,
+      "loss": 6.713,
+      "step": 20000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.98564298899296e-06,
+      "loss": 6.703,
+      "step": 20500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.984959321802148e-06,
+      "loss": 6.6975,
+      "step": 21000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.984275654611336e-06,
+      "loss": 6.6913,
+      "step": 21500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.983591987420524e-06,
+      "loss": 6.6759,
+      "step": 22000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.982908320229712e-06,
+      "loss": 6.6744,
+      "step": 22500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.9822246530389e-06,
+      "loss": 6.664,
+      "step": 23000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.98154098584809e-06,
+      "loss": 6.6572,
+      "step": 23500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.980857318657278e-06,
+      "loss": 6.6452,
+      "step": 24000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.980173651466468e-06,
+      "loss": 6.6432,
+      "step": 24500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.979489984275656e-06,
+      "loss": 6.6371,
+      "step": 25000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.978806317084844e-06,
+      "loss": 6.6298,
+      "step": 25500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.978122649894032e-06,
+      "loss": 6.6217,
+      "step": 26000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.977438982703222e-06,
+      "loss": 6.6183,
+      "step": 26500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.97675531551241e-06,
+      "loss": 6.6071,
+      "step": 27000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.976071648321598e-06,
+      "loss": 6.5981,
+      "step": 27500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.975387981130786e-06,
+      "loss": 6.5907,
+      "step": 28000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.974704313939974e-06,
+      "loss": 6.5957,
+      "step": 28500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.974020646749164e-06,
+      "loss": 6.5843,
+      "step": 29000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.973336979558352e-06,
+      "loss": 6.5783,
+      "step": 29500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.972653312367541e-06,
+      "loss": 6.5729,
+      "step": 30000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.97196964517673e-06,
+      "loss": 6.5692,
+      "step": 30500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.971285977985918e-06,
+      "loss": 6.5644,
+      "step": 31000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.970602310795106e-06,
+      "loss": 6.5518,
+      "step": 31500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.969918643604294e-06,
+      "loss": 6.5433,
+      "step": 32000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.969234976413483e-06,
+      "loss": 6.547,
+      "step": 32500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.968551309222671e-06,
+      "loss": 6.5378,
+      "step": 33000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.96786764203186e-06,
+      "loss": 6.5331,
+      "step": 33500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.967183974841048e-06,
+      "loss": 6.5352,
+      "step": 34000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.966500307650237e-06,
+      "loss": 6.5288,
+      "step": 34500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.965816640459425e-06,
+      "loss": 6.5173,
+      "step": 35000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.965132973268613e-06,
+      "loss": 6.5179,
+      "step": 35500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.964449306077803e-06,
+      "loss": 6.5172,
+      "step": 36000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.963765638886991e-06,
+      "loss": 6.5101,
+      "step": 36500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.96308197169618e-06,
+      "loss": 6.5061,
+      "step": 37000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.962398304505367e-06,
+      "loss": 6.5004,
+      "step": 37500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961714637314555e-06,
+      "loss": 6.4972,
+      "step": 38000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.961030970123745e-06,
+      "loss": 6.4884,
+      "step": 38500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.960347302932933e-06,
+      "loss": 6.4864,
+      "step": 39000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.959663635742121e-06,
+      "loss": 6.4921,
+      "step": 39500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.95897996855131e-06,
+      "loss": 6.4805,
+      "step": 40000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.958296301360499e-06,
+      "loss": 6.4801,
+      "step": 40500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.957612634169687e-06,
+      "loss": 6.4634,
+      "step": 41000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.956928966978877e-06,
+      "loss": 6.4642,
+      "step": 41500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.956245299788065e-06,
+      "loss": 6.4615,
+      "step": 42000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.955561632597253e-06,
+      "loss": 6.463,
+      "step": 42500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.954877965406441e-06,
+      "loss": 6.4568,
+      "step": 43000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.954194298215629e-06,
+      "loss": 6.4548,
+      "step": 43500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.953510631024817e-06,
+      "loss": 6.4519,
+      "step": 44000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.952826963834007e-06,
+      "loss": 6.4511,
+      "step": 44500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.952143296643195e-06,
+      "loss": 6.444,
+      "step": 45000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.951459629452383e-06,
+      "loss": 6.4376,
+      "step": 45500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.950775962261573e-06,
+      "loss": 6.4374,
+      "step": 46000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.95009229507076e-06,
+      "loss": 6.4345,
+      "step": 46500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.949408627879949e-06,
+      "loss": 6.4298,
+      "step": 47000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.948724960689138e-06,
+      "loss": 6.4256,
+      "step": 47500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.948041293498326e-06,
+      "loss": 6.4305,
+      "step": 48000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.947357626307515e-06,
+      "loss": 6.4183,
+      "step": 48500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.946673959116703e-06,
+      "loss": 6.4175,
+      "step": 49000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.94599029192589e-06,
+      "loss": 6.4148,
+      "step": 49500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.945306624735079e-06,
+      "loss": 6.4048,
+      "step": 50000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.944622957544268e-06,
+      "loss": 6.409,
+      "step": 50500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.943939290353456e-06,
+      "loss": 6.4053,
+      "step": 51000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.943255623162645e-06,
+      "loss": 6.4042,
+      "step": 51500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.942571955971834e-06,
+      "loss": 6.3958,
+      "step": 52000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.941888288781022e-06,
+      "loss": 6.4009,
+      "step": 52500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.94120462159021e-06,
+      "loss": 6.3901,
+      "step": 53000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.9405209543994e-06,
+      "loss": 6.3928,
+      "step": 53500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.939837287208588e-06,
+      "loss": 6.3881,
+      "step": 54000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.939153620017776e-06,
+      "loss": 6.3878,
+      "step": 54500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.938469952826964e-06,
+      "loss": 6.3814,
+      "step": 55000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.937786285636152e-06,
+      "loss": 6.3781,
+      "step": 55500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.93710261844534e-06,
+      "loss": 6.3787,
+      "step": 56000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.93641895125453e-06,
+      "loss": 6.3794,
+      "step": 56500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.935735284063718e-06,
+      "loss": 6.3657,
+      "step": 57000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.935051616872908e-06,
+      "loss": 6.3719,
+      "step": 57500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.934367949682096e-06,
+      "loss": 6.3669,
+      "step": 58000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.933684282491284e-06,
+      "loss": 6.3654,
+      "step": 58500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.933000615300472e-06,
+      "loss": 6.365,
+      "step": 59000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.932316948109662e-06,
+      "loss": 6.3628,
+      "step": 59500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.93163328091885e-06,
+      "loss": 6.3582,
+      "step": 60000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.930949613728038e-06,
+      "loss": 6.352,
+      "step": 60500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.930265946537226e-06,
+      "loss": 6.3538,
+      "step": 61000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.929582279346414e-06,
+      "loss": 6.3532,
+      "step": 61500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.928898612155604e-06,
+      "loss": 6.349,
+      "step": 62000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.928214944964792e-06,
+      "loss": 6.3481,
+      "step": 62500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.927531277773982e-06,
+      "loss": 6.3392,
+      "step": 63000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.92684761058317e-06,
+      "loss": 6.3442,
+      "step": 63500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.926163943392358e-06,
+      "loss": 6.345,
+      "step": 64000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.925480276201546e-06,
+      "loss": 6.346,
+      "step": 64500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.924796609010734e-06,
+      "loss": 6.336,
+      "step": 65000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.924112941819923e-06,
+      "loss": 6.3371,
+      "step": 65500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.923429274629112e-06,
+      "loss": 6.3332,
+      "step": 66000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.9227456074383e-06,
+      "loss": 6.3305,
+      "step": 66500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.922061940247488e-06,
+      "loss": 6.3311,
+      "step": 67000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.921378273056677e-06,
+      "loss": 6.3302,
+      "step": 67500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.920694605865865e-06,
+      "loss": 6.3308,
+      "step": 68000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.920010938675053e-06,
+      "loss": 6.3291,
+      "step": 68500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.919327271484243e-06,
+      "loss": 6.3207,
+      "step": 69000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.918643604293431e-06,
+      "loss": 6.325,
+      "step": 69500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.91795993710262e-06,
+      "loss": 6.3192,
+      "step": 70000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.917276269911807e-06,
+      "loss": 6.3215,
+      "step": 70500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.916592602720995e-06,
+      "loss": 6.3154,
+      "step": 71000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.915908935530185e-06,
+      "loss": 6.3175,
+      "step": 71500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.915225268339373e-06,
+      "loss": 6.3108,
+      "step": 72000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.914541601148561e-06,
+      "loss": 6.3109,
+      "step": 72500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.91385793395775e-06,
+      "loss": 6.3087,
+      "step": 73000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.913174266766939e-06,
+      "loss": 6.3064,
+      "step": 73500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.912490599576127e-06,
+      "loss": 6.3092,
+      "step": 74000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.911806932385317e-06,
+      "loss": 6.3052,
+      "step": 74500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.911123265194505e-06,
+      "loss": 6.3046,
+      "step": 75000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.910439598003693e-06,
+      "loss": 6.3058,
+      "step": 75500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.909755930812881e-06,
+      "loss": 6.2966,
+      "step": 76000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.909072263622069e-06,
+      "loss": 6.2924,
+      "step": 76500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.908388596431257e-06,
+      "loss": 6.2923,
+      "step": 77000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.907704929240447e-06,
+      "loss": 6.2968,
+      "step": 77500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.907021262049635e-06,
+      "loss": 6.2951,
+      "step": 78000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.906337594858823e-06,
+      "loss": 6.2943,
+      "step": 78500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.905653927668013e-06,
+      "loss": 6.2905,
+      "step": 79000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.9049702604772e-06,
+      "loss": 6.2916,
+      "step": 79500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.904286593286389e-06,
+      "loss": 6.2853,
+      "step": 80000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.903602926095579e-06,
+      "loss": 6.2889,
+      "step": 80500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.902919258904767e-06,
+      "loss": 6.2781,
+      "step": 81000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.902235591713955e-06,
+      "loss": 6.2828,
+      "step": 81500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.901551924523143e-06,
+      "loss": 6.2839,
+      "step": 82000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.90086825733233e-06,
+      "loss": 6.2813,
+      "step": 82500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.900184590141519e-06,
+      "loss": 6.2765,
+      "step": 83000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.899500922950709e-06,
+      "loss": 6.2794,
+      "step": 83500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.898817255759897e-06,
+      "loss": 6.2794,
+      "step": 84000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.898133588569086e-06,
+      "loss": 6.2756,
+      "step": 84500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.897449921378274e-06,
+      "loss": 6.2711,
+      "step": 85000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.896766254187462e-06,
+      "loss": 6.2709,
+      "step": 85500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.89608258699665e-06,
+      "loss": 6.2729,
+      "step": 86000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.89539891980584e-06,
+      "loss": 6.269,
+      "step": 86500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.894715252615028e-06,
+      "loss": 6.2654,
+      "step": 87000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.894031585424216e-06,
+      "loss": 6.2653,
+      "step": 87500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.893347918233404e-06,
+      "loss": 6.2639,
+      "step": 88000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.892664251042592e-06,
+      "loss": 6.2585,
+      "step": 88500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.89198058385178e-06,
+      "loss": 6.2631,
+      "step": 89000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.89129691666097e-06,
+      "loss": 6.2615,
+      "step": 89500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.890613249470158e-06,
+      "loss": 6.2572,
+      "step": 90000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.889929582279348e-06,
+      "loss": 6.2558,
+      "step": 90500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.889245915088536e-06,
+      "loss": 6.2582,
+      "step": 91000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.888562247897724e-06,
+      "loss": 6.2517,
+      "step": 91500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.887878580706912e-06,
+      "loss": 6.253,
+      "step": 92000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.887194913516102e-06,
+      "loss": 6.2554,
+      "step": 92500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.88651124632529e-06,
+      "loss": 6.2512,
+      "step": 93000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.885827579134478e-06,
+      "loss": 6.2525,
+      "step": 93500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.885143911943666e-06,
+      "loss": 6.2551,
+      "step": 94000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.884460244752854e-06,
+      "loss": 6.2537,
+      "step": 94500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.883776577562044e-06,
+      "loss": 6.2463,
+      "step": 95000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.883092910371232e-06,
+      "loss": 6.2454,
+      "step": 95500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.882409243180422e-06,
+      "loss": 6.2527,
+      "step": 96000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.88172557598961e-06,
+      "loss": 6.2468,
+      "step": 96500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.881041908798798e-06,
+      "loss": 6.2393,
+      "step": 97000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.880358241607986e-06,
+      "loss": 6.2451,
+      "step": 97500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.879674574417174e-06,
+      "loss": 6.2376,
+      "step": 98000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.878990907226364e-06,
+      "loss": 6.2408,
+      "step": 98500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.878307240035552e-06,
+      "loss": 6.2449,
+      "step": 99000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.87762357284474e-06,
+      "loss": 6.2411,
+      "step": 99500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.876939905653928e-06,
+      "loss": 6.237,
+      "step": 100000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.876256238463117e-06,
+      "loss": 6.2297,
+      "step": 100500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.875572571272305e-06,
+      "loss": 6.2332,
+      "step": 101000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.874888904081494e-06,
+      "loss": 6.2328,
+      "step": 101500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.874205236890683e-06,
+      "loss": 6.2376,
+      "step": 102000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.873521569699871e-06,
+      "loss": 6.2347,
+      "step": 102500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.87283790250906e-06,
+      "loss": 6.2292,
+      "step": 103000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.872154235318247e-06,
+      "loss": 6.2265,
+      "step": 103500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.871470568127435e-06,
+      "loss": 6.2294,
+      "step": 104000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.870786900936625e-06,
+      "loss": 6.2262,
+      "step": 104500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.870103233745813e-06,
+      "loss": 6.2273,
+      "step": 105000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.869419566555001e-06,
+      "loss": 6.2303,
+      "step": 105500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.86873589936419e-06,
+      "loss": 6.2277,
+      "step": 106000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.868052232173379e-06,
+      "loss": 6.2268,
+      "step": 106500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.867368564982567e-06,
+      "loss": 6.2238,
+      "step": 107000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.866684897791757e-06,
+      "loss": 6.2216,
+      "step": 107500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.866001230600945e-06,
+      "loss": 6.221,
+      "step": 108000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.865317563410133e-06,
+      "loss": 6.2186,
+      "step": 108500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.864633896219321e-06,
+      "loss": 6.2208,
+      "step": 109000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.863950229028509e-06,
+      "loss": 6.2109,
+      "step": 109500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.863266561837697e-06,
+      "loss": 6.2177,
+      "step": 110000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.862582894646887e-06,
+      "loss": 6.2115,
+      "step": 110500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.861899227456075e-06,
+      "loss": 6.2165,
+      "step": 111000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.861215560265263e-06,
+      "loss": 6.2114,
+      "step": 111500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.860531893074453e-06,
+      "loss": 6.2107,
+      "step": 112000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.85984822588364e-06,
+      "loss": 6.2133,
+      "step": 112500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.859164558692829e-06,
+      "loss": 6.2183,
+      "step": 113000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.858480891502019e-06,
+      "loss": 6.2109,
+      "step": 113500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857797224311207e-06,
+      "loss": 6.2172,
+      "step": 114000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857113557120395e-06,
+      "loss": 6.215,
+      "step": 114500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.856429889929583e-06,
+      "loss": 6.2081,
+      "step": 115000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.85574622273877e-06,
+      "loss": 6.2102,
+      "step": 115500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.855062555547959e-06,
+      "loss": 6.2085,
+      "step": 116000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.854378888357149e-06,
+      "loss": 6.2131,
+      "step": 116500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.853695221166337e-06,
+      "loss": 6.2129,
+      "step": 117000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.853011553975526e-06,
+      "loss": 6.2095,
+      "step": 117500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.852327886784714e-06,
+      "loss": 6.2064,
+      "step": 118000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.851644219593902e-06,
+      "loss": 6.206,
+      "step": 118500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.85096055240309e-06,
+      "loss": 6.2033,
+      "step": 119000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.85027688521228e-06,
+      "loss": 6.1988,
+      "step": 119500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.849593218021468e-06,
+      "loss": 6.2086,
+      "step": 120000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.848909550830656e-06,
+      "loss": 6.1993,
+      "step": 120500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.848225883639844e-06,
+      "loss": 6.2061,
+      "step": 121000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.847542216449032e-06,
+      "loss": 6.2018,
+      "step": 121500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.846858549258222e-06,
+      "loss": 6.1964,
+      "step": 122000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.84617488206741e-06,
+      "loss": 6.199,
+      "step": 122500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.845491214876598e-06,
+      "loss": 6.1959,
+      "step": 123000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.844807547685788e-06,
+      "loss": 6.2011,
+      "step": 123500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.844123880494976e-06,
+      "loss": 6.1905,
+      "step": 124000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.843440213304164e-06,
+      "loss": 6.1956,
+      "step": 124500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.842756546113352e-06,
+      "loss": 6.1905,
+      "step": 125000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.842072878922542e-06,
+      "loss": 6.1932,
+      "step": 125500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.84138921173173e-06,
+      "loss": 6.1909,
+      "step": 126000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.840705544540918e-06,
+      "loss": 6.1912,
+      "step": 126500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.840021877350106e-06,
+      "loss": 6.1927,
+      "step": 127000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.839338210159294e-06,
+      "loss": 6.1935,
+      "step": 127500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.838654542968484e-06,
+      "loss": 6.1897,
+      "step": 128000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.837970875777672e-06,
+      "loss": 6.1891,
+      "step": 128500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.837287208586862e-06,
+      "loss": 6.1853,
+      "step": 129000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.83660354139605e-06,
+      "loss": 6.1915,
+      "step": 129500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.835919874205238e-06,
+      "loss": 6.1883,
+      "step": 130000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.835236207014426e-06,
+      "loss": 6.1819,
+      "step": 130500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.834552539823614e-06,
+      "loss": 6.1865,
+      "step": 131000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.833868872632804e-06,
+      "loss": 6.1807,
+      "step": 131500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.833185205441992e-06,
+      "loss": 6.1823,
+      "step": 132000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.83250153825118e-06,
+      "loss": 6.1849,
+      "step": 132500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.831817871060368e-06,
+      "loss": 6.1861,
+      "step": 133000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.831134203869558e-06,
+      "loss": 6.1847,
+      "step": 133500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.830450536678746e-06,
+      "loss": 6.1789,
+      "step": 134000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.829766869487934e-06,
+      "loss": 6.1845,
+      "step": 134500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.829083202297123e-06,
+      "loss": 6.182,
+      "step": 135000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.828399535106311e-06,
+      "loss": 6.1806,
+      "step": 135500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.8277158679155e-06,
+      "loss": 6.1822,
+      "step": 136000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.827032200724688e-06,
+      "loss": 6.184,
+      "step": 136500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.826348533533876e-06,
+      "loss": 6.1804,
+      "step": 137000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.825664866343065e-06,
+      "loss": 6.1794,
+      "step": 137500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.824981199152253e-06,
+      "loss": 6.1784,
+      "step": 138000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.824297531961441e-06,
+      "loss": 6.1743,
+      "step": 138500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.82361386477063e-06,
+      "loss": 6.1807,
+      "step": 139000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.82293019757982e-06,
+      "loss": 6.1765,
+      "step": 139500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.822246530389007e-06,
+      "loss": 6.1731,
+      "step": 140000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.821562863198197e-06,
+      "loss": 6.1708,
+      "step": 140500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.820879196007385e-06,
+      "loss": 6.1721,
+      "step": 141000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.820195528816573e-06,
+      "loss": 6.172,
+      "step": 141500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.819511861625761e-06,
+      "loss": 6.1726,
+      "step": 142000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.81882819443495e-06,
+      "loss": 6.1738,
+      "step": 142500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.818144527244139e-06,
+      "loss": 6.1701,
+      "step": 143000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.817460860053327e-06,
+      "loss": 6.1697,
+      "step": 143500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.816777192862515e-06,
+      "loss": 6.1713,
+      "step": 144000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.816093525671703e-06,
+      "loss": 6.1662,
+      "step": 144500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.815409858480893e-06,
+      "loss": 6.1705,
+      "step": 145000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.814726191290081e-06,
+      "loss": 6.1671,
+      "step": 145500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.81404252409927e-06,
+      "loss": 6.1678,
+      "step": 146000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.813358856908459e-06,
+      "loss": 6.1646,
+      "step": 146500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.812675189717647e-06,
+      "loss": 6.17,
+      "step": 147000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.811991522526835e-06,
+      "loss": 6.1644,
+      "step": 147500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.811307855336023e-06,
+      "loss": 6.1674,
+      "step": 148000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.810624188145211e-06,
+      "loss": 6.1578,
+      "step": 148500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.8099405209544e-06,
+      "loss": 6.1664,
+      "step": 149000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.809256853763589e-06,
+      "loss": 6.1652,
+      "step": 149500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.808573186572777e-06,
+      "loss": 6.1624,
+      "step": 150000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.807889519381966e-06,
+      "loss": 6.1598,
+      "step": 150500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.807205852191155e-06,
+      "loss": 6.1625,
+      "step": 151000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.806522185000343e-06,
+      "loss": 6.1598,
+      "step": 151500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.805838517809532e-06,
+      "loss": 6.1608,
+      "step": 152000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.80515485061872e-06,
+      "loss": 6.1576,
+      "step": 152500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.804471183427908e-06,
+      "loss": 6.1616,
+      "step": 153000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.803787516237096e-06,
+      "loss": 6.1591,
+      "step": 153500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.803103849046285e-06,
+      "loss": 6.1598,
+      "step": 154000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.802420181855473e-06,
+      "loss": 6.1656,
+      "step": 154500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.801736514664662e-06,
+      "loss": 6.1549,
+      "step": 155000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.80105284747385e-06,
+      "loss": 6.1552,
+      "step": 155500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.800369180283038e-06,
+      "loss": 6.1547,
+      "step": 156000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.799685513092228e-06,
+      "loss": 6.1577,
+      "step": 156500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.799001845901416e-06,
+      "loss": 6.1584,
+      "step": 157000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.798318178710604e-06,
+      "loss": 6.1573,
+      "step": 157500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.797634511519794e-06,
+      "loss": 6.1541,
+      "step": 158000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.796950844328982e-06,
+      "loss": 6.1598,
+      "step": 158500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.79626717713817e-06,
+      "loss": 6.1528,
+      "step": 159000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.795583509947358e-06,
+      "loss": 6.1547,
+      "step": 159500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.794899842756546e-06,
+      "loss": 6.1561,
+      "step": 160000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.794216175565734e-06,
+      "loss": 6.1586,
+      "step": 160500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.793532508374924e-06,
+      "loss": 6.1501,
+      "step": 161000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.792848841184112e-06,
+      "loss": 6.1507,
+      "step": 161500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.792165173993302e-06,
+      "loss": 6.1529,
+      "step": 162000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.79148150680249e-06,
+      "loss": 6.1533,
+      "step": 162500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.790797839611678e-06,
+      "loss": 6.1545,
+      "step": 163000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.790114172420866e-06,
+      "loss": 6.1501,
+      "step": 163500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.789430505230056e-06,
+      "loss": 6.1505,
+      "step": 164000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.788746838039244e-06,
+      "loss": 6.1462,
+      "step": 164500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.788063170848432e-06,
+      "loss": 6.1444,
+      "step": 165000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.78737950365762e-06,
+      "loss": 6.1509,
+      "step": 165500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.786695836466808e-06,
+      "loss": 6.1466,
+      "step": 166000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.786012169275998e-06,
+      "loss": 6.1484,
+      "step": 166500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.785328502085186e-06,
+      "loss": 6.1529,
+      "step": 167000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.784644834894374e-06,
+      "loss": 6.1462,
+      "step": 167500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.783961167703563e-06,
+      "loss": 6.1459,
+      "step": 168000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.783277500512752e-06,
+      "loss": 6.1467,
+      "step": 168500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.78259383332194e-06,
+      "loss": 6.149,
+      "step": 169000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.781910166131128e-06,
+      "loss": 6.1467,
+      "step": 169500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.781226498940317e-06,
+      "loss": 6.1408,
+      "step": 170000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.780542831749505e-06,
+      "loss": 6.1432,
+      "step": 170500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.779859164558693e-06,
+      "loss": 6.1451,
+      "step": 171000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.779175497367882e-06,
+      "loss": 6.1524,
+      "step": 171500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.77849183017707e-06,
+      "loss": 6.1431,
+      "step": 172000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.77780816298626e-06,
+      "loss": 6.1414,
+      "step": 172500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.777124495795447e-06,
+      "loss": 6.1377,
+      "step": 173000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.776440828604637e-06,
+      "loss": 6.1438,
+      "step": 173500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.775757161413825e-06,
+      "loss": 6.1418,
+      "step": 174000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.775073494223013e-06,
+      "loss": 6.1406,
+      "step": 174500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.774389827032201e-06,
+      "loss": 6.1402,
+      "step": 175000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.77370615984139e-06,
+      "loss": 6.1407,
+      "step": 175500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.773022492650579e-06,
+      "loss": 6.141,
+      "step": 176000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.772338825459767e-06,
+      "loss": 6.1391,
+      "step": 176500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.771655158268955e-06,
+      "loss": 6.1411,
+      "step": 177000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.770971491078143e-06,
+      "loss": 6.1354,
+      "step": 177500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.770287823887333e-06,
+      "loss": 6.138,
+      "step": 178000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.769604156696521e-06,
+      "loss": 6.1387,
+      "step": 178500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.76892048950571e-06,
+      "loss": 6.1369,
+      "step": 179000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.768236822314899e-06,
+      "loss": 6.1326,
+      "step": 179500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.767553155124087e-06,
+      "loss": 6.1391,
+      "step": 180000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.766869487933275e-06,
+      "loss": 6.1337,
+      "step": 180500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.766185820742463e-06,
+      "loss": 6.1349,
+      "step": 181000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.765502153551651e-06,
+      "loss": 6.1385,
+      "step": 181500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.76481848636084e-06,
+      "loss": 6.1322,
+      "step": 182000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.764134819170029e-06,
+      "loss": 6.1335,
+      "step": 182500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.763451151979217e-06,
+      "loss": 6.1342,
+      "step": 183000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.762767484788407e-06,
+      "loss": 6.1329,
+      "step": 183500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.762083817597595e-06,
+      "loss": 6.1343,
+      "step": 184000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.761400150406783e-06,
+      "loss": 6.1257,
+      "step": 184500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.760716483215972e-06,
+      "loss": 6.1288,
+      "step": 185000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.76003281602516e-06,
+      "loss": 6.1368,
+      "step": 185500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.759349148834348e-06,
+      "loss": 6.1301,
+      "step": 186000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.758665481643537e-06,
+      "loss": 6.1268,
+      "step": 186500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.757981814452725e-06,
+      "loss": 6.1276,
+      "step": 187000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.757298147261913e-06,
+      "loss": 6.1277,
+      "step": 187500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.756614480071102e-06,
+      "loss": 6.1312,
+      "step": 188000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.75593081288029e-06,
+      "loss": 6.127,
+      "step": 188500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.755247145689478e-06,
+      "loss": 6.1286,
+      "step": 189000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.754563478498668e-06,
+      "loss": 6.1267,
+      "step": 189500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.753879811307856e-06,
+      "loss": 6.1265,
+      "step": 190000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.753196144117044e-06,
+      "loss": 6.1258,
+      "step": 190500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.752512476926234e-06,
+      "loss": 6.1288,
+      "step": 191000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.751828809735422e-06,
+      "loss": 6.1261,
+      "step": 191500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.75114514254461e-06,
+      "loss": 6.1274,
+      "step": 192000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.750461475353798e-06,
+      "loss": 6.1324,
+      "step": 192500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.749777808162986e-06,
+      "loss": 6.1263,
+      "step": 193000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.749094140972174e-06,
+      "loss": 6.1187,
+      "step": 193500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.748410473781364e-06,
+      "loss": 6.1265,
+      "step": 194000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.747726806590552e-06,
+      "loss": 6.1197,
+      "step": 194500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.747043139399742e-06,
+      "loss": 6.1204,
+      "step": 195000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.74635947220893e-06,
+      "loss": 6.126,
+      "step": 195500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.745675805018118e-06,
+      "loss": 6.1209,
+      "step": 196000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.744992137827306e-06,
+      "loss": 6.1242,
+      "step": 196500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.744308470636496e-06,
+      "loss": 6.1244,
+      "step": 197000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.743624803445684e-06,
+      "loss": 6.1216,
+      "step": 197500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.742941136254872e-06,
+      "loss": 6.1252,
+      "step": 198000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.74225746906406e-06,
+      "loss": 6.1214,
+      "step": 198500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.741573801873248e-06,
+      "loss": 6.1185,
+      "step": 199000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.740890134682438e-06,
+      "loss": 6.1259,
+      "step": 199500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.740206467491626e-06,
+      "loss": 6.1213,
+      "step": 200000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.739522800300815e-06,
+      "loss": 6.1194,
+      "step": 200500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.738839133110004e-06,
+      "loss": 6.1177,
+      "step": 201000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.738155465919192e-06,
+      "loss": 6.1177,
+      "step": 201500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.73747179872838e-06,
+      "loss": 6.1214,
+      "step": 202000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.736788131537568e-06,
+      "loss": 6.1135,
+      "step": 202500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.736104464346757e-06,
+      "loss": 6.1212,
+      "step": 203000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.735420797155945e-06,
+      "loss": 6.1192,
+      "step": 203500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.734737129965134e-06,
+      "loss": 6.1138,
+      "step": 204000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.734053462774322e-06,
+      "loss": 6.1204,
+      "step": 204500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.73336979558351e-06,
+      "loss": 6.1155,
+      "step": 205000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.7326861283927e-06,
+      "loss": 6.1196,
+      "step": 205500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.732002461201887e-06,
+      "loss": 6.1127,
+      "step": 206000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.731318794011077e-06,
+      "loss": 6.1179,
+      "step": 206500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.730635126820265e-06,
+      "loss": 6.1203,
+      "step": 207000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.729951459629453e-06,
+      "loss": 6.1181,
+      "step": 207500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.729267792438641e-06,
+      "loss": 6.1109,
+      "step": 208000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.72858412524783e-06,
+      "loss": 6.1182,
+      "step": 208500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.727900458057019e-06,
+      "loss": 6.1101,
+      "step": 209000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.727216790866207e-06,
+      "loss": 6.1131,
+      "step": 209500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.726533123675395e-06,
+      "loss": 6.1109,
+      "step": 210000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.725849456484583e-06,
+      "loss": 6.1165,
+      "step": 210500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.725165789293773e-06,
+      "loss": 6.1062,
+      "step": 211000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.724482122102961e-06,
+      "loss": 6.1156,
+      "step": 211500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.72379845491215e-06,
+      "loss": 6.1096,
+      "step": 212000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.723114787721339e-06,
+      "loss": 6.1109,
+      "step": 212500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.722431120530527e-06,
+      "loss": 6.1096,
+      "step": 213000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.721747453339715e-06,
+      "loss": 6.1071,
+      "step": 213500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.721063786148903e-06,
+      "loss": 6.1112,
+      "step": 214000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.720380118958091e-06,
+      "loss": 6.1148,
+      "step": 214500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.71969645176728e-06,
+      "loss": 6.108,
+      "step": 215000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.719012784576469e-06,
+      "loss": 6.111,
+      "step": 215500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.718329117385657e-06,
+      "loss": 6.112,
+      "step": 216000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.717645450194847e-06,
+      "loss": 6.1068,
+      "step": 216500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.716961783004035e-06,
+      "loss": 6.1075,
+      "step": 217000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.716278115813223e-06,
+      "loss": 6.1082,
+      "step": 217500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.715594448622412e-06,
+      "loss": 6.1066,
+      "step": 218000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.7149107814316e-06,
+      "loss": 6.1072,
+      "step": 218500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.714227114240789e-06,
+      "loss": 6.109,
+      "step": 219000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.713543447049977e-06,
+      "loss": 6.1036,
+      "step": 219500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.712859779859165e-06,
+      "loss": 6.1048,
+      "step": 220000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.712176112668353e-06,
+      "loss": 6.1113,
+      "step": 220500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.711492445477542e-06,
+      "loss": 6.1052,
+      "step": 221000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.71080877828673e-06,
+      "loss": 6.1086,
+      "step": 221500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.710125111095919e-06,
+      "loss": 6.1089,
+      "step": 222000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.709441443905108e-06,
+      "loss": 6.1054,
+      "step": 222500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.708757776714296e-06,
+      "loss": 6.1095,
+      "step": 223000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.708074109523484e-06,
+      "loss": 6.1041,
+      "step": 223500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.707390442332674e-06,
+      "loss": 6.1037,
+      "step": 224000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.706706775141862e-06,
+      "loss": 6.1069,
+      "step": 224500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.70602310795105e-06,
+      "loss": 6.1093,
+      "step": 225000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.705339440760238e-06,
+      "loss": 6.1048,
+      "step": 225500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.704655773569426e-06,
+      "loss": 6.1024,
+      "step": 226000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.703972106378614e-06,
+      "loss": 6.0995,
+      "step": 226500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.703288439187804e-06,
+      "loss": 6.1057,
+      "step": 227000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.702604771996992e-06,
+      "loss": 6.1034,
+      "step": 227500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.701921104806182e-06,
+      "loss": 6.1044,
+      "step": 228000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.70123743761537e-06,
+      "loss": 6.108,
+      "step": 228500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.700553770424558e-06,
+      "loss": 6.1026,
+      "step": 229000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.699870103233746e-06,
+      "loss": 6.0995,
+      "step": 229500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.699186436042936e-06,
+      "loss": 6.0923,
+      "step": 230000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.698502768852124e-06,
+      "loss": 6.1001,
+      "step": 230500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.697819101661312e-06,
+      "loss": 6.0996,
+      "step": 231000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.6971354344705e-06,
+      "loss": 6.1032,
+      "step": 231500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.696451767279688e-06,
+      "loss": 6.1033,
+      "step": 232000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.695768100088878e-06,
+      "loss": 6.1012,
+      "step": 232500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.695084432898066e-06,
+      "loss": 6.1,
+      "step": 233000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.694400765707256e-06,
+      "loss": 6.1017,
+      "step": 233500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.693717098516444e-06,
+      "loss": 6.1001,
+      "step": 234000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.693033431325632e-06,
+      "loss": 6.1023,
+      "step": 234500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.69234976413482e-06,
+      "loss": 6.0977,
+      "step": 235000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.691666096944008e-06,
+      "loss": 6.0986,
+      "step": 235500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.690982429753198e-06,
+      "loss": 6.0963,
+      "step": 236000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.690298762562386e-06,
+      "loss": 6.0936,
+      "step": 236500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.689615095371574e-06,
+      "loss": 6.0959,
+      "step": 237000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.688931428180762e-06,
+      "loss": 6.1009,
+      "step": 237500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.68824776098995e-06,
+      "loss": 6.0983,
+      "step": 238000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.68756409379914e-06,
+      "loss": 6.1014,
+      "step": 238500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.686880426608328e-06,
+      "loss": 6.0965,
+      "step": 239000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.686196759417517e-06,
+      "loss": 6.0927,
+      "step": 239500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.685513092226705e-06,
+      "loss": 6.0954,
+      "step": 240000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.684829425035893e-06,
+      "loss": 6.094,
+      "step": 240500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.684145757845081e-06,
+      "loss": 6.0949,
+      "step": 241000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.68346209065427e-06,
+      "loss": 6.0968,
+      "step": 241500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.68277842346346e-06,
+      "loss": 6.0932,
+      "step": 242000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.682094756272647e-06,
+      "loss": 6.0924,
+      "step": 242500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.681411089081835e-06,
+      "loss": 6.0952,
+      "step": 243000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.680727421891023e-06,
+      "loss": 6.09,
+      "step": 243500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.680043754700213e-06,
+      "loss": 6.0955,
+      "step": 244000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.679360087509401e-06,
+      "loss": 6.0963,
+      "step": 244500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.678676420318591e-06,
+      "loss": 6.0918,
+      "step": 245000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.677992753127779e-06,
+      "loss": 6.0874,
+      "step": 245500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 9.677309085936967e-06,
+      "loss": 6.0935,
+      "step": 246000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.676625418746155e-06,
+      "loss": 6.0881,
+      "step": 246500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.675941751555343e-06,
+      "loss": 6.0926,
+      "step": 247000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.675258084364531e-06,
+      "loss": 6.0888,
+      "step": 247500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.674574417173721e-06,
+      "loss": 6.0866,
+      "step": 248000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.673890749982909e-06,
+      "loss": 6.0931,
+      "step": 248500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.673207082792097e-06,
+      "loss": 6.0953,
+      "step": 249000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.672523415601287e-06,
+      "loss": 6.0936,
+      "step": 249500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.671839748410475e-06,
+      "loss": 6.0871,
+      "step": 250000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.671156081219663e-06,
+      "loss": 6.0945,
+      "step": 250500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.670472414028853e-06,
+      "loss": 6.0857,
+      "step": 251000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.66978874683804e-06,
+      "loss": 6.0894,
+      "step": 251500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.669105079647229e-06,
+      "loss": 6.0888,
+      "step": 252000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.668421412456417e-06,
+      "loss": 6.09,
+      "step": 252500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.667737745265605e-06,
+      "loss": 6.0959,
+      "step": 253000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.667054078074793e-06,
+      "loss": 6.0866,
+      "step": 253500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.666370410883983e-06,
+      "loss": 6.0897,
+      "step": 254000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.66568674369317e-06,
+      "loss": 6.0908,
+      "step": 254500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.665003076502359e-06,
+      "loss": 6.0908,
+      "step": 255000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.664319409311548e-06,
+      "loss": 6.0908,
+      "step": 255500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.663635742120736e-06,
+      "loss": 6.0847,
+      "step": 256000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.662952074929925e-06,
+      "loss": 6.0901,
+      "step": 256500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 9.662268407739114e-06,
+      "loss": 6.0849,
+      "step": 257000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.661584740548302e-06,
+      "loss": 6.0864,
+      "step": 257500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.66090107335749e-06,
+      "loss": 6.0904,
+      "step": 258000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 9.660217406166678e-06,
+      "loss": 6.0884,
+      "step": 258500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 9.659533738975866e-06,
+      "loss": 6.0862,
+      "step": 259000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 9.658850071785054e-06,
+      "loss": 6.0874,
+      "step": 259500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 9.658166404594244e-06,
+      "loss": 6.0836,
+      "step": 260000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.657482737403432e-06,
+      "loss": 6.0844,
+      "step": 260500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.656799070212622e-06,
+      "loss": 6.091,
+      "step": 261000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.65611540302181e-06,
+      "loss": 6.0807,
+      "step": 261500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.655431735830998e-06,
+      "loss": 6.0856,
+      "step": 262000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.654748068640186e-06,
+      "loss": 6.0799,
+      "step": 262500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.654064401449376e-06,
+      "loss": 6.0813,
+      "step": 263000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.653380734258564e-06,
+      "loss": 6.0834,
+      "step": 263500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.652697067067752e-06,
+      "loss": 6.0872,
+      "step": 264000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.65201339987694e-06,
+      "loss": 6.083,
+      "step": 264500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.651329732686128e-06,
+      "loss": 6.0825,
+      "step": 265000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.650646065495318e-06,
+      "loss": 6.0798,
+      "step": 265500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.649962398304506e-06,
+      "loss": 6.0879,
+      "step": 266000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.649278731113696e-06,
+      "loss": 6.08,
+      "step": 266500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.648595063922884e-06,
+      "loss": 6.0817,
+      "step": 267000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.647911396732072e-06,
+      "loss": 6.0861,
+      "step": 267500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.64722772954126e-06,
+      "loss": 6.078,
+      "step": 268000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.646544062350448e-06,
+      "loss": 6.0812,
+      "step": 268500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.645860395159638e-06,
+      "loss": 6.081,
+      "step": 269000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.645176727968826e-06,
+      "loss": 6.0811,
+      "step": 269500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.644493060778014e-06,
+      "loss": 6.0784,
+      "step": 270000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.643809393587202e-06,
+      "loss": 6.0779,
+      "step": 270500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.643125726396392e-06,
+      "loss": 6.083,
+      "step": 271000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.64244205920558e-06,
+      "loss": 6.0789,
+      "step": 271500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.641758392014768e-06,
+      "loss": 6.0811,
+      "step": 272000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.641074724823957e-06,
+      "loss": 6.0792,
+      "step": 272500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.640391057633145e-06,
+      "loss": 6.0835,
+      "step": 273000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.639707390442333e-06,
+      "loss": 6.0799,
+      "step": 273500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.639023723251521e-06,
+      "loss": 6.0788,
+      "step": 274000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.63834005606071e-06,
+      "loss": 6.0803,
+      "step": 274500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.6376563888699e-06,
+      "loss": 6.0784,
+      "step": 275000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.636972721679087e-06,
+      "loss": 6.0784,
+      "step": 275500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.636289054488275e-06,
+      "loss": 6.0748,
+      "step": 276000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.635605387297463e-06,
+      "loss": 6.0756,
+      "step": 276500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.634921720106653e-06,
+      "loss": 6.0814,
+      "step": 277000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.634238052915841e-06,
+      "loss": 6.0818,
+      "step": 277500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.633554385725031e-06,
+      "loss": 6.0709,
+      "step": 278000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.632870718534219e-06,
+      "loss": 6.0846,
+      "step": 278500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.632187051343407e-06,
+      "loss": 6.0778,
+      "step": 279000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.631503384152595e-06,
+      "loss": 6.0812,
+      "step": 279500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.630819716961783e-06,
+      "loss": 6.0739,
+      "step": 280000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.630136049770973e-06,
+      "loss": 6.0701,
+      "step": 280500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.629452382580161e-06,
+      "loss": 6.0742,
+      "step": 281000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.628768715389349e-06,
+      "loss": 6.0691,
+      "step": 281500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 9.628085048198537e-06,
+      "loss": 6.0792,
+      "step": 282000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.627401381007727e-06,
+      "loss": 6.0802,
+      "step": 282500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.626717713816915e-06,
+      "loss": 6.0787,
+      "step": 283000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 9.626034046626103e-06,
+      "loss": 6.0802,
+      "step": 283500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.625350379435293e-06,
+      "loss": 6.0739,
+      "step": 284000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.62466671224448e-06,
+      "loss": 6.0721,
+      "step": 284500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.623983045053669e-06,
+      "loss": 6.0758,
+      "step": 285000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 9.623299377862857e-06,
+      "loss": 6.0732,
+      "step": 285500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.622615710672045e-06,
+      "loss": 6.0721,
+      "step": 286000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.621932043481235e-06,
+      "loss": 6.0692,
+      "step": 286500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.621248376290423e-06,
+      "loss": 6.0702,
+      "step": 287000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.62056470909961e-06,
+      "loss": 6.075,
+      "step": 287500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.619881041908799e-06,
+      "loss": 6.0736,
+      "step": 288000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.619197374717988e-06,
+      "loss": 6.0741,
+      "step": 288500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.618513707527177e-06,
+      "loss": 6.0738,
+      "step": 289000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.617830040336366e-06,
+      "loss": 6.0758,
+      "step": 289500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.617146373145554e-06,
+      "loss": 6.0714,
+      "step": 290000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.616462705954742e-06,
+      "loss": 6.0755,
+      "step": 290500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.61577903876393e-06,
+      "loss": 6.0724,
+      "step": 291000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 9.615095371573118e-06,
+      "loss": 6.0708,
+      "step": 291500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.614411704382307e-06,
+      "loss": 6.0708,
+      "step": 292000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.613728037191496e-06,
+      "loss": 6.0685,
+      "step": 292500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.613044370000684e-06,
+      "loss": 6.0691,
+      "step": 293000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.612360702809872e-06,
+      "loss": 6.0705,
+      "step": 293500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.611677035619062e-06,
+      "loss": 6.0749,
+      "step": 294000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.61099336842825e-06,
+      "loss": 6.071,
+      "step": 294500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.610309701237438e-06,
+      "loss": 6.0695,
+      "step": 295000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.609626034046628e-06,
+      "loss": 6.0714,
+      "step": 295500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.608942366855816e-06,
+      "loss": 6.0743,
+      "step": 296000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.608258699665004e-06,
+      "loss": 6.0679,
+      "step": 296500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.607575032474192e-06,
+      "loss": 6.0683,
+      "step": 297000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.60689136528338e-06,
+      "loss": 6.069,
+      "step": 297500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.606207698092568e-06,
+      "loss": 6.0645,
+      "step": 298000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.605524030901758e-06,
+      "loss": 6.0748,
+      "step": 298500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.604840363710946e-06,
+      "loss": 6.0693,
+      "step": 299000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.604156696520136e-06,
+      "loss": 6.068,
+      "step": 299500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.603473029329324e-06,
+      "loss": 6.0703,
+      "step": 300000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.602789362138512e-06,
+      "loss": 6.0722,
+      "step": 300500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.6021056949477e-06,
+      "loss": 6.0655,
+      "step": 301000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.60142202775689e-06,
+      "loss": 6.0671,
+      "step": 301500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.600738360566078e-06,
+      "loss": 6.0639,
+      "step": 302000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.600054693375266e-06,
+      "loss": 6.0632,
+      "step": 302500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.599371026184454e-06,
+      "loss": 6.0655,
+      "step": 303000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.598687358993642e-06,
+      "loss": 6.0682,
+      "step": 303500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.598003691802832e-06,
+      "loss": 6.065,
+      "step": 304000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.59732002461202e-06,
+      "loss": 6.0681,
+      "step": 304500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.596636357421208e-06,
+      "loss": 6.0638,
+      "step": 305000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.595952690230397e-06,
+      "loss": 6.0653,
+      "step": 305500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.595269023039585e-06,
+      "loss": 6.0665,
+      "step": 306000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.594585355848774e-06,
+      "loss": 6.0656,
+      "step": 306500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.593901688657962e-06,
+      "loss": 6.0591,
+      "step": 307000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.593218021467151e-06,
+      "loss": 6.0651,
+      "step": 307500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.59253435427634e-06,
+      "loss": 6.0638,
+      "step": 308000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.591850687085527e-06,
+      "loss": 6.0629,
+      "step": 308500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 9.591167019894715e-06,
+      "loss": 6.0668,
+      "step": 309000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.590483352703904e-06,
+      "loss": 6.0663,
+      "step": 309500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.589799685513093e-06,
+      "loss": 6.0667,
+      "step": 310000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 9.589116018322281e-06,
+      "loss": 6.0629,
+      "step": 310500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.588432351131471e-06,
+      "loss": 6.0659,
+      "step": 311000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.587748683940659e-06,
+      "loss": 6.0634,
+      "step": 311500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.587065016749847e-06,
+      "loss": 6.0652,
+      "step": 312000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.586381349559035e-06,
+      "loss": 6.0589,
+      "step": 312500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.585697682368223e-06,
+      "loss": 6.0595,
+      "step": 313000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.585014015177413e-06,
+      "loss": 6.0633,
+      "step": 313500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.584330347986601e-06,
+      "loss": 6.066,
+      "step": 314000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.583646680795789e-06,
+      "loss": 6.0645,
+      "step": 314500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.582963013604977e-06,
+      "loss": 6.0598,
+      "step": 315000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.582279346414167e-06,
+      "loss": 6.0633,
+      "step": 315500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.581595679223355e-06,
+      "loss": 6.0649,
+      "step": 316000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.580912012032545e-06,
+      "loss": 6.0654,
+      "step": 316500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.580228344841733e-06,
+      "loss": 6.0653,
+      "step": 317000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.57954467765092e-06,
+      "loss": 6.058,
+      "step": 317500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.578861010460109e-06,
+      "loss": 6.0665,
+      "step": 318000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.578177343269297e-06,
+      "loss": 6.0592,
+      "step": 318500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.577493676078485e-06,
+      "loss": 6.0609,
+      "step": 319000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.576810008887675e-06,
+      "loss": 6.0607,
+      "step": 319500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 9.576126341696863e-06,
+      "loss": 6.0587,
+      "step": 320000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.57544267450605e-06,
+      "loss": 6.0587,
+      "step": 320500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.574759007315239e-06,
+      "loss": 6.0585,
+      "step": 321000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 9.574075340124429e-06,
+      "loss": 6.0533,
+      "step": 321500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 9.573391672933617e-06,
+      "loss": 6.0601,
+      "step": 322000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 9.572708005742806e-06,
+      "loss": 6.0638,
+      "step": 322500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 9.572024338551994e-06,
+      "loss": 6.0557,
+      "step": 323000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.571340671361182e-06,
+      "loss": 6.0669,
+      "step": 323500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.57065700417037e-06,
+      "loss": 6.0586,
+      "step": 324000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.569973336979559e-06,
+      "loss": 6.0593,
+      "step": 324500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.569289669788747e-06,
+      "loss": 6.0569,
+      "step": 325000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.568606002597936e-06,
+      "loss": 6.0574,
+      "step": 325500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.567922335407124e-06,
+      "loss": 6.0604,
+      "step": 326000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 9.567238668216312e-06,
+      "loss": 6.0538,
+      "step": 326500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.566555001025502e-06,
+      "loss": 6.0568,
+      "step": 327000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.56587133383469e-06,
+      "loss": 6.0601,
+      "step": 327500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 9.565187666643878e-06,
+      "loss": 6.0636,
+      "step": 328000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 9.564503999453068e-06,
+      "loss": 6.058,
+      "step": 328500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 9.563820332262256e-06,
+      "loss": 6.0568,
+      "step": 329000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 9.563136665071444e-06,
+      "loss": 6.0573,
+      "step": 329500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.562452997880632e-06,
+      "loss": 6.0627,
+      "step": 330000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.56176933068982e-06,
+      "loss": 6.0548,
+      "step": 330500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.561085663499008e-06,
+      "loss": 6.0527,
+      "step": 331000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.560401996308198e-06,
+      "loss": 6.0531,
+      "step": 331500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.559718329117386e-06,
+      "loss": 6.0599,
+      "step": 332000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.559034661926576e-06,
+      "loss": 6.0582,
+      "step": 332500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.558350994735764e-06,
+      "loss": 6.0548,
+      "step": 333000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.557667327544952e-06,
+      "loss": 6.0551,
+      "step": 333500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.55698366035414e-06,
+      "loss": 6.0473,
+      "step": 334000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.55629999316333e-06,
+      "loss": 6.0566,
+      "step": 334500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.555616325972518e-06,
+      "loss": 6.0519,
+      "step": 335000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.554932658781706e-06,
+      "loss": 6.0575,
+      "step": 335500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.554248991590894e-06,
+      "loss": 6.0591,
+      "step": 336000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.553565324400082e-06,
+      "loss": 6.0592,
+      "step": 336500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.552881657209272e-06,
+      "loss": 6.0621,
+      "step": 337000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.55219799001846e-06,
+      "loss": 6.0551,
+      "step": 337500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.551514322827648e-06,
+      "loss": 6.0546,
+      "step": 338000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.550830655636838e-06,
+      "loss": 6.0537,
+      "step": 338500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.550146988446026e-06,
+      "loss": 6.0548,
+      "step": 339000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.549463321255214e-06,
+      "loss": 6.0529,
+      "step": 339500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.548779654064402e-06,
+      "loss": 6.0548,
+      "step": 340000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.548095986873591e-06,
+      "loss": 6.0576,
+      "step": 340500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.54741231968278e-06,
+      "loss": 6.053,
+      "step": 341000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.546728652491968e-06,
+      "loss": 6.052,
+      "step": 341500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.546044985301156e-06,
+      "loss": 6.0555,
+      "step": 342000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.545361318110344e-06,
+      "loss": 6.0479,
+      "step": 342500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.544677650919533e-06,
+      "loss": 6.0557,
+      "step": 343000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.543993983728721e-06,
+      "loss": 6.049,
+      "step": 343500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.543310316537911e-06,
+      "loss": 6.0452,
+      "step": 344000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.5426266493471e-06,
+      "loss": 6.0541,
+      "step": 344500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.541942982156287e-06,
+      "loss": 6.0577,
+      "step": 345000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.541259314965475e-06,
+      "loss": 6.0508,
+      "step": 345500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.540575647774663e-06,
+      "loss": 6.0551,
+      "step": 346000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.539891980583853e-06,
+      "loss": 6.0478,
+      "step": 346500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 9.539208313393041e-06,
+      "loss": 6.0537,
+      "step": 347000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 9.53852464620223e-06,
+      "loss": 6.05,
+      "step": 347500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.537840979011417e-06,
+      "loss": 6.0481,
+      "step": 348000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.537157311820607e-06,
+      "loss": 6.0521,
+      "step": 348500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.536473644629795e-06,
+      "loss": 6.0517,
+      "step": 349000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.535789977438985e-06,
+      "loss": 6.0463,
+      "step": 349500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.535106310248173e-06,
+      "loss": 6.0554,
+      "step": 350000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.534422643057361e-06,
+      "loss": 6.0482,
+      "step": 350500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.533738975866549e-06,
+      "loss": 6.0511,
+      "step": 351000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 9.533055308675737e-06,
+      "loss": 6.0503,
+      "step": 351500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.532371641484925e-06,
+      "loss": 6.0517,
+      "step": 352000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.531687974294115e-06,
+      "loss": 6.0507,
+      "step": 352500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.531004307103303e-06,
+      "loss": 6.0528,
+      "step": 353000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.530320639912491e-06,
+      "loss": 6.0519,
+      "step": 353500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.529636972721679e-06,
+      "loss": 6.0508,
+      "step": 354000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 9.528953305530869e-06,
+      "loss": 6.0497,
+      "step": 354500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.528269638340057e-06,
+      "loss": 6.049,
+      "step": 355000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.527585971149246e-06,
+      "loss": 6.0546,
+      "step": 355500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.526902303958435e-06,
+      "loss": 6.0413,
+      "step": 356000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.526218636767623e-06,
+      "loss": 6.0467,
+      "step": 356500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.52553496957681e-06,
+      "loss": 6.0459,
+      "step": 357000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.524851302385999e-06,
+      "loss": 6.0457,
+      "step": 357500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.524167635195187e-06,
+      "loss": 6.0421,
+      "step": 358000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.523483968004376e-06,
+      "loss": 6.0503,
+      "step": 358500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.522800300813564e-06,
+      "loss": 6.0461,
+      "step": 359000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.522116633622753e-06,
+      "loss": 6.042,
+      "step": 359500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.521432966431942e-06,
+      "loss": 6.0441,
+      "step": 360000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.52074929924113e-06,
+      "loss": 6.0439,
+      "step": 360500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.520065632050318e-06,
+      "loss": 6.0427,
+      "step": 361000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.519381964859508e-06,
+      "loss": 6.0493,
+      "step": 361500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.518698297668696e-06,
+      "loss": 6.0471,
+      "step": 362000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 9.518014630477884e-06,
+      "loss": 6.0536,
+      "step": 362500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.517330963287072e-06,
+      "loss": 6.045,
+      "step": 363000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.51664729609626e-06,
+      "loss": 6.051,
+      "step": 363500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 9.515963628905448e-06,
+      "loss": 6.0429,
+      "step": 364000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.515279961714638e-06,
+      "loss": 6.0424,
+      "step": 364500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.514596294523826e-06,
+      "loss": 6.0487,
+      "step": 365000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.513912627333016e-06,
+      "loss": 6.0477,
+      "step": 365500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.513228960142204e-06,
+      "loss": 6.0501,
+      "step": 366000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.512545292951392e-06,
+      "loss": 6.0353,
+      "step": 366500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.51186162576058e-06,
+      "loss": 6.0465,
+      "step": 367000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.51117795856977e-06,
+      "loss": 6.0436,
+      "step": 367500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.510494291378958e-06,
+      "loss": 6.0454,
+      "step": 368000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.509810624188146e-06,
+      "loss": 6.0463,
+      "step": 368500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.509126956997334e-06,
+      "loss": 6.0442,
+      "step": 369000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.508443289806522e-06,
+      "loss": 6.0462,
+      "step": 369500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.507759622615712e-06,
+      "loss": 6.0458,
+      "step": 370000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.5070759554249e-06,
+      "loss": 6.0452,
+      "step": 370500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.506392288234088e-06,
+      "loss": 6.0435,
+      "step": 371000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.505708621043278e-06,
+      "loss": 6.0422,
+      "step": 371500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.505024953852466e-06,
+      "loss": 6.0359,
+      "step": 372000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.504341286661654e-06,
+      "loss": 6.0398,
+      "step": 372500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.503657619470842e-06,
+      "loss": 6.044,
+      "step": 373000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.502973952280031e-06,
+      "loss": 6.0423,
+      "step": 373500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.50229028508922e-06,
+      "loss": 6.0404,
+      "step": 374000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.501606617898408e-06,
+      "loss": 6.0493,
+      "step": 374500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.500922950707596e-06,
+      "loss": 6.0402,
+      "step": 375000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.500239283516784e-06,
+      "loss": 6.0459,
+      "step": 375500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.499555616325973e-06,
+      "loss": 6.042,
+      "step": 376000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.498871949135161e-06,
+      "loss": 6.0424,
+      "step": 376500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.498188281944351e-06,
+      "loss": 6.0431,
+      "step": 377000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.49750461475354e-06,
+      "loss": 6.0445,
+      "step": 377500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.496820947562727e-06,
+      "loss": 6.0464,
+      "step": 378000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.496137280371915e-06,
+      "loss": 6.0407,
+      "step": 378500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.495453613181103e-06,
+      "loss": 6.0395,
+      "step": 379000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.494769945990293e-06,
+      "loss": 6.0395,
+      "step": 379500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.494086278799481e-06,
+      "loss": 6.0402,
+      "step": 380000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.49340261160867e-06,
+      "loss": 6.0392,
+      "step": 380500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.492718944417857e-06,
+      "loss": 6.0417,
+      "step": 381000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.492035277227047e-06,
+      "loss": 6.0442,
+      "step": 381500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.491351610036235e-06,
+      "loss": 6.0378,
+      "step": 382000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.490667942845425e-06,
+      "loss": 6.0417,
+      "step": 382500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.489984275654613e-06,
+      "loss": 6.0358,
+      "step": 383000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.489300608463801e-06,
+      "loss": 6.034,
+      "step": 383500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.488616941272989e-06,
+      "loss": 6.0343,
+      "step": 384000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.487933274082177e-06,
+      "loss": 6.0383,
+      "step": 384500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.487249606891365e-06,
+      "loss": 6.0403,
+      "step": 385000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.486565939700555e-06,
+      "loss": 6.0388,
+      "step": 385500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.485882272509743e-06,
+      "loss": 6.038,
+      "step": 386000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.485198605318931e-06,
+      "loss": 6.037,
+      "step": 386500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.48451493812812e-06,
+      "loss": 6.0428,
+      "step": 387000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.483831270937309e-06,
+      "loss": 6.0405,
+      "step": 387500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.483147603746497e-06,
+      "loss": 6.0431,
+      "step": 388000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.482463936555687e-06,
+      "loss": 6.0359,
+      "step": 388500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.481780269364875e-06,
+      "loss": 6.0372,
+      "step": 389000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.481096602174063e-06,
+      "loss": 6.0367,
+      "step": 389500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.48041293498325e-06,
+      "loss": 6.0395,
+      "step": 390000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.479729267792439e-06,
+      "loss": 6.0382,
+      "step": 390500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.479045600601627e-06,
+      "loss": 6.0407,
+      "step": 391000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.478361933410817e-06,
+      "loss": 6.0376,
+      "step": 391500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.477678266220005e-06,
+      "loss": 6.036,
+      "step": 392000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.476994599029193e-06,
+      "loss": 6.0381,
+      "step": 392500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.476310931838382e-06,
+      "loss": 6.0339,
+      "step": 393000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.47562726464757e-06,
+      "loss": 6.0387,
+      "step": 393500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.474943597456758e-06,
+      "loss": 6.0328,
+      "step": 394000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.474259930265948e-06,
+      "loss": 6.0338,
+      "step": 394500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.473576263075136e-06,
+      "loss": 6.0361,
+      "step": 395000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.472892595884324e-06,
+      "loss": 6.0375,
+      "step": 395500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.472208928693512e-06,
+      "loss": 6.0383,
+      "step": 396000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.4715252615027e-06,
+      "loss": 6.0361,
+      "step": 396500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.470841594311888e-06,
+      "loss": 6.0288,
+      "step": 397000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.470157927121078e-06,
+      "loss": 6.0347,
+      "step": 397500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.469474259930266e-06,
+      "loss": 6.0351,
+      "step": 398000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.468790592739456e-06,
+      "loss": 6.0356,
+      "step": 398500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.468106925548644e-06,
+      "loss": 6.0284,
+      "step": 399000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.467423258357832e-06,
+      "loss": 6.0288,
+      "step": 399500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.46673959116702e-06,
+      "loss": 6.034,
+      "step": 400000
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.46605592397621e-06,
+      "loss": 6.0313,
+      "step": 400500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.465372256785398e-06,
+      "loss": 6.0371,
+      "step": 401000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.464688589594586e-06,
+      "loss": 6.0305,
+      "step": 401500
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.464004922403774e-06,
+      "loss": 6.035,
+      "step": 402000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.463321255212962e-06,
+      "loss": 6.0367,
+      "step": 402500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.462637588022152e-06,
+      "loss": 6.0301,
+      "step": 403000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 9.46195392083134e-06,
+      "loss": 6.0362,
+      "step": 403500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 9.461270253640528e-06,
+      "loss": 6.0266,
+      "step": 404000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 9.460586586449718e-06,
+      "loss": 6.0325,
+      "step": 404500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 9.459902919258906e-06,
+      "loss": 6.0345,
+      "step": 405000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 9.459219252068094e-06,
+      "loss": 6.037,
+      "step": 405500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 9.458535584877282e-06,
+      "loss": 6.032,
+      "step": 406000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 9.457851917686472e-06,
+      "loss": 6.0298,
+      "step": 406500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 9.45716825049566e-06,
+      "loss": 6.0303,
+      "step": 407000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 9.456484583304848e-06,
+      "loss": 6.0324,
+      "step": 407500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 9.455800916114036e-06,
+      "loss": 6.0311,
+      "step": 408000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 9.455117248923224e-06,
+      "loss": 6.029,
+      "step": 408500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 9.454433581732414e-06,
+      "loss": 6.0321,
+      "step": 409000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 9.453749914541602e-06,
+      "loss": 6.0352,
+      "step": 409500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 9.453066247350791e-06,
+      "loss": 6.0381,
+      "step": 410000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 9.45238258015998e-06,
+      "loss": 6.0296,
+      "step": 410500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 9.451698912969167e-06,
+      "loss": 6.0294,
+      "step": 411000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 9.451015245778355e-06,
+      "loss": 6.0326,
+      "step": 411500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 9.450331578587545e-06,
+      "loss": 6.0252,
+      "step": 412000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 9.449647911396733e-06,
+      "loss": 6.0255,
+      "step": 412500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 9.448964244205921e-06,
+      "loss": 6.0318,
+      "step": 413000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 9.44828057701511e-06,
+      "loss": 6.0281,
+      "step": 413500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 9.447596909824297e-06,
+      "loss": 6.028,
+      "step": 414000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 9.446913242633487e-06,
+      "loss": 6.0259,
+      "step": 414500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 9.446229575442675e-06,
+      "loss": 6.0294,
+      "step": 415000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 9.445545908251865e-06,
+      "loss": 6.0304,
+      "step": 415500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 9.444862241061053e-06,
+      "loss": 6.031,
+      "step": 416000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 9.444178573870241e-06,
+      "loss": 6.0265,
+      "step": 416500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 9.443494906679429e-06,
+      "loss": 6.0249,
+      "step": 417000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 9.442811239488617e-06,
+      "loss": 6.0274,
+      "step": 417500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 9.442127572297807e-06,
+      "loss": 6.027,
+      "step": 418000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 9.441443905106995e-06,
+      "loss": 6.0314,
+      "step": 418500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 9.440760237916183e-06,
+      "loss": 6.0229,
+      "step": 419000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.440076570725371e-06,
+      "loss": 6.0267,
+      "step": 419500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.43939290353456e-06,
+      "loss": 6.0202,
+      "step": 420000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.438709236343749e-06,
+      "loss": 6.0257,
+      "step": 420500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.438025569152937e-06,
+      "loss": 6.0236,
+      "step": 421000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 9.437341901962127e-06,
+      "loss": 6.0215,
+      "step": 421500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 9.436658234771315e-06,
+      "loss": 6.0314,
+      "step": 422000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 9.435974567580503e-06,
+      "loss": 6.0275,
+      "step": 422500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 9.43529090038969e-06,
+      "loss": 6.0275,
+      "step": 423000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 9.434607233198879e-06,
+      "loss": 6.0165,
+      "step": 423500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 9.433923566008069e-06,
+      "loss": 6.0208,
+      "step": 424000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.433239898817257e-06,
+      "loss": 6.0221,
+      "step": 424500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.432556231626445e-06,
+      "loss": 6.0225,
+      "step": 425000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.431872564435633e-06,
+      "loss": 6.0186,
+      "step": 425500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.431188897244822e-06,
+      "loss": 6.0208,
+      "step": 426000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.43050523005401e-06,
+      "loss": 6.0172,
+      "step": 426500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 9.4298215628632e-06,
+      "loss": 6.0255,
+      "step": 427000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 9.429137895672388e-06,
+      "loss": 6.0201,
+      "step": 427500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 9.428454228481576e-06,
+      "loss": 6.0222,
+      "step": 428000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 9.427770561290764e-06,
+      "loss": 6.0177,
+      "step": 428500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 9.427086894099952e-06,
+      "loss": 6.0191,
+      "step": 429000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 9.42640322690914e-06,
+      "loss": 6.0191,
+      "step": 429500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 9.42571955971833e-06,
+      "loss": 6.0173,
+      "step": 430000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 9.425035892527518e-06,
+      "loss": 6.0219,
+      "step": 430500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 9.424352225336706e-06,
+      "loss": 6.013,
+      "step": 431000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 9.423668558145896e-06,
+      "loss": 6.0146,
+      "step": 431500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 9.422984890955084e-06,
+      "loss": 6.0113,
+      "step": 432000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.422301223764272e-06,
+      "loss": 6.0198,
+      "step": 432500
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.421617556573462e-06,
+      "loss": 6.0184,
+      "step": 433000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.42093388938265e-06,
+      "loss": 6.0137,
+      "step": 433500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 9.420250222191838e-06,
+      "loss": 6.0149,
+      "step": 434000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 9.419566555001026e-06,
+      "loss": 6.0084,
+      "step": 434500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 9.418882887810214e-06,
+      "loss": 6.0091,
+      "step": 435000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 9.418199220619402e-06,
+      "loss": 6.0065,
+      "step": 435500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 9.417515553428592e-06,
+      "loss": 6.0168,
+      "step": 436000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 9.41683188623778e-06,
+      "loss": 6.0042,
+      "step": 436500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.416148219046968e-06,
+      "loss": 6.0034,
+      "step": 437000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.415464551856158e-06,
+      "loss": 6.0064,
+      "step": 437500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 9.414780884665346e-06,
+      "loss": 6.0072,
+      "step": 438000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 9.414097217474534e-06,
+      "loss": 6.0033,
+      "step": 438500
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 9.413413550283724e-06,
+      "loss": 6.0028,
+      "step": 439000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 9.412729883092912e-06,
+      "loss": 6.0012,
+      "step": 439500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 9.4120462159021e-06,
+      "loss": 5.9984,
+      "step": 440000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 9.411362548711288e-06,
+      "loss": 5.9979,
+      "step": 440500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 9.410678881520476e-06,
+      "loss": 5.9934,
+      "step": 441000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 9.409995214329664e-06,
+      "loss": 6.0005,
+      "step": 441500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 9.409311547138854e-06,
+      "loss": 5.9979,
+      "step": 442000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 9.408627879948042e-06,
+      "loss": 5.9985,
+      "step": 442500
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 9.407944212757231e-06,
+      "loss": 5.9915,
+      "step": 443000
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 9.40726054556642e-06,
+      "loss": 5.9927,
+      "step": 443500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 9.406576878375607e-06,
+      "loss": 5.9927,
+      "step": 444000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 9.405893211184796e-06,
+      "loss": 5.9935,
+      "step": 444500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 9.405209543993985e-06,
+      "loss": 5.9918,
+      "step": 445000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 9.404525876803173e-06,
+      "loss": 5.9886,
+      "step": 445500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 9.403842209612361e-06,
+      "loss": 5.9878,
+      "step": 446000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.40315854242155e-06,
+      "loss": 5.9827,
+      "step": 446500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.402474875230737e-06,
+      "loss": 5.9828,
+      "step": 447000
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 9.401791208039927e-06,
+      "loss": 5.9874,
+      "step": 447500
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 9.401107540849115e-06,
+      "loss": 5.9843,
+      "step": 448000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 9.400423873658305e-06,
+      "loss": 5.9858,
+      "step": 448500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 9.399740206467493e-06,
+      "loss": 5.9866,
+      "step": 449000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 9.399056539276681e-06,
+      "loss": 5.984,
+      "step": 449500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 9.39837287208587e-06,
+      "loss": 5.9804,
+      "step": 450000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.397689204895057e-06,
+      "loss": 5.975,
+      "step": 450500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 9.397005537704247e-06,
+      "loss": 5.9824,
+      "step": 451000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 9.396321870513435e-06,
+      "loss": 5.9789,
+      "step": 451500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 9.395638203322623e-06,
+      "loss": 5.979,
+      "step": 452000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 9.394954536131811e-06,
+      "loss": 5.9745,
+      "step": 452500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 9.394270868941001e-06,
+      "loss": 5.9773,
+      "step": 453000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 9.393587201750189e-06,
+      "loss": 5.9778,
+      "step": 453500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 9.392903534559377e-06,
+      "loss": 5.9778,
+      "step": 454000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 9.392219867368567e-06,
+      "loss": 5.9747,
+      "step": 454500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 9.391536200177755e-06,
+      "loss": 5.9731,
+      "step": 455000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.390852532986943e-06,
+      "loss": 5.9842,
+      "step": 455500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.390168865796131e-06,
+      "loss": 5.9713,
+      "step": 456000
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 9.389485198605319e-06,
+      "loss": 5.9731,
+      "step": 456500
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 9.388801531414509e-06,
+      "loss": 5.9659,
+      "step": 457000
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 9.388117864223697e-06,
+      "loss": 5.9707,
+      "step": 457500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 9.387434197032885e-06,
+      "loss": 5.97,
+      "step": 458000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 9.386750529842073e-06,
+      "loss": 5.9698,
+      "step": 458500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 9.386066862651263e-06,
+      "loss": 5.9698,
+      "step": 459000
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 9.38538319546045e-06,
+      "loss": 5.9661,
+      "step": 459500
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 9.38469952826964e-06,
+      "loss": 5.9689,
+      "step": 460000
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.384015861078828e-06,
+      "loss": 5.9709,
+      "step": 460500
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 9.383332193888016e-06,
+      "loss": 5.9643,
+      "step": 461000
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 9.382648526697204e-06,
+      "loss": 5.9687,
+      "step": 461500
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 9.381964859506393e-06,
+      "loss": 5.9655,
+      "step": 462000
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 9.38128119231558e-06,
+      "loss": 5.9686,
+      "step": 462500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 9.38059752512477e-06,
+      "loss": 5.9599,
+      "step": 463000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 9.379913857933958e-06,
+      "loss": 5.9617,
+      "step": 463500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 9.379230190743146e-06,
+      "loss": 5.9611,
+      "step": 464000
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 9.378546523552336e-06,
+      "loss": 5.9629,
+      "step": 464500
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 9.377862856361524e-06,
+      "loss": 5.9647,
+      "step": 465000
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 9.377179189170712e-06,
+      "loss": 5.9594,
+      "step": 465500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 9.376495521979902e-06,
+      "loss": 5.9583,
+      "step": 466000
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 9.37581185478909e-06,
+      "loss": 5.959,
+      "step": 466500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 9.375128187598278e-06,
+      "loss": 5.9597,
+      "step": 467000
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 9.374444520407466e-06,
+      "loss": 5.9601,
+      "step": 467500
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 9.373760853216654e-06,
+      "loss": 5.956,
+      "step": 468000
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 9.373077186025842e-06,
+      "loss": 5.9533,
+      "step": 468500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 9.372393518835032e-06,
+      "loss": 5.9575,
+      "step": 469000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 9.37170985164422e-06,
+      "loss": 5.9522,
+      "step": 469500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 9.371026184453408e-06,
+      "loss": 5.9535,
+      "step": 470000
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 9.370342517262598e-06,
+      "loss": 5.9491,
+      "step": 470500
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 9.369658850071786e-06,
+      "loss": 5.9525,
+      "step": 471000
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 9.368975182880974e-06,
+      "loss": 5.9514,
+      "step": 471500
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 9.368291515690164e-06,
+      "loss": 5.9557,
+      "step": 472000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 9.367607848499352e-06,
+      "loss": 5.9464,
+      "step": 472500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 9.36692418130854e-06,
+      "loss": 5.9487,
+      "step": 473000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 9.366240514117728e-06,
+      "loss": 5.9491,
+      "step": 473500
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 9.365556846926916e-06,
+      "loss": 5.9474,
+      "step": 474000
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 9.364873179736104e-06,
+      "loss": 5.9449,
+      "step": 474500
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 9.364189512545294e-06,
+      "loss": 5.945,
+      "step": 475000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 9.363505845354482e-06,
+      "loss": 5.9483,
+      "step": 475500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 9.362822178163671e-06,
+      "loss": 5.9466,
+      "step": 476000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 9.36213851097286e-06,
+      "loss": 5.9456,
+      "step": 476500
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 9.361454843782048e-06,
+      "loss": 5.9489,
+      "step": 477000
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 9.360771176591236e-06,
+      "loss": 5.944,
+      "step": 477500
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 9.360087509400425e-06,
+      "loss": 5.9419,
+      "step": 478000
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 9.359403842209613e-06,
+      "loss": 5.9402,
+      "step": 478500
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 9.358720175018801e-06,
+      "loss": 5.9412,
+      "step": 479000
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 9.35803650782799e-06,
+      "loss": 5.9385,
+      "step": 479500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 9.357352840637178e-06,
+      "loss": 5.9386,
+      "step": 480000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 9.356669173446367e-06,
+      "loss": 5.9439,
+      "step": 480500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 9.355985506255555e-06,
+      "loss": 5.9425,
+      "step": 481000
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 9.355301839064745e-06,
+      "loss": 5.9396,
+      "step": 481500
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 9.354618171873933e-06,
+      "loss": 5.9428,
+      "step": 482000
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 9.353934504683121e-06,
+      "loss": 5.9399,
+      "step": 482500
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 9.35325083749231e-06,
+      "loss": 5.9384,
+      "step": 483000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 9.352567170301497e-06,
+      "loss": 5.9384,
+      "step": 483500
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 9.351883503110687e-06,
+      "loss": 5.9323,
+      "step": 484000
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 9.351199835919875e-06,
+      "loss": 5.9386,
+      "step": 484500
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 9.350516168729063e-06,
+      "loss": 5.9331,
+      "step": 485000
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 9.349832501538251e-06,
+      "loss": 5.9375,
+      "step": 485500
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 9.349148834347441e-06,
+      "loss": 5.9371,
+      "step": 486000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 9.348465167156629e-06,
+      "loss": 5.9286,
+      "step": 486500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 9.347781499965817e-06,
+      "loss": 5.9336,
+      "step": 487000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 9.347097832775007e-06,
+      "loss": 5.9309,
+      "step": 487500
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 9.346414165584195e-06,
+      "loss": 5.9319,
+      "step": 488000
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 9.345730498393383e-06,
+      "loss": 5.9297,
+      "step": 488500
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.345046831202571e-06,
+      "loss": 5.9369,
+      "step": 489000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 9.344363164011759e-06,
+      "loss": 5.9264,
+      "step": 489500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 9.343679496820949e-06,
+      "loss": 5.9286,
+      "step": 490000
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.342995829630137e-06,
+      "loss": 5.9308,
+      "step": 490500
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 9.342312162439325e-06,
+      "loss": 5.927,
+      "step": 491000
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 9.341628495248513e-06,
+      "loss": 5.9236,
+      "step": 491500
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.340944828057703e-06,
+      "loss": 5.9277,
+      "step": 492000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 9.34026116086689e-06,
+      "loss": 5.9237,
+      "step": 492500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 9.33957749367608e-06,
+      "loss": 5.9262,
+      "step": 493000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 9.338893826485268e-06,
+      "loss": 5.9228,
+      "step": 493500
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.338210159294457e-06,
+      "loss": 5.9228,
+      "step": 494000
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 9.337526492103645e-06,
+      "loss": 5.9279,
+      "step": 494500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 9.336842824912833e-06,
+      "loss": 5.9185,
+      "step": 495000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.33615915772202e-06,
+      "loss": 5.9237,
+      "step": 495500
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 9.33547549053121e-06,
+      "loss": 5.9261,
+      "step": 496000
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 9.334791823340398e-06,
+      "loss": 5.9241,
+      "step": 496500
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.334108156149587e-06,
+      "loss": 5.9227,
+      "step": 497000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 9.333424488958776e-06,
+      "loss": 5.918,
+      "step": 497500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 9.332740821767964e-06,
+      "loss": 5.9159,
+      "step": 498000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.332057154577152e-06,
+      "loss": 5.9154,
+      "step": 498500
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 9.331373487386342e-06,
+      "loss": 5.9199,
+      "step": 499000
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 9.33068982019553e-06,
+      "loss": 5.9181,
+      "step": 499500
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.330006153004718e-06,
+      "loss": 5.9175,
+      "step": 500000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 9.329322485813906e-06,
+      "loss": 5.9158,
+      "step": 500500
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 9.328638818623094e-06,
+      "loss": 5.9132,
+      "step": 501000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.327955151432282e-06,
+      "loss": 5.913,
+      "step": 501500
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 9.327271484241472e-06,
+      "loss": 5.9074,
+      "step": 502000
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 9.32658781705066e-06,
+      "loss": 5.9162,
+      "step": 502500
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 9.32590414985985e-06,
+      "loss": 5.9133,
+      "step": 503000
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.325220482669038e-06,
+      "loss": 5.9132,
+      "step": 503500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 9.324536815478226e-06,
+      "loss": 5.9076,
+      "step": 504000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 9.323853148287414e-06,
+      "loss": 5.9034,
+      "step": 504500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.323169481096604e-06,
+      "loss": 5.9072,
+      "step": 505000
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 9.322485813905792e-06,
+      "loss": 5.9064,
+      "step": 505500
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 9.32180214671498e-06,
+      "loss": 5.9125,
+      "step": 506000
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.321118479524168e-06,
+      "loss": 5.9111,
+      "step": 506500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 9.320434812333356e-06,
+      "loss": 5.9048,
+      "step": 507000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 9.319751145142544e-06,
+      "loss": 5.9038,
+      "step": 507500
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.319067477951734e-06,
+      "loss": 5.9031,
+      "step": 508000
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 9.318383810760922e-06,
+      "loss": 5.9058,
+      "step": 508500
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 9.317700143570112e-06,
+      "loss": 5.9047,
+      "step": 509000
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.3170164763793e-06,
+      "loss": 5.9045,
+      "step": 509500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 9.316332809188488e-06,
+      "loss": 5.9045,
+      "step": 510000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 9.315649141997676e-06,
+      "loss": 5.9053,
+      "step": 510500
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.314965474806865e-06,
+      "loss": 5.9022,
+      "step": 511000
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 9.314281807616054e-06,
+      "loss": 5.9026,
+      "step": 511500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 9.313598140425242e-06,
+      "loss": 5.9005,
+      "step": 512000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.31291447323443e-06,
+      "loss": 5.8989,
+      "step": 512500
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 9.312230806043618e-06,
+      "loss": 5.8971,
+      "step": 513000
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 9.311547138852807e-06,
+      "loss": 5.8995,
+      "step": 513500
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 9.310863471661995e-06,
+      "loss": 5.8988,
+      "step": 514000
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.310179804471185e-06,
+      "loss": 5.8953,
+      "step": 514500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 9.309496137280373e-06,
+      "loss": 5.896,
+      "step": 515000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 9.308812470089561e-06,
+      "loss": 5.8942,
+      "step": 515500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.30812880289875e-06,
+      "loss": 5.8981,
+      "step": 516000
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 9.307445135707937e-06,
+      "loss": 5.8965,
+      "step": 516500
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 9.306761468517127e-06,
+      "loss": 5.893,
+      "step": 517000
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.306077801326315e-06,
+      "loss": 5.8917,
+      "step": 517500
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 9.305394134135503e-06,
+      "loss": 5.8947,
+      "step": 518000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 9.304710466944691e-06,
+      "loss": 5.8944,
+      "step": 518500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.304026799753881e-06,
+      "loss": 5.895,
+      "step": 519000
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 9.303343132563069e-06,
+      "loss": 5.8906,
+      "step": 519500
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 9.302659465372257e-06,
+      "loss": 5.895,
+      "step": 520000
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.301975798181447e-06,
+      "loss": 5.8848,
+      "step": 520500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 9.301292130990635e-06,
+      "loss": 5.8916,
+      "step": 521000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 9.300608463799823e-06,
+      "loss": 5.8867,
+      "step": 521500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.299924796609011e-06,
+      "loss": 5.8875,
+      "step": 522000
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 9.299241129418199e-06,
+      "loss": 5.8868,
+      "step": 522500
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 9.298557462227389e-06,
+      "loss": 5.8856,
+      "step": 523000
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 9.297873795036577e-06,
+      "loss": 5.8832,
+      "step": 523500
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.297190127845765e-06,
+      "loss": 5.8878,
+      "step": 524000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 9.296506460654953e-06,
+      "loss": 5.8876,
+      "step": 524500
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 9.295822793464143e-06,
+      "loss": 5.8852,
+      "step": 525000
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.29513912627333e-06,
+      "loss": 5.88,
+      "step": 525500
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 9.29445545908252e-06,
+      "loss": 5.8841,
+      "step": 526000
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 9.293771791891709e-06,
+      "loss": 5.8836,
+      "step": 526500
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.293088124700897e-06,
+      "loss": 5.8858,
+      "step": 527000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 9.292404457510085e-06,
+      "loss": 5.8792,
+      "step": 527500
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 9.291720790319273e-06,
+      "loss": 5.8797,
+      "step": 528000
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.29103712312846e-06,
+      "loss": 5.887,
+      "step": 528500
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 9.29035345593765e-06,
+      "loss": 5.8798,
+      "step": 529000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 9.289669788746839e-06,
+      "loss": 5.8761,
+      "step": 529500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.288986121556027e-06,
+      "loss": 5.8746,
+      "step": 530000
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 9.288302454365216e-06,
+      "loss": 5.8749,
+      "step": 530500
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 9.287618787174404e-06,
+      "loss": 5.8779,
+      "step": 531000
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.286935119983592e-06,
+      "loss": 5.8755,
+      "step": 531500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 9.286251452792782e-06,
+      "loss": 5.8767,
+      "step": 532000
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 9.28556778560197e-06,
+      "loss": 5.8718,
+      "step": 532500
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.284884118411158e-06,
+      "loss": 5.874,
+      "step": 533000
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 9.284200451220346e-06,
+      "loss": 5.8699,
+      "step": 533500
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 9.283516784029534e-06,
+      "loss": 5.8713,
+      "step": 534000
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 9.282833116838722e-06,
+      "loss": 5.8766,
+      "step": 534500
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 9.282149449647912e-06,
+      "loss": 5.8737,
+      "step": 535000
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 9.2814657824571e-06,
+      "loss": 5.875,
+      "step": 535500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 9.28078211526629e-06,
+      "loss": 5.8669,
+      "step": 536000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 9.280098448075478e-06,
+      "loss": 5.8733,
+      "step": 536500
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 9.279414780884666e-06,
+      "loss": 5.874,
+      "step": 537000
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 9.278731113693854e-06,
+      "loss": 5.8697,
+      "step": 537500
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 9.278047446503044e-06,
+      "loss": 5.8661,
+      "step": 538000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 9.277363779312232e-06,
+      "loss": 5.8684,
+      "step": 538500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 9.27668011212142e-06,
+      "loss": 5.8701,
+      "step": 539000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 9.275996444930608e-06,
+      "loss": 5.8686,
+      "step": 539500
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 9.275312777739796e-06,
+      "loss": 5.8685,
+      "step": 540000
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 9.274629110548986e-06,
+      "loss": 5.8735,
+      "step": 540500
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 9.273945443358174e-06,
+      "loss": 5.8702,
+      "step": 541000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 9.273261776167362e-06,
+      "loss": 5.8627,
+      "step": 541500
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 9.272578108976552e-06,
+      "loss": 5.8634,
+      "step": 542000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 9.27189444178574e-06,
+      "loss": 5.8624,
+      "step": 542500
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 9.271210774594928e-06,
+      "loss": 5.8616,
+      "step": 543000
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 9.270527107404117e-06,
+      "loss": 5.865,
+      "step": 543500
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 9.269843440213306e-06,
+      "loss": 5.8615,
+      "step": 544000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 9.269159773022494e-06,
+      "loss": 5.8617,
+      "step": 544500
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 9.268476105831682e-06,
+      "loss": 5.8606,
+      "step": 545000
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 9.26779243864087e-06,
+      "loss": 5.8613,
+      "step": 545500
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 9.267108771450058e-06,
+      "loss": 5.8646,
+      "step": 546000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 9.266425104259247e-06,
+      "loss": 5.8626,
+      "step": 546500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 9.265741437068436e-06,
+      "loss": 5.8559,
+      "step": 547000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 9.265057769877625e-06,
+      "loss": 5.8604,
+      "step": 547500
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 9.264374102686813e-06,
+      "loss": 5.8622,
+      "step": 548000
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 9.263690435496001e-06,
+      "loss": 5.8642,
+      "step": 548500
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 9.26300676830519e-06,
+      "loss": 5.8626,
+      "step": 549000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 9.26232310111438e-06,
+      "loss": 5.855,
+      "step": 549500
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 9.261639433923567e-06,
+      "loss": 5.8544,
+      "step": 550000
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 9.260955766732755e-06,
+      "loss": 5.8607,
+      "step": 550500
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 9.260272099541943e-06,
+      "loss": 5.865,
+      "step": 551000
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 9.259588432351131e-06,
+      "loss": 5.8541,
+      "step": 551500
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 9.258904765160321e-06,
+      "loss": 5.8552,
+      "step": 552000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 9.25822109796951e-06,
+      "loss": 5.8555,
+      "step": 552500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 9.257537430778697e-06,
+      "loss": 5.8544,
+      "step": 553000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 9.256853763587887e-06,
+      "loss": 5.8584,
+      "step": 553500
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 9.256170096397075e-06,
+      "loss": 5.8593,
+      "step": 554000
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 9.255486429206263e-06,
+      "loss": 5.8521,
+      "step": 554500
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 9.254802762015451e-06,
+      "loss": 5.8505,
+      "step": 555000
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 9.254119094824641e-06,
+      "loss": 5.8511,
+      "step": 555500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 9.253435427633829e-06,
+      "loss": 5.8536,
+      "step": 556000
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 9.252751760443017e-06,
+      "loss": 5.8498,
+      "step": 556500
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 9.252068093252205e-06,
+      "loss": 5.8513,
+      "step": 557000
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 9.251384426061393e-06,
+      "loss": 5.8487,
+      "step": 557500
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 9.250700758870583e-06,
+      "loss": 5.851,
+      "step": 558000
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 9.25001709167977e-06,
+      "loss": 5.8513,
+      "step": 558500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 9.24933342448896e-06,
+      "loss": 5.8503,
+      "step": 559000
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 9.248649757298149e-06,
+      "loss": 5.8491,
+      "step": 559500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 9.247966090107337e-06,
+      "loss": 5.8458,
+      "step": 560000
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 9.247282422916525e-06,
+      "loss": 5.8434,
+      "step": 560500
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 9.246598755725713e-06,
+      "loss": 5.8475,
+      "step": 561000
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 9.245915088534903e-06,
+      "loss": 5.8439,
+      "step": 561500
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 9.24523142134409e-06,
+      "loss": 5.8456,
+      "step": 562000
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 9.244547754153279e-06,
+      "loss": 5.8422,
+      "step": 562500
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 9.243864086962467e-06,
+      "loss": 5.8422,
+      "step": 563000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 9.243180419771656e-06,
+      "loss": 5.8447,
+      "step": 563500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 9.242496752580844e-06,
+      "loss": 5.845,
+      "step": 564000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 9.241813085390034e-06,
+      "loss": 5.846,
+      "step": 564500
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 9.241129418199222e-06,
+      "loss": 5.8435,
+      "step": 565000
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 9.24044575100841e-06,
+      "loss": 5.8368,
+      "step": 565500
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 9.239762083817598e-06,
+      "loss": 5.8471,
+      "step": 566000
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.239078416626786e-06,
+      "loss": 5.8385,
+      "step": 566500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 9.238394749435974e-06,
+      "loss": 5.8367,
+      "step": 567000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 9.237711082245164e-06,
+      "loss": 5.8399,
+      "step": 567500
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.237027415054352e-06,
+      "loss": 5.8403,
+      "step": 568000
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 9.23634374786354e-06,
+      "loss": 5.837,
+      "step": 568500
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 9.23566008067273e-06,
+      "loss": 5.8415,
+      "step": 569000
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.234976413481918e-06,
+      "loss": 5.8459,
+      "step": 569500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 9.234292746291106e-06,
+      "loss": 5.8356,
+      "step": 570000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 9.233609079100296e-06,
+      "loss": 5.8402,
+      "step": 570500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.232925411909484e-06,
+      "loss": 5.8404,
+      "step": 571000
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 9.232241744718672e-06,
+      "loss": 5.8347,
+      "step": 571500
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 9.23155807752786e-06,
+      "loss": 5.8388,
+      "step": 572000
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.230874410337048e-06,
+      "loss": 5.833,
+      "step": 572500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 9.230190743146236e-06,
+      "loss": 5.8362,
+      "step": 573000
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 9.229507075955426e-06,
+      "loss": 5.8389,
+      "step": 573500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.228823408764614e-06,
+      "loss": 5.8321,
+      "step": 574000
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 9.228139741573802e-06,
+      "loss": 5.8356,
+      "step": 574500
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 9.227456074382992e-06,
+      "loss": 5.835,
+      "step": 575000
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 9.22677240719218e-06,
+      "loss": 5.8339,
+      "step": 575500
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.226088740001368e-06,
+      "loss": 5.8272,
+      "step": 576000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 9.225405072810558e-06,
+      "loss": 5.8301,
+      "step": 576500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 9.224721405619746e-06,
+      "loss": 5.8338,
+      "step": 577000
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.224037738428934e-06,
+      "loss": 5.8293,
+      "step": 577500
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 9.223354071238122e-06,
+      "loss": 5.8306,
+      "step": 578000
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 9.22267040404731e-06,
+      "loss": 5.8323,
+      "step": 578500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.221986736856498e-06,
+      "loss": 5.8301,
+      "step": 579000
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 9.221303069665688e-06,
+      "loss": 5.8296,
+      "step": 579500
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 9.220619402474876e-06,
+      "loss": 5.8294,
+      "step": 580000
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.219935735284065e-06,
+      "loss": 5.8273,
+      "step": 580500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 9.219252068093253e-06,
+      "loss": 5.8277,
+      "step": 581000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 9.218568400902441e-06,
+      "loss": 5.8265,
+      "step": 581500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.21788473371163e-06,
+      "loss": 5.828,
+      "step": 582000
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 9.21720106652082e-06,
+      "loss": 5.827,
+      "step": 582500
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 9.216517399330007e-06,
+      "loss": 5.8254,
+      "step": 583000
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 9.215833732139195e-06,
+      "loss": 5.826,
+      "step": 583500
+    },
+    {
+      "epoch": 7.97,
+      "step": 583837,
+      "total_flos": 3.919564805232394e+17,
+      "train_loss": 6.144668718638239,
+      "train_runtime": 197998.5867,
+      "train_samples_per_second": 2958.99,
+      "train_steps_per_second": 36.988
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 7323500,
+  "num_train_epochs": 100,
+  "save_steps": 1000000,
+  "total_flos": 3.919564805232394e+17,
+  "trial_name": null,
+  "trial_params": null
+}