diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5494 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 6.222953505837372,
+  "eval_steps": 1000000,
+  "global_step": 455738,
+  "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": 10.2368,
+      "step": 500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 9.9729,
+      "step": 1000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5e-06,
+      "loss": 9.7932,
+      "step": 1500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 9.5809,
+      "step": 2000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.5e-06,
+      "loss": 9.3091,
+      "step": 2500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3e-06,
+      "loss": 8.9945,
+      "step": 3000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.5e-06,
+      "loss": 8.6527,
+      "step": 3500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 8.3065,
+      "step": 4000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.5e-06,
+      "loss": 7.9756,
+      "step": 4500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5e-06,
+      "loss": 7.6748,
+      "step": 5000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 7.4353,
+      "step": 5500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 6e-06,
+      "loss": 7.2603,
+      "step": 6000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 7.1461,
+      "step": 6500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7e-06,
+      "loss": 7.0636,
+      "step": 7000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 7.0055,
+      "step": 7500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 6.9559,
+      "step": 8000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.5e-06,
+      "loss": 6.9146,
+      "step": 8500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9e-06,
+      "loss": 6.8756,
+      "step": 9000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.5e-06,
+      "loss": 6.842,
+      "step": 9500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1e-05,
+      "loss": 6.8219,
+      "step": 10000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.999316332809189e-06,
+      "loss": 6.7882,
+      "step": 10500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.998632665618377e-06,
+      "loss": 6.7594,
+      "step": 11000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.997948998427567e-06,
+      "loss": 6.74,
+      "step": 11500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.997265331236755e-06,
+      "loss": 6.7131,
+      "step": 12000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.996581664045943e-06,
+      "loss": 6.6966,
+      "step": 12500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.995897996855133e-06,
+      "loss": 6.6779,
+      "step": 13000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.99521432966432e-06,
+      "loss": 6.6602,
+      "step": 13500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.994530662473509e-06,
+      "loss": 6.648,
+      "step": 14000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.993846995282698e-06,
+      "loss": 6.6332,
+      "step": 14500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.993163328091886e-06,
+      "loss": 6.6176,
+      "step": 15000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.992479660901074e-06,
+      "loss": 6.6059,
+      "step": 15500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.991795993710262e-06,
+      "loss": 6.5995,
+      "step": 16000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.99111232651945e-06,
+      "loss": 6.5824,
+      "step": 16500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.990428659328639e-06,
+      "loss": 6.5744,
+      "step": 17000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.989744992137828e-06,
+      "loss": 6.5615,
+      "step": 17500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.989061324947016e-06,
+      "loss": 6.5557,
+      "step": 18000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.988377657756204e-06,
+      "loss": 6.5419,
+      "step": 18500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.987693990565394e-06,
+      "loss": 6.5298,
+      "step": 19000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.987010323374582e-06,
+      "loss": 6.5268,
+      "step": 19500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.98632665618377e-06,
+      "loss": 6.5117,
+      "step": 20000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.98564298899296e-06,
+      "loss": 6.5034,
+      "step": 20500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.984959321802148e-06,
+      "loss": 6.4983,
+      "step": 21000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.984275654611336e-06,
+      "loss": 6.4926,
+      "step": 21500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.983591987420524e-06,
+      "loss": 6.4785,
+      "step": 22000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.982908320229712e-06,
+      "loss": 6.4782,
+      "step": 22500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.9822246530389e-06,
+      "loss": 6.4682,
+      "step": 23000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.98154098584809e-06,
+      "loss": 6.4628,
+      "step": 23500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.980857318657278e-06,
+      "loss": 6.4513,
+      "step": 24000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.980173651466468e-06,
+      "loss": 6.4485,
+      "step": 24500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.979489984275656e-06,
+      "loss": 6.4429,
+      "step": 25000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.978806317084844e-06,
+      "loss": 6.4352,
+      "step": 25500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.978122649894032e-06,
+      "loss": 6.4294,
+      "step": 26000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.977438982703222e-06,
+      "loss": 6.4259,
+      "step": 26500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.97675531551241e-06,
+      "loss": 6.4145,
+      "step": 27000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.976071648321598e-06,
+      "loss": 6.407,
+      "step": 27500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.975387981130786e-06,
+      "loss": 6.4003,
+      "step": 28000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.974704313939974e-06,
+      "loss": 6.405,
+      "step": 28500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.974020646749164e-06,
+      "loss": 6.3946,
+      "step": 29000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.973336979558352e-06,
+      "loss": 6.3884,
+      "step": 29500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.972653312367541e-06,
+      "loss": 6.3852,
+      "step": 30000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.97196964517673e-06,
+      "loss": 6.3799,
+      "step": 30500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.971285977985918e-06,
+      "loss": 6.3758,
+      "step": 31000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.970602310795106e-06,
+      "loss": 6.3645,
+      "step": 31500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.969918643604294e-06,
+      "loss": 6.3563,
+      "step": 32000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.969234976413483e-06,
+      "loss": 6.3604,
+      "step": 32500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.968551309222671e-06,
+      "loss": 6.3518,
+      "step": 33000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.96786764203186e-06,
+      "loss": 6.3477,
+      "step": 33500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.967183974841048e-06,
+      "loss": 6.3504,
+      "step": 34000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.966500307650237e-06,
+      "loss": 6.3435,
+      "step": 34500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.965816640459425e-06,
+      "loss": 6.3337,
+      "step": 35000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.965132973268613e-06,
+      "loss": 6.3341,
+      "step": 35500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.964449306077803e-06,
+      "loss": 6.3344,
+      "step": 36000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.963765638886991e-06,
+      "loss": 6.3278,
+      "step": 36500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.96308197169618e-06,
+      "loss": 6.3238,
+      "step": 37000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.962398304505367e-06,
+      "loss": 6.3191,
+      "step": 37500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961714637314555e-06,
+      "loss": 6.3159,
+      "step": 38000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.961030970123745e-06,
+      "loss": 6.3086,
+      "step": 38500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.960347302932933e-06,
+      "loss": 6.3078,
+      "step": 39000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.959663635742121e-06,
+      "loss": 6.3129,
+      "step": 39500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.95897996855131e-06,
+      "loss": 6.3021,
+      "step": 40000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.958296301360499e-06,
+      "loss": 6.3024,
+      "step": 40500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.957612634169687e-06,
+      "loss": 6.2857,
+      "step": 41000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.956928966978877e-06,
+      "loss": 6.2886,
+      "step": 41500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.956245299788065e-06,
+      "loss": 6.2859,
+      "step": 42000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.955561632597253e-06,
+      "loss": 6.2885,
+      "step": 42500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.954877965406441e-06,
+      "loss": 6.282,
+      "step": 43000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.954194298215629e-06,
+      "loss": 6.2811,
+      "step": 43500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.953510631024817e-06,
+      "loss": 6.2791,
+      "step": 44000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.952826963834007e-06,
+      "loss": 6.2789,
+      "step": 44500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.952143296643195e-06,
+      "loss": 6.2708,
+      "step": 45000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.951459629452383e-06,
+      "loss": 6.2658,
+      "step": 45500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.950775962261573e-06,
+      "loss": 6.2669,
+      "step": 46000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.95009229507076e-06,
+      "loss": 6.2636,
+      "step": 46500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.949408627879949e-06,
+      "loss": 6.2613,
+      "step": 47000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.948724960689138e-06,
+      "loss": 6.2563,
+      "step": 47500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.948041293498326e-06,
+      "loss": 6.2617,
+      "step": 48000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.947357626307515e-06,
+      "loss": 6.251,
+      "step": 48500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.946673959116703e-06,
+      "loss": 6.2501,
+      "step": 49000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.94599029192589e-06,
+      "loss": 6.2479,
+      "step": 49500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 9.945306624735079e-06,
+      "loss": 6.2392,
+      "step": 50000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 9.944622957544268e-06,
+      "loss": 6.2433,
+      "step": 50500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.943939290353456e-06,
+      "loss": 6.2414,
+      "step": 51000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 9.943255623162645e-06,
+      "loss": 6.2404,
+      "step": 51500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 9.942571955971834e-06,
+      "loss": 6.233,
+      "step": 52000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.941888288781022e-06,
+      "loss": 6.2372,
+      "step": 52500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 9.94120462159021e-06,
+      "loss": 6.228,
+      "step": 53000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 9.9405209543994e-06,
+      "loss": 6.2317,
+      "step": 53500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.939837287208588e-06,
+      "loss": 6.2273,
+      "step": 54000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 9.939153620017776e-06,
+      "loss": 6.2273,
+      "step": 54500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 9.938469952826964e-06,
+      "loss": 6.2224,
+      "step": 55000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.937786285636152e-06,
+      "loss": 6.2187,
+      "step": 55500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 9.93710261844534e-06,
+      "loss": 6.2192,
+      "step": 56000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 9.93641895125453e-06,
+      "loss": 6.2215,
+      "step": 56500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 9.935735284063718e-06,
+      "loss": 6.2079,
+      "step": 57000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.935051616872908e-06,
+      "loss": 6.2151,
+      "step": 57500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 9.934367949682096e-06,
+      "loss": 6.2102,
+      "step": 58000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.933684282491284e-06,
+      "loss": 6.2094,
+      "step": 58500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.933000615300472e-06,
+      "loss": 6.2094,
+      "step": 59000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.932316948109662e-06,
+      "loss": 6.2078,
+      "step": 59500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 9.93163328091885e-06,
+      "loss": 6.2042,
+      "step": 60000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.930949613728038e-06,
+      "loss": 6.1982,
+      "step": 60500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 9.930265946537226e-06,
+      "loss": 6.2002,
+      "step": 61000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 9.929582279346414e-06,
+      "loss": 6.2002,
+      "step": 61500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.928898612155604e-06,
+      "loss": 6.1965,
+      "step": 62000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 9.928214944964792e-06,
+      "loss": 6.1965,
+      "step": 62500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.927531277773982e-06,
+      "loss": 6.1878,
+      "step": 63000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.92684761058317e-06,
+      "loss": 6.1942,
+      "step": 63500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.926163943392358e-06,
+      "loss": 6.1948,
+      "step": 64000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 9.925480276201546e-06,
+      "loss": 6.1964,
+      "step": 64500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.924796609010734e-06,
+      "loss": 6.1861,
+      "step": 65000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 9.924112941819923e-06,
+      "loss": 6.188,
+      "step": 65500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 9.923429274629112e-06,
+      "loss": 6.1834,
+      "step": 66000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.9227456074383e-06,
+      "loss": 6.1823,
+      "step": 66500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 9.922061940247488e-06,
+      "loss": 6.1827,
+      "step": 67000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 9.921378273056677e-06,
+      "loss": 6.1829,
+      "step": 67500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 9.920694605865865e-06,
+      "loss": 6.1833,
+      "step": 68000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.920010938675053e-06,
+      "loss": 6.1819,
+      "step": 68500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.919327271484243e-06,
+      "loss": 6.175,
+      "step": 69000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 9.918643604293431e-06,
+      "loss": 6.1783,
+      "step": 69500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.91795993710262e-06,
+      "loss": 6.1741,
+      "step": 70000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.917276269911807e-06,
+      "loss": 6.1765,
+      "step": 70500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 9.916592602720995e-06,
+      "loss": 6.1717,
+      "step": 71000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.915908935530185e-06,
+      "loss": 6.1732,
+      "step": 71500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.915225268339373e-06,
+      "loss": 6.1685,
+      "step": 72000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.914541601148561e-06,
+      "loss": 6.1679,
+      "step": 72500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.91385793395775e-06,
+      "loss": 6.1651,
+      "step": 73000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.913174266766939e-06,
+      "loss": 6.1638,
+      "step": 73500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 9.912490599576127e-06,
+      "loss": 6.1671,
+      "step": 74000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.911806932385317e-06,
+      "loss": 6.1635,
+      "step": 74500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 9.911123265194505e-06,
+      "loss": 6.1628,
+      "step": 75000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 9.910439598003693e-06,
+      "loss": 6.165,
+      "step": 75500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.909755930812881e-06,
+      "loss": 6.156,
+      "step": 76000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 9.909072263622069e-06,
+      "loss": 6.1527,
+      "step": 76500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 9.908388596431257e-06,
+      "loss": 6.1524,
+      "step": 77000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 9.907704929240447e-06,
+      "loss": 6.1571,
+      "step": 77500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.907021262049635e-06,
+      "loss": 6.1552,
+      "step": 78000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 9.906337594858823e-06,
+      "loss": 6.1551,
+      "step": 78500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 9.905653927668013e-06,
+      "loss": 6.1516,
+      "step": 79000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.9049702604772e-06,
+      "loss": 6.1531,
+      "step": 79500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 9.904286593286389e-06,
+      "loss": 6.1479,
+      "step": 80000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 9.903602926095579e-06,
+      "loss": 6.151,
+      "step": 80500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.902919258904767e-06,
+      "loss": 6.1412,
+      "step": 81000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 9.902235591713955e-06,
+      "loss": 6.1468,
+      "step": 81500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 9.901551924523143e-06,
+      "loss": 6.1469,
+      "step": 82000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.90086825733233e-06,
+      "loss": 6.1449,
+      "step": 82500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 9.900184590141519e-06,
+      "loss": 6.1408,
+      "step": 83000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 9.899500922950709e-06,
+      "loss": 6.1439,
+      "step": 83500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.898817255759897e-06,
+      "loss": 6.1434,
+      "step": 84000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 9.898133588569086e-06,
+      "loss": 6.1408,
+      "step": 84500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 9.897449921378274e-06,
+      "loss": 6.1368,
+      "step": 85000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.896766254187462e-06,
+      "loss": 6.1368,
+      "step": 85500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 9.89608258699665e-06,
+      "loss": 6.1394,
+      "step": 86000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 9.89539891980584e-06,
+      "loss": 6.1349,
+      "step": 86500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.894715252615028e-06,
+      "loss": 6.1323,
+      "step": 87000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 9.894031585424216e-06,
+      "loss": 6.1318,
+      "step": 87500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 9.893347918233404e-06,
+      "loss": 6.1316,
+      "step": 88000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 9.892664251042592e-06,
+      "loss": 6.1267,
+      "step": 88500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.89198058385178e-06,
+      "loss": 6.1305,
+      "step": 89000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 9.89129691666097e-06,
+      "loss": 6.1294,
+      "step": 89500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 9.890613249470158e-06,
+      "loss": 6.1258,
+      "step": 90000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.889929582279348e-06,
+      "loss": 6.1244,
+      "step": 90500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 9.889245915088536e-06,
+      "loss": 6.1274,
+      "step": 91000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 9.888562247897724e-06,
+      "loss": 6.1209,
+      "step": 91500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.887878580706912e-06,
+      "loss": 6.1223,
+      "step": 92000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 9.887194913516102e-06,
+      "loss": 6.1252,
+      "step": 92500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 9.88651124632529e-06,
+      "loss": 6.1211,
+      "step": 93000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.885827579134478e-06,
+      "loss": 6.1229,
+      "step": 93500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 9.885143911943666e-06,
+      "loss": 6.1256,
+      "step": 94000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 9.884460244752854e-06,
+      "loss": 6.1242,
+      "step": 94500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.883776577562044e-06,
+      "loss": 6.117,
+      "step": 95000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 9.883092910371232e-06,
+      "loss": 6.1159,
+      "step": 95500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 9.882409243180422e-06,
+      "loss": 6.1242,
+      "step": 96000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.88172557598961e-06,
+      "loss": 6.1178,
+      "step": 96500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 9.881041908798798e-06,
+      "loss": 6.1119,
+      "step": 97000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 9.880358241607986e-06,
+      "loss": 6.1178,
+      "step": 97500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.879674574417174e-06,
+      "loss": 6.1096,
+      "step": 98000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 9.878990907226364e-06,
+      "loss": 6.1132,
+      "step": 98500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 9.878307240035552e-06,
+      "loss": 6.1174,
+      "step": 99000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 9.87762357284474e-06,
+      "loss": 6.1139,
+      "step": 99500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.876939905653928e-06,
+      "loss": 6.1105,
+      "step": 100000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 9.876256238463117e-06,
+      "loss": 6.1031,
+      "step": 100500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 9.875572571272305e-06,
+      "loss": 6.1064,
+      "step": 101000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.874888904081494e-06,
+      "loss": 6.1069,
+      "step": 101500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 9.874205236890683e-06,
+      "loss": 6.1118,
+      "step": 102000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 9.873521569699871e-06,
+      "loss": 6.1089,
+      "step": 102500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.87283790250906e-06,
+      "loss": 6.1034,
+      "step": 103000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 9.872154235318247e-06,
+      "loss": 6.1012,
+      "step": 103500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 9.871470568127435e-06,
+      "loss": 6.1033,
+      "step": 104000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.870786900936625e-06,
+      "loss": 6.1016,
+      "step": 104500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 9.870103233745813e-06,
+      "loss": 6.102,
+      "step": 105000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 9.869419566555001e-06,
+      "loss": 6.1053,
+      "step": 105500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.86873589936419e-06,
+      "loss": 6.103,
+      "step": 106000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 9.868052232173379e-06,
+      "loss": 6.102,
+      "step": 106500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 9.867368564982567e-06,
+      "loss": 6.099,
+      "step": 107000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.866684897791757e-06,
+      "loss": 6.098,
+      "step": 107500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 9.866001230600945e-06,
+      "loss": 6.0973,
+      "step": 108000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 9.865317563410133e-06,
+      "loss": 6.095,
+      "step": 108500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 9.864633896219321e-06,
+      "loss": 6.0967,
+      "step": 109000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.863950229028509e-06,
+      "loss": 6.0887,
+      "step": 109500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.863266561837697e-06,
+      "loss": 6.0951,
+      "step": 110000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.862582894646887e-06,
+      "loss": 6.0893,
+      "step": 110500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.861899227456075e-06,
+      "loss": 6.0944,
+      "step": 111000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.861215560265263e-06,
+      "loss": 6.089,
+      "step": 111500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.860531893074453e-06,
+      "loss": 6.0884,
+      "step": 112000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.85984822588364e-06,
+      "loss": 6.0915,
+      "step": 112500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.859164558692829e-06,
+      "loss": 6.0964,
+      "step": 113000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.858480891502019e-06,
+      "loss": 6.0891,
+      "step": 113500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857797224311207e-06,
+      "loss": 6.0958,
+      "step": 114000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.857113557120395e-06,
+      "loss": 6.0937,
+      "step": 114500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.856429889929583e-06,
+      "loss": 6.0862,
+      "step": 115000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.85574622273877e-06,
+      "loss": 6.0882,
+      "step": 115500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.855062555547959e-06,
+      "loss": 6.0873,
+      "step": 116000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.854378888357149e-06,
+      "loss": 6.0918,
+      "step": 116500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.853695221166337e-06,
+      "loss": 6.0921,
+      "step": 117000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.853011553975526e-06,
+      "loss": 6.0874,
+      "step": 117500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.852327886784714e-06,
+      "loss": 6.0852,
+      "step": 118000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.851644219593902e-06,
+      "loss": 6.0852,
+      "step": 118500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.85096055240309e-06,
+      "loss": 6.0831,
+      "step": 119000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.85027688521228e-06,
+      "loss": 6.0789,
+      "step": 119500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.849593218021468e-06,
+      "loss": 6.0874,
+      "step": 120000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.848909550830656e-06,
+      "loss": 6.0796,
+      "step": 120500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 9.848225883639844e-06,
+      "loss": 6.0862,
+      "step": 121000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 9.847542216449032e-06,
+      "loss": 6.0817,
+      "step": 121500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.846858549258222e-06,
+      "loss": 6.0776,
+      "step": 122000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 9.84617488206741e-06,
+      "loss": 6.0797,
+      "step": 122500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 9.845491214876598e-06,
+      "loss": 6.0764,
+      "step": 123000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.844807547685788e-06,
+      "loss": 6.0812,
+      "step": 123500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 9.844123880494976e-06,
+      "loss": 6.0722,
+      "step": 124000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 9.843440213304164e-06,
+      "loss": 6.0762,
+      "step": 124500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.842756546113352e-06,
+      "loss": 6.0715,
+      "step": 125000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 9.842072878922542e-06,
+      "loss": 6.0739,
+      "step": 125500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 9.84138921173173e-06,
+      "loss": 6.0718,
+      "step": 126000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.840705544540918e-06,
+      "loss": 6.073,
+      "step": 126500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 9.840021877350106e-06,
+      "loss": 6.0743,
+      "step": 127000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 9.839338210159294e-06,
+      "loss": 6.0749,
+      "step": 127500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.838654542968484e-06,
+      "loss": 6.0711,
+      "step": 128000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 9.837970875777672e-06,
+      "loss": 6.0714,
+      "step": 128500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 9.837287208586862e-06,
+      "loss": 6.0682,
+      "step": 129000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 9.83660354139605e-06,
+      "loss": 6.073,
+      "step": 129500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.835919874205238e-06,
+      "loss": 6.0705,
+      "step": 130000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 9.835236207014426e-06,
+      "loss": 6.0636,
+      "step": 130500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 9.834552539823614e-06,
+      "loss": 6.069,
+      "step": 131000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.833868872632804e-06,
+      "loss": 6.063,
+      "step": 131500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 9.833185205441992e-06,
+      "loss": 6.0649,
+      "step": 132000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 9.83250153825118e-06,
+      "loss": 6.0674,
+      "step": 132500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.831817871060368e-06,
+      "loss": 6.068,
+      "step": 133000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 9.831134203869558e-06,
+      "loss": 6.0669,
+      "step": 133500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 9.830450536678746e-06,
+      "loss": 6.0622,
+      "step": 134000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.829766869487934e-06,
+      "loss": 6.0675,
+      "step": 134500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 9.829083202297123e-06,
+      "loss": 6.0659,
+      "step": 135000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 9.828399535106311e-06,
+      "loss": 6.0638,
+      "step": 135500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.8277158679155e-06,
+      "loss": 6.0653,
+      "step": 136000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 9.827032200724688e-06,
+      "loss": 6.0661,
+      "step": 136500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 9.826348533533876e-06,
+      "loss": 6.0637,
+      "step": 137000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.825664866343065e-06,
+      "loss": 6.0626,
+      "step": 137500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 9.824981199152253e-06,
+      "loss": 6.061,
+      "step": 138000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 9.824297531961441e-06,
+      "loss": 6.0585,
+      "step": 138500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.82361386477063e-06,
+      "loss": 6.0645,
+      "step": 139000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 9.82293019757982e-06,
+      "loss": 6.0598,
+      "step": 139500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 9.822246530389007e-06,
+      "loss": 6.0563,
+      "step": 140000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 9.821562863198197e-06,
+      "loss": 6.0555,
+      "step": 140500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.820879196007385e-06,
+      "loss": 6.0561,
+      "step": 141000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 9.820195528816573e-06,
+      "loss": 6.0563,
+      "step": 141500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 9.819511861625761e-06,
+      "loss": 6.0568,
+      "step": 142000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.81882819443495e-06,
+      "loss": 6.0574,
+      "step": 142500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.818144527244139e-06,
+      "loss": 6.0544,
+      "step": 143000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.817460860053327e-06,
+      "loss": 6.0542,
+      "step": 143500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.816777192862515e-06,
+      "loss": 6.0555,
+      "step": 144000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 9.816093525671703e-06,
+      "loss": 6.0507,
+      "step": 144500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 9.815409858480893e-06,
+      "loss": 6.054,
+      "step": 145000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.814726191290081e-06,
+      "loss": 6.051,
+      "step": 145500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 9.81404252409927e-06,
+      "loss": 6.0523,
+      "step": 146000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 9.813358856908459e-06,
+      "loss": 6.0495,
+      "step": 146500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.812675189717647e-06,
+      "loss": 6.0531,
+      "step": 147000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 9.811991522526835e-06,
+      "loss": 6.049,
+      "step": 147500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 9.811307855336023e-06,
+      "loss": 6.0523,
+      "step": 148000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.810624188145211e-06,
+      "loss": 6.0429,
+      "step": 148500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 9.8099405209544e-06,
+      "loss": 6.051,
+      "step": 149000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 9.809256853763589e-06,
+      "loss": 6.0496,
+      "step": 149500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 9.808573186572777e-06,
+      "loss": 6.0468,
+      "step": 150000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.807889519381966e-06,
+      "loss": 6.0447,
+      "step": 150500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 9.807205852191155e-06,
+      "loss": 6.0483,
+      "step": 151000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 9.806522185000343e-06,
+      "loss": 6.0441,
+      "step": 151500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.805838517809532e-06,
+      "loss": 6.046,
+      "step": 152000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 9.80515485061872e-06,
+      "loss": 6.043,
+      "step": 152500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 9.804471183427908e-06,
+      "loss": 6.0462,
+      "step": 153000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.803787516237096e-06,
+      "loss": 6.0445,
+      "step": 153500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 9.803103849046285e-06,
+      "loss": 6.0442,
+      "step": 154000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 9.802420181855473e-06,
+      "loss": 6.0506,
+      "step": 154500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.801736514664662e-06,
+      "loss": 6.0401,
+      "step": 155000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 9.80105284747385e-06,
+      "loss": 6.0402,
+      "step": 155500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 9.800369180283038e-06,
+      "loss": 6.0401,
+      "step": 156000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.799685513092228e-06,
+      "loss": 6.043,
+      "step": 156500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 9.799001845901416e-06,
+      "loss": 6.0436,
+      "step": 157000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 9.798318178710604e-06,
+      "loss": 6.0428,
+      "step": 157500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.797634511519794e-06,
+      "loss": 6.0401,
+      "step": 158000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 9.796950844328982e-06,
+      "loss": 6.0455,
+      "step": 158500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 9.79626717713817e-06,
+      "loss": 6.0377,
+      "step": 159000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.795583509947358e-06,
+      "loss": 6.0403,
+      "step": 159500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 9.794899842756546e-06,
+      "loss": 6.041,
+      "step": 160000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 9.794216175565734e-06,
+      "loss": 6.0443,
+      "step": 160500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 9.793532508374924e-06,
+      "loss": 6.0353,
+      "step": 161000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.792848841184112e-06,
+      "loss": 6.0365,
+      "step": 161500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 9.792165173993302e-06,
+      "loss": 6.039,
+      "step": 162000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 9.79148150680249e-06,
+      "loss": 6.0393,
+      "step": 162500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.790797839611678e-06,
+      "loss": 6.0404,
+      "step": 163000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 9.790114172420866e-06,
+      "loss": 6.0363,
+      "step": 163500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 9.789430505230056e-06,
+      "loss": 6.0369,
+      "step": 164000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.788746838039244e-06,
+      "loss": 6.0319,
+      "step": 164500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 9.788063170848432e-06,
+      "loss": 6.0306,
+      "step": 165000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 9.78737950365762e-06,
+      "loss": 6.0374,
+      "step": 165500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.786695836466808e-06,
+      "loss": 6.0333,
+      "step": 166000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 9.786012169275998e-06,
+      "loss": 6.0343,
+      "step": 166500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 9.785328502085186e-06,
+      "loss": 6.0387,
+      "step": 167000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.784644834894374e-06,
+      "loss": 6.0325,
+      "step": 167500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 9.783961167703563e-06,
+      "loss": 6.0326,
+      "step": 168000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 9.783277500512752e-06,
+      "loss": 6.0336,
+      "step": 168500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.78259383332194e-06,
+      "loss": 6.0351,
+      "step": 169000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 9.781910166131128e-06,
+      "loss": 6.0334,
+      "step": 169500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 9.781226498940317e-06,
+      "loss": 6.0268,
+      "step": 170000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.780542831749505e-06,
+      "loss": 6.0304,
+      "step": 170500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 9.779859164558693e-06,
+      "loss": 6.0321,
+      "step": 171000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 9.779175497367882e-06,
+      "loss": 6.0386,
+      "step": 171500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 9.77849183017707e-06,
+      "loss": 6.0302,
+      "step": 172000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.77780816298626e-06,
+      "loss": 6.0286,
+      "step": 172500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 9.777124495795447e-06,
+      "loss": 6.0241,
+      "step": 173000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 9.776440828604637e-06,
+      "loss": 6.0306,
+      "step": 173500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.775757161413825e-06,
+      "loss": 6.0282,
+      "step": 174000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 9.775073494223013e-06,
+      "loss": 6.027,
+      "step": 174500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 9.774389827032201e-06,
+      "loss": 6.0265,
+      "step": 175000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.77370615984139e-06,
+      "loss": 6.028,
+      "step": 175500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.773022492650579e-06,
+      "loss": 6.0279,
+      "step": 176000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.772338825459767e-06,
+      "loss": 6.027,
+      "step": 176500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.771655158268955e-06,
+      "loss": 6.0276,
+      "step": 177000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.770971491078143e-06,
+      "loss": 6.0233,
+      "step": 177500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.770287823887333e-06,
+      "loss": 6.0254,
+      "step": 178000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.769604156696521e-06,
+      "loss": 6.0261,
+      "step": 178500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.76892048950571e-06,
+      "loss": 6.0231,
+      "step": 179000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.768236822314899e-06,
+      "loss": 6.02,
+      "step": 179500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.767553155124087e-06,
+      "loss": 6.0264,
+      "step": 180000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.766869487933275e-06,
+      "loss": 6.0207,
+      "step": 180500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 9.766185820742463e-06,
+      "loss": 6.022,
+      "step": 181000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 9.765502153551651e-06,
+      "loss": 6.0249,
+      "step": 181500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.76481848636084e-06,
+      "loss": 6.0195,
+      "step": 182000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 9.764134819170029e-06,
+      "loss": 6.0208,
+      "step": 182500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.763451151979217e-06,
+      "loss": 6.0215,
+      "step": 183000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.762767484788407e-06,
+      "loss": 6.0199,
+      "step": 183500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.762083817597595e-06,
+      "loss": 6.0222,
+      "step": 184000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.761400150406783e-06,
+      "loss": 6.0136,
+      "step": 184500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.760716483215972e-06,
+      "loss": 6.0168,
+      "step": 185000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.76003281602516e-06,
+      "loss": 6.024,
+      "step": 185500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.759349148834348e-06,
+      "loss": 6.0185,
+      "step": 186000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.758665481643537e-06,
+      "loss": 6.0148,
+      "step": 186500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.757981814452725e-06,
+      "loss": 6.0156,
+      "step": 187000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.757298147261913e-06,
+      "loss": 6.0152,
+      "step": 187500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.756614480071102e-06,
+      "loss": 6.0192,
+      "step": 188000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.75593081288029e-06,
+      "loss": 6.0148,
+      "step": 188500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.755247145689478e-06,
+      "loss": 6.0156,
+      "step": 189000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.754563478498668e-06,
+      "loss": 6.0145,
+      "step": 189500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.753879811307856e-06,
+      "loss": 6.0147,
+      "step": 190000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.753196144117044e-06,
+      "loss": 6.0128,
+      "step": 190500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.752512476926234e-06,
+      "loss": 6.0167,
+      "step": 191000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.751828809735422e-06,
+      "loss": 6.0144,
+      "step": 191500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.75114514254461e-06,
+      "loss": 6.0158,
+      "step": 192000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.750461475353798e-06,
+      "loss": 6.0203,
+      "step": 192500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.749777808162986e-06,
+      "loss": 6.0145,
+      "step": 193000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.749094140972174e-06,
+      "loss": 6.0072,
+      "step": 193500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.748410473781364e-06,
+      "loss": 6.0144,
+      "step": 194000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.747726806590552e-06,
+      "loss": 6.0079,
+      "step": 194500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.747043139399742e-06,
+      "loss": 6.0079,
+      "step": 195000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.74635947220893e-06,
+      "loss": 6.0134,
+      "step": 195500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.745675805018118e-06,
+      "loss": 6.009,
+      "step": 196000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.744992137827306e-06,
+      "loss": 6.0119,
+      "step": 196500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.744308470636496e-06,
+      "loss": 6.0123,
+      "step": 197000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.743624803445684e-06,
+      "loss": 6.0098,
+      "step": 197500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.742941136254872e-06,
+      "loss": 6.013,
+      "step": 198000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.74225746906406e-06,
+      "loss": 6.0097,
+      "step": 198500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.741573801873248e-06,
+      "loss": 6.0071,
+      "step": 199000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.740890134682438e-06,
+      "loss": 6.0137,
+      "step": 199500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.740206467491626e-06,
+      "loss": 6.0097,
+      "step": 200000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.739522800300815e-06,
+      "loss": 6.0073,
+      "step": 200500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.738839133110004e-06,
+      "loss": 6.0059,
+      "step": 201000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.738155465919192e-06,
+      "loss": 6.0049,
+      "step": 201500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 9.73747179872838e-06,
+      "loss": 6.0102,
+      "step": 202000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.736788131537568e-06,
+      "loss": 6.0021,
+      "step": 202500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 9.736104464346757e-06,
+      "loss": 6.0092,
+      "step": 203000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 9.735420797155945e-06,
+      "loss": 6.0067,
+      "step": 203500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.734737129965134e-06,
+      "loss": 6.002,
+      "step": 204000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 9.734053462774322e-06,
+      "loss": 6.0097,
+      "step": 204500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 9.73336979558351e-06,
+      "loss": 6.0039,
+      "step": 205000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.7326861283927e-06,
+      "loss": 6.0069,
+      "step": 205500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 9.732002461201887e-06,
+      "loss": 6.001,
+      "step": 206000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 9.731318794011077e-06,
+      "loss": 6.0056,
+      "step": 206500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.730635126820265e-06,
+      "loss": 6.0081,
+      "step": 207000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 9.729951459629453e-06,
+      "loss": 6.006,
+      "step": 207500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 9.729267792438641e-06,
+      "loss": 5.9988,
+      "step": 208000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.72858412524783e-06,
+      "loss": 6.0057,
+      "step": 208500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 9.727900458057019e-06,
+      "loss": 5.9984,
+      "step": 209000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 9.727216790866207e-06,
+      "loss": 6.0002,
+      "step": 209500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.726533123675395e-06,
+      "loss": 5.9995,
+      "step": 210000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 9.725849456484583e-06,
+      "loss": 6.0047,
+      "step": 210500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 9.725165789293773e-06,
+      "loss": 5.9944,
+      "step": 211000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.724482122102961e-06,
+      "loss": 6.0034,
+      "step": 211500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 9.72379845491215e-06,
+      "loss": 5.9981,
+      "step": 212000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 9.723114787721339e-06,
+      "loss": 5.9984,
+      "step": 212500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 9.722431120530527e-06,
+      "loss": 5.9974,
+      "step": 213000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.721747453339715e-06,
+      "loss": 5.9954,
+      "step": 213500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 9.721063786148903e-06,
+      "loss": 5.9999,
+      "step": 214000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 9.720380118958091e-06,
+      "loss": 6.0039,
+      "step": 214500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.71969645176728e-06,
+      "loss": 5.9961,
+      "step": 215000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.719012784576469e-06,
+      "loss": 5.9985,
+      "step": 215500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.718329117385657e-06,
+      "loss": 6.0006,
+      "step": 216000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.717645450194847e-06,
+      "loss": 5.995,
+      "step": 216500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 9.716961783004035e-06,
+      "loss": 5.9959,
+      "step": 217000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 9.716278115813223e-06,
+      "loss": 5.9963,
+      "step": 217500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.715594448622412e-06,
+      "loss": 5.995,
+      "step": 218000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 9.7149107814316e-06,
+      "loss": 5.9952,
+      "step": 218500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 9.714227114240789e-06,
+      "loss": 5.9966,
+      "step": 219000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.713543447049977e-06,
+      "loss": 5.9929,
+      "step": 219500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 9.712859779859165e-06,
+      "loss": 5.9925,
+      "step": 220000
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 9.712176112668353e-06,
+      "loss": 5.9991,
+      "step": 220500
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.711492445477542e-06,
+      "loss": 5.9928,
+      "step": 221000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 9.71080877828673e-06,
+      "loss": 5.996,
+      "step": 221500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 9.710125111095919e-06,
+      "loss": 5.9966,
+      "step": 222000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.709441443905108e-06,
+      "loss": 5.9935,
+      "step": 222500
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 9.708757776714296e-06,
+      "loss": 5.9967,
+      "step": 223000
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 9.708074109523484e-06,
+      "loss": 5.9913,
+      "step": 223500
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 9.707390442332674e-06,
+      "loss": 5.9909,
+      "step": 224000
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.706706775141862e-06,
+      "loss": 5.9941,
+      "step": 224500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 9.70602310795105e-06,
+      "loss": 5.9972,
+      "step": 225000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 9.705339440760238e-06,
+      "loss": 5.9925,
+      "step": 225500
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.704655773569426e-06,
+      "loss": 5.9891,
+      "step": 226000
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 9.703972106378614e-06,
+      "loss": 5.9851,
+      "step": 226500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 9.703288439187804e-06,
+      "loss": 5.9911,
+      "step": 227000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.702604771996992e-06,
+      "loss": 5.9884,
+      "step": 227500
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 9.701921104806182e-06,
+      "loss": 5.9883,
+      "step": 228000
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 9.70123743761537e-06,
+      "loss": 5.9901,
+      "step": 228500
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.700553770424558e-06,
+      "loss": 5.983,
+      "step": 229000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 9.699870103233746e-06,
+      "loss": 5.9778,
+      "step": 229500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 9.699186436042936e-06,
+      "loss": 5.9698,
+      "step": 230000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.698502768852124e-06,
+      "loss": 5.9746,
+      "step": 230500
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 9.697819101661312e-06,
+      "loss": 5.972,
+      "step": 231000
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 9.6971354344705e-06,
+      "loss": 5.9724,
+      "step": 231500
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.696451767279688e-06,
+      "loss": 5.968,
+      "step": 232000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 9.695768100088878e-06,
+      "loss": 5.9622,
+      "step": 232500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 9.695084432898066e-06,
+      "loss": 5.9568,
+      "step": 233000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 9.694400765707256e-06,
+      "loss": 5.9537,
+      "step": 233500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.693717098516444e-06,
+      "loss": 5.9481,
+      "step": 234000
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 9.693033431325632e-06,
+      "loss": 5.9461,
+      "step": 234500
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 9.69234976413482e-06,
+      "loss": 5.9377,
+      "step": 235000
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.691666096944008e-06,
+      "loss": 5.9338,
+      "step": 235500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 9.690982429753198e-06,
+      "loss": 5.9283,
+      "step": 236000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 9.690298762562386e-06,
+      "loss": 5.922,
+      "step": 236500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.689615095371574e-06,
+      "loss": 5.9186,
+      "step": 237000
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 9.688931428180762e-06,
+      "loss": 5.9195,
+      "step": 237500
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 9.68824776098995e-06,
+      "loss": 5.9136,
+      "step": 238000
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.68756409379914e-06,
+      "loss": 5.9127,
+      "step": 238500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 9.686880426608328e-06,
+      "loss": 5.9042,
+      "step": 239000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 9.686196759417517e-06,
+      "loss": 5.8971,
+      "step": 239500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.685513092226705e-06,
+      "loss": 5.8954,
+      "step": 240000
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 9.684829425035893e-06,
+      "loss": 5.891,
+      "step": 240500
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 9.684145757845081e-06,
+      "loss": 5.8882,
+      "step": 241000
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.68346209065427e-06,
+      "loss": 5.8858,
+      "step": 241500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 9.68277842346346e-06,
+      "loss": 5.88,
+      "step": 242000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 9.682094756272647e-06,
+      "loss": 5.8769,
+      "step": 242500
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.681411089081835e-06,
+      "loss": 5.8778,
+      "step": 243000
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 9.680727421891023e-06,
+      "loss": 5.8688,
+      "step": 243500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 9.680043754700213e-06,
+      "loss": 5.8709,
+      "step": 244000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 9.679360087509401e-06,
+      "loss": 5.8704,
+      "step": 244500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.678676420318591e-06,
+      "loss": 5.8628,
+      "step": 245000
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 9.677992753127779e-06,
+      "loss": 5.8591,
+      "step": 245500
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 9.677309085936967e-06,
+      "loss": 5.8611,
+      "step": 246000
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.676625418746155e-06,
+      "loss": 5.8526,
+      "step": 246500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 9.675941751555343e-06,
+      "loss": 5.8551,
+      "step": 247000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 9.675258084364531e-06,
+      "loss": 5.8503,
+      "step": 247500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.674574417173721e-06,
+      "loss": 5.8461,
+      "step": 248000
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 9.673890749982909e-06,
+      "loss": 5.8496,
+      "step": 248500
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 9.673207082792097e-06,
+      "loss": 5.85,
+      "step": 249000
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.672523415601287e-06,
+      "loss": 5.8442,
+      "step": 249500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 9.671839748410475e-06,
+      "loss": 5.8379,
+      "step": 250000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 9.671156081219663e-06,
+      "loss": 5.8439,
+      "step": 250500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.670472414028853e-06,
+      "loss": 5.8339,
+      "step": 251000
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 9.66978874683804e-06,
+      "loss": 5.8348,
+      "step": 251500
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 9.669105079647229e-06,
+      "loss": 5.8319,
+      "step": 252000
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.668421412456417e-06,
+      "loss": 5.831,
+      "step": 252500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 9.667737745265605e-06,
+      "loss": 5.8335,
+      "step": 253000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 9.667054078074793e-06,
+      "loss": 5.8254,
+      "step": 253500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 9.666370410883983e-06,
+      "loss": 5.8275,
+      "step": 254000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.66568674369317e-06,
+      "loss": 5.8246,
+      "step": 254500
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 9.665003076502359e-06,
+      "loss": 5.824,
+      "step": 255000
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 9.664319409311548e-06,
+      "loss": 5.8221,
+      "step": 255500
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.663635742120736e-06,
+      "loss": 5.8144,
+      "step": 256000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 9.662952074929925e-06,
+      "loss": 5.8185,
+      "step": 256500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 9.662268407739114e-06,
+      "loss": 5.8122,
+      "step": 257000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.661584740548302e-06,
+      "loss": 5.8105,
+      "step": 257500
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 9.66090107335749e-06,
+      "loss": 5.8134,
+      "step": 258000
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 9.660217406166678e-06,
+      "loss": 5.8112,
+      "step": 258500
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 9.659533738975866e-06,
+      "loss": 5.8045,
+      "step": 259000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 9.658850071785054e-06,
+      "loss": 5.8051,
+      "step": 259500
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 9.658166404594244e-06,
+      "loss": 5.8001,
+      "step": 260000
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.657482737403432e-06,
+      "loss": 5.7982,
+      "step": 260500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 9.656799070212622e-06,
+      "loss": 5.8032,
+      "step": 261000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 9.65611540302181e-06,
+      "loss": 5.7913,
+      "step": 261500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.655431735830998e-06,
+      "loss": 5.795,
+      "step": 262000
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 9.654748068640186e-06,
+      "loss": 5.7876,
+      "step": 262500
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 9.654064401449376e-06,
+      "loss": 5.7882,
+      "step": 263000
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.653380734258564e-06,
+      "loss": 5.7874,
+      "step": 263500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 9.652697067067752e-06,
+      "loss": 5.791,
+      "step": 264000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 9.65201339987694e-06,
+      "loss": 5.7845,
+      "step": 264500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 9.651329732686128e-06,
+      "loss": 5.7827,
+      "step": 265000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.650646065495318e-06,
+      "loss": 5.7784,
+      "step": 265500
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 9.649962398304506e-06,
+      "loss": 5.7842,
+      "step": 266000
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 9.649278731113696e-06,
+      "loss": 5.7746,
+      "step": 266500
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.648595063922884e-06,
+      "loss": 5.7766,
+      "step": 267000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 9.647911396732072e-06,
+      "loss": 5.7765,
+      "step": 267500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 9.64722772954126e-06,
+      "loss": 5.7691,
+      "step": 268000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.646544062350448e-06,
+      "loss": 5.7707,
+      "step": 268500
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 9.645860395159638e-06,
+      "loss": 5.7689,
+      "step": 269000
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 9.645176727968826e-06,
+      "loss": 5.7675,
+      "step": 269500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.644493060778014e-06,
+      "loss": 5.7628,
+      "step": 270000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 9.643809393587202e-06,
+      "loss": 5.7622,
+      "step": 270500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 9.643125726396392e-06,
+      "loss": 5.7646,
+      "step": 271000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.64244205920558e-06,
+      "loss": 5.7587,
+      "step": 271500
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 9.641758392014768e-06,
+      "loss": 5.7586,
+      "step": 272000
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 9.641074724823957e-06,
+      "loss": 5.7537,
+      "step": 272500
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.640391057633145e-06,
+      "loss": 5.7577,
+      "step": 273000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 9.639707390442333e-06,
+      "loss": 5.754,
+      "step": 273500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 9.639023723251521e-06,
+      "loss": 5.7523,
+      "step": 274000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 9.63834005606071e-06,
+      "loss": 5.7497,
+      "step": 274500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.6376563888699e-06,
+      "loss": 5.7467,
+      "step": 275000
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 9.636972721679087e-06,
+      "loss": 5.7459,
+      "step": 275500
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 9.636289054488275e-06,
+      "loss": 5.741,
+      "step": 276000
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.635605387297463e-06,
+      "loss": 5.7404,
+      "step": 276500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 9.634921720106653e-06,
+      "loss": 5.7435,
+      "step": 277000
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 9.634238052915841e-06,
+      "loss": 5.7428,
+      "step": 277500
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.633554385725031e-06,
+      "loss": 5.7324,
+      "step": 278000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 9.632870718534219e-06,
+      "loss": 5.7422,
+      "step": 278500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 9.632187051343407e-06,
+      "loss": 5.7338,
+      "step": 279000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.631503384152595e-06,
+      "loss": 5.7366,
+      "step": 279500
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 9.630819716961783e-06,
+      "loss": 5.7286,
+      "step": 280000
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 9.630136049770973e-06,
+      "loss": 5.7248,
+      "step": 280500
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.629452382580161e-06,
+      "loss": 5.7239,
+      "step": 281000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 9.628768715389349e-06,
+      "loss": 5.7197,
+      "step": 281500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 9.628085048198537e-06,
+      "loss": 5.7239,
+      "step": 282000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.627401381007727e-06,
+      "loss": 5.7284,
+      "step": 282500
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 9.626717713816915e-06,
+      "loss": 5.724,
+      "step": 283000
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 9.626034046626103e-06,
+      "loss": 5.7228,
+      "step": 283500
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.625350379435293e-06,
+      "loss": 5.7175,
+      "step": 284000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 9.62466671224448e-06,
+      "loss": 5.7125,
+      "step": 284500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 9.623983045053669e-06,
+      "loss": 5.7142,
+      "step": 285000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 9.623299377862857e-06,
+      "loss": 5.7114,
+      "step": 285500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.622615710672045e-06,
+      "loss": 5.7099,
+      "step": 286000
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 9.621932043481235e-06,
+      "loss": 5.7053,
+      "step": 286500
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 9.621248376290423e-06,
+      "loss": 5.7062,
+      "step": 287000
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.62056470909961e-06,
+      "loss": 5.7075,
+      "step": 287500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 9.619881041908799e-06,
+      "loss": 5.705,
+      "step": 288000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 9.619197374717988e-06,
+      "loss": 5.7016,
+      "step": 288500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.618513707527177e-06,
+      "loss": 5.7028,
+      "step": 289000
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 9.617830040336366e-06,
+      "loss": 5.7015,
+      "step": 289500
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 9.617146373145554e-06,
+      "loss": 5.6953,
+      "step": 290000
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.616462705954742e-06,
+      "loss": 5.6988,
+      "step": 290500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 9.61577903876393e-06,
+      "loss": 5.6945,
+      "step": 291000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 9.615095371573118e-06,
+      "loss": 5.6901,
+      "step": 291500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.614411704382307e-06,
+      "loss": 5.6891,
+      "step": 292000
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 9.613728037191496e-06,
+      "loss": 5.6857,
+      "step": 292500
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 9.613044370000684e-06,
+      "loss": 5.6851,
+      "step": 293000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.612360702809872e-06,
+      "loss": 5.6834,
+      "step": 293500
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 9.611677035619062e-06,
+      "loss": 5.6859,
+      "step": 294000
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 9.61099336842825e-06,
+      "loss": 5.6802,
+      "step": 294500
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.610309701237438e-06,
+      "loss": 5.6789,
+      "step": 295000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 9.609626034046628e-06,
+      "loss": 5.679,
+      "step": 295500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 9.608942366855816e-06,
+      "loss": 5.6785,
+      "step": 296000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 9.608258699665004e-06,
+      "loss": 5.6731,
+      "step": 296500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.607575032474192e-06,
+      "loss": 5.6705,
+      "step": 297000
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 9.60689136528338e-06,
+      "loss": 5.6697,
+      "step": 297500
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 9.606207698092568e-06,
+      "loss": 5.6652,
+      "step": 298000
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.605524030901758e-06,
+      "loss": 5.6702,
+      "step": 298500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 9.604840363710946e-06,
+      "loss": 5.6662,
+      "step": 299000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 9.604156696520136e-06,
+      "loss": 5.663,
+      "step": 299500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.603473029329324e-06,
+      "loss": 5.6624,
+      "step": 300000
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 9.602789362138512e-06,
+      "loss": 5.6622,
+      "step": 300500
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 9.6021056949477e-06,
+      "loss": 5.6552,
+      "step": 301000
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.60142202775689e-06,
+      "loss": 5.6541,
+      "step": 301500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 9.600738360566078e-06,
+      "loss": 5.65,
+      "step": 302000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 9.600054693375266e-06,
+      "loss": 5.6477,
+      "step": 302500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.599371026184454e-06,
+      "loss": 5.6489,
+      "step": 303000
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 9.598687358993642e-06,
+      "loss": 5.6472,
+      "step": 303500
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 9.598003691802832e-06,
+      "loss": 5.6429,
+      "step": 304000
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.59732002461202e-06,
+      "loss": 5.6446,
+      "step": 304500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 9.596636357421208e-06,
+      "loss": 5.6393,
+      "step": 305000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 9.595952690230397e-06,
+      "loss": 5.6387,
+      "step": 305500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 9.595269023039585e-06,
+      "loss": 5.636,
+      "step": 306000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.594585355848774e-06,
+      "loss": 5.636,
+      "step": 306500
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 9.593901688657962e-06,
+      "loss": 5.6275,
+      "step": 307000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 9.593218021467151e-06,
+      "loss": 5.6309,
+      "step": 307500
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.59253435427634e-06,
+      "loss": 5.6288,
+      "step": 308000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 9.591850687085527e-06,
+      "loss": 5.6269,
+      "step": 308500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 9.591167019894715e-06,
+      "loss": 5.6268,
+      "step": 309000
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.590483352703904e-06,
+      "loss": 5.6235,
+      "step": 309500
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 9.589799685513093e-06,
+      "loss": 5.6234,
+      "step": 310000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 9.589116018322281e-06,
+      "loss": 5.6159,
+      "step": 310500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.588432351131471e-06,
+      "loss": 5.6195,
+      "step": 311000
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 9.587748683940659e-06,
+      "loss": 5.6146,
+      "step": 311500
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 9.587065016749847e-06,
+      "loss": 5.6144,
+      "step": 312000
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.586381349559035e-06,
+      "loss": 5.6068,
+      "step": 312500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 9.585697682368223e-06,
+      "loss": 5.6054,
+      "step": 313000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 9.585014015177413e-06,
+      "loss": 5.6072,
+      "step": 313500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.584330347986601e-06,
+      "loss": 5.6092,
+      "step": 314000
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 9.583646680795789e-06,
+      "loss": 5.6032,
+      "step": 314500
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.582963013604977e-06,
+      "loss": 5.5979,
+      "step": 315000
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.582279346414167e-06,
+      "loss": 5.5999,
+      "step": 315500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.581595679223355e-06,
+      "loss": 5.5998,
+      "step": 316000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.580912012032545e-06,
+      "loss": 5.5968,
+      "step": 316500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.580228344841733e-06,
+      "loss": 5.5962,
+      "step": 317000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.57954467765092e-06,
+      "loss": 5.5882,
+      "step": 317500
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.578861010460109e-06,
+      "loss": 5.5934,
+      "step": 318000
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 9.578177343269297e-06,
+      "loss": 5.5827,
+      "step": 318500
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.577493676078485e-06,
+      "loss": 5.5848,
+      "step": 319000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 9.576810008887675e-06,
+      "loss": 5.5834,
+      "step": 319500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 9.576126341696863e-06,
+      "loss": 5.5776,
+      "step": 320000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.57544267450605e-06,
+      "loss": 5.5761,
+      "step": 320500
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 9.574759007315239e-06,
+      "loss": 5.5744,
+      "step": 321000
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 9.574075340124429e-06,
+      "loss": 5.5681,
+      "step": 321500
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 9.573391672933617e-06,
+      "loss": 5.5701,
+      "step": 322000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 9.572708005742806e-06,
+      "loss": 5.5743,
+      "step": 322500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 9.572024338551994e-06,
+      "loss": 5.5632,
+      "step": 323000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.571340671361182e-06,
+      "loss": 5.5713,
+      "step": 323500
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 9.57065700417037e-06,
+      "loss": 5.5644,
+      "step": 324000
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 9.569973336979559e-06,
+      "loss": 5.5614,
+      "step": 324500
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.569289669788747e-06,
+      "loss": 5.5573,
+      "step": 325000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 9.568606002597936e-06,
+      "loss": 5.5565,
+      "step": 325500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 9.567922335407124e-06,
+      "loss": 5.5542,
+      "step": 326000
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 9.567238668216312e-06,
+      "loss": 5.5458,
+      "step": 326500
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.566555001025502e-06,
+      "loss": 5.5485,
+      "step": 327000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 9.56587133383469e-06,
+      "loss": 5.5499,
+      "step": 327500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 9.565187666643878e-06,
+      "loss": 5.548,
+      "step": 328000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 9.564503999453068e-06,
+      "loss": 5.5448,
+      "step": 328500
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 9.563820332262256e-06,
+      "loss": 5.5397,
+      "step": 329000
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 9.563136665071444e-06,
+      "loss": 5.5373,
+      "step": 329500
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.562452997880632e-06,
+      "loss": 5.5393,
+      "step": 330000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 9.56176933068982e-06,
+      "loss": 5.531,
+      "step": 330500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 9.561085663499008e-06,
+      "loss": 5.5275,
+      "step": 331000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.560401996308198e-06,
+      "loss": 5.5218,
+      "step": 331500
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 9.559718329117386e-06,
+      "loss": 5.5278,
+      "step": 332000
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 9.559034661926576e-06,
+      "loss": 5.5243,
+      "step": 332500
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.558350994735764e-06,
+      "loss": 5.5169,
+      "step": 333000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 9.557667327544952e-06,
+      "loss": 5.5161,
+      "step": 333500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 9.55698366035414e-06,
+      "loss": 5.5092,
+      "step": 334000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.55629999316333e-06,
+      "loss": 5.5116,
+      "step": 334500
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 9.555616325972518e-06,
+      "loss": 5.5049,
+      "step": 335000
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 9.554932658781706e-06,
+      "loss": 5.5104,
+      "step": 335500
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.554248991590894e-06,
+      "loss": 5.5079,
+      "step": 336000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 9.553565324400082e-06,
+      "loss": 5.5062,
+      "step": 336500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 9.552881657209272e-06,
+      "loss": 5.506,
+      "step": 337000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 9.55219799001846e-06,
+      "loss": 5.4969,
+      "step": 337500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.551514322827648e-06,
+      "loss": 5.496,
+      "step": 338000
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 9.550830655636838e-06,
+      "loss": 5.4895,
+      "step": 338500
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 9.550146988446026e-06,
+      "loss": 5.4883,
+      "step": 339000
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.549463321255214e-06,
+      "loss": 5.4851,
+      "step": 339500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 9.548779654064402e-06,
+      "loss": 5.4868,
+      "step": 340000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 9.548095986873591e-06,
+      "loss": 5.4849,
+      "step": 340500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.54741231968278e-06,
+      "loss": 5.4793,
+      "step": 341000
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 9.546728652491968e-06,
+      "loss": 5.4778,
+      "step": 341500
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 9.546044985301156e-06,
+      "loss": 5.4756,
+      "step": 342000
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.545361318110344e-06,
+      "loss": 5.4677,
+      "step": 342500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 9.544677650919533e-06,
+      "loss": 5.4696,
+      "step": 343000
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 9.543993983728721e-06,
+      "loss": 5.4649,
+      "step": 343500
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.543310316537911e-06,
+      "loss": 5.4588,
+      "step": 344000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 9.5426266493471e-06,
+      "loss": 5.4627,
+      "step": 344500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 9.541942982156287e-06,
+      "loss": 5.4648,
+      "step": 345000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.541259314965475e-06,
+      "loss": 5.4562,
+      "step": 345500
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 9.540575647774663e-06,
+      "loss": 5.4581,
+      "step": 346000
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 9.539891980583853e-06,
+      "loss": 5.4463,
+      "step": 346500
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 9.539208313393041e-06,
+      "loss": 5.4502,
+      "step": 347000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 9.53852464620223e-06,
+      "loss": 5.4462,
+      "step": 347500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.537840979011417e-06,
+      "loss": 5.4394,
+      "step": 348000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.537157311820607e-06,
+      "loss": 5.4412,
+      "step": 348500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.536473644629795e-06,
+      "loss": 5.4369,
+      "step": 349000
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.535789977438985e-06,
+      "loss": 5.4318,
+      "step": 349500
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.535106310248173e-06,
+      "loss": 5.4346,
+      "step": 350000
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.534422643057361e-06,
+      "loss": 5.4276,
+      "step": 350500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.533738975866549e-06,
+      "loss": 5.4279,
+      "step": 351000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 9.533055308675737e-06,
+      "loss": 5.4242,
+      "step": 351500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.532371641484925e-06,
+      "loss": 5.4223,
+      "step": 352000
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 9.531687974294115e-06,
+      "loss": 5.4185,
+      "step": 352500
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 9.531004307103303e-06,
+      "loss": 5.4173,
+      "step": 353000
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.530320639912491e-06,
+      "loss": 5.4145,
+      "step": 353500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 9.529636972721679e-06,
+      "loss": 5.4117,
+      "step": 354000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 9.528953305530869e-06,
+      "loss": 5.4088,
+      "step": 354500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.528269638340057e-06,
+      "loss": 5.4058,
+      "step": 355000
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 9.527585971149246e-06,
+      "loss": 5.4068,
+      "step": 355500
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 9.526902303958435e-06,
+      "loss": 5.3917,
+      "step": 356000
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.526218636767623e-06,
+      "loss": 5.395,
+      "step": 356500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 9.52553496957681e-06,
+      "loss": 5.3933,
+      "step": 357000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 9.524851302385999e-06,
+      "loss": 5.3877,
+      "step": 357500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 9.524167635195187e-06,
+      "loss": 5.3837,
+      "step": 358000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.523483968004376e-06,
+      "loss": 5.3847,
+      "step": 358500
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 9.522800300813564e-06,
+      "loss": 5.3795,
+      "step": 359000
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 9.522116633622753e-06,
+      "loss": 5.3741,
+      "step": 359500
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.521432966431942e-06,
+      "loss": 5.3728,
+      "step": 360000
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 9.52074929924113e-06,
+      "loss": 5.3701,
+      "step": 360500
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.520065632050318e-06,
+      "loss": 5.3642,
+      "step": 361000
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.519381964859508e-06,
+      "loss": 5.3679,
+      "step": 361500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 9.518698297668696e-06,
+      "loss": 5.3635,
+      "step": 362000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 9.518014630477884e-06,
+      "loss": 5.3651,
+      "step": 362500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.517330963287072e-06,
+      "loss": 5.3561,
+      "step": 363000
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 9.51664729609626e-06,
+      "loss": 5.3603,
+      "step": 363500
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 9.515963628905448e-06,
+      "loss": 5.3475,
+      "step": 364000
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.515279961714638e-06,
+      "loss": 5.3422,
+      "step": 364500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.514596294523826e-06,
+      "loss": 5.3477,
+      "step": 365000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 9.513912627333016e-06,
+      "loss": 5.3421,
+      "step": 365500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.513228960142204e-06,
+      "loss": 5.3403,
+      "step": 366000
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 9.512545292951392e-06,
+      "loss": 5.3226,
+      "step": 366500
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 9.51186162576058e-06,
+      "loss": 5.331,
+      "step": 367000
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.51117795856977e-06,
+      "loss": 5.3239,
+      "step": 367500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 9.510494291378958e-06,
+      "loss": 5.3238,
+      "step": 368000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 9.509810624188146e-06,
+      "loss": 5.3215,
+      "step": 368500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 9.509126956997334e-06,
+      "loss": 5.3155,
+      "step": 369000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.508443289806522e-06,
+      "loss": 5.3155,
+      "step": 369500
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 9.507759622615712e-06,
+      "loss": 5.3107,
+      "step": 370000
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 9.5070759554249e-06,
+      "loss": 5.3067,
+      "step": 370500
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.506392288234088e-06,
+      "loss": 5.3033,
+      "step": 371000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 9.505708621043278e-06,
+      "loss": 5.2961,
+      "step": 371500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 9.505024953852466e-06,
+      "loss": 5.2888,
+      "step": 372000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.504341286661654e-06,
+      "loss": 5.2889,
+      "step": 372500
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 9.503657619470842e-06,
+      "loss": 5.289,
+      "step": 373000
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 9.502973952280031e-06,
+      "loss": 5.2866,
+      "step": 373500
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.50229028508922e-06,
+      "loss": 5.2788,
+      "step": 374000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 9.501606617898408e-06,
+      "loss": 5.2836,
+      "step": 374500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 9.500922950707596e-06,
+      "loss": 5.2718,
+      "step": 375000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.500239283516784e-06,
+      "loss": 5.2735,
+      "step": 375500
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 9.499555616325973e-06,
+      "loss": 5.2677,
+      "step": 376000
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 9.498871949135161e-06,
+      "loss": 5.265,
+      "step": 376500
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.498188281944351e-06,
+      "loss": 5.2599,
+      "step": 377000
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.49750461475354e-06,
+      "loss": 5.2593,
+      "step": 377500
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.496820947562727e-06,
+      "loss": 5.2592,
+      "step": 378000
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.496137280371915e-06,
+      "loss": 5.2495,
+      "step": 378500
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.495453613181103e-06,
+      "loss": 5.2441,
+      "step": 379000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.494769945990293e-06,
+      "loss": 5.2419,
+      "step": 379500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.494086278799481e-06,
+      "loss": 5.2396,
+      "step": 380000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.49340261160867e-06,
+      "loss": 5.233,
+      "step": 380500
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.492718944417857e-06,
+      "loss": 5.2329,
+      "step": 381000
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.492035277227047e-06,
+      "loss": 5.2302,
+      "step": 381500
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.491351610036235e-06,
+      "loss": 5.2234,
+      "step": 382000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.490667942845425e-06,
+      "loss": 5.2228,
+      "step": 382500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.489984275654613e-06,
+      "loss": 5.2124,
+      "step": 383000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.489300608463801e-06,
+      "loss": 5.2087,
+      "step": 383500
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.488616941272989e-06,
+      "loss": 5.2036,
+      "step": 384000
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.487933274082177e-06,
+      "loss": 5.2026,
+      "step": 384500
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.487249606891365e-06,
+      "loss": 5.2026,
+      "step": 385000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.486565939700555e-06,
+      "loss": 5.1969,
+      "step": 385500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.485882272509743e-06,
+      "loss": 5.1949,
+      "step": 386000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.485198605318931e-06,
+      "loss": 5.1847,
+      "step": 386500
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.48451493812812e-06,
+      "loss": 5.189,
+      "step": 387000
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.483831270937309e-06,
+      "loss": 5.184,
+      "step": 387500
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.483147603746497e-06,
+      "loss": 5.1835,
+      "step": 388000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.482463936555687e-06,
+      "loss": 5.1711,
+      "step": 388500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.481780269364875e-06,
+      "loss": 5.1693,
+      "step": 389000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.481096602174063e-06,
+      "loss": 5.1664,
+      "step": 389500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.48041293498325e-06,
+      "loss": 5.1648,
+      "step": 390000
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.479729267792439e-06,
+      "loss": 5.1587,
+      "step": 390500
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.479045600601627e-06,
+      "loss": 5.1552,
+      "step": 391000
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.478361933410817e-06,
+      "loss": 5.1484,
+      "step": 391500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.477678266220005e-06,
+      "loss": 5.1457,
+      "step": 392000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.476994599029193e-06,
+      "loss": 5.1421,
+      "step": 392500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.476310931838382e-06,
+      "loss": 5.132,
+      "step": 393000
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.47562726464757e-06,
+      "loss": 5.1347,
+      "step": 393500
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.474943597456758e-06,
+      "loss": 5.1253,
+      "step": 394000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.474259930265948e-06,
+      "loss": 5.1215,
+      "step": 394500
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.473576263075136e-06,
+      "loss": 5.1199,
+      "step": 395000
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.472892595884324e-06,
+      "loss": 5.119,
+      "step": 395500
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.472208928693512e-06,
+      "loss": 5.1134,
+      "step": 396000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.4715252615027e-06,
+      "loss": 5.1084,
+      "step": 396500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.470841594311888e-06,
+      "loss": 5.0966,
+      "step": 397000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.470157927121078e-06,
+      "loss": 5.0981,
+      "step": 397500
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.469474259930266e-06,
+      "loss": 5.0936,
+      "step": 398000
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.468790592739456e-06,
+      "loss": 5.0914,
+      "step": 398500
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.468106925548644e-06,
+      "loss": 5.0792,
+      "step": 399000
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.467423258357832e-06,
+      "loss": 5.0786,
+      "step": 399500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 9.46673959116702e-06,
+      "loss": 5.0784,
+      "step": 400000
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 9.46605592397621e-06,
+      "loss": 5.0718,
+      "step": 400500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.465372256785398e-06,
+      "loss": 5.0725,
+      "step": 401000
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 9.464688589594586e-06,
+      "loss": 5.0648,
+      "step": 401500
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 9.464004922403774e-06,
+      "loss": 5.0627,
+      "step": 402000
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.463321255212962e-06,
+      "loss": 5.0609,
+      "step": 402500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 9.462637588022152e-06,
+      "loss": 5.0494,
+      "step": 403000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 9.46195392083134e-06,
+      "loss": 5.05,
+      "step": 403500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 9.461270253640528e-06,
+      "loss": 5.0394,
+      "step": 404000
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 9.460586586449718e-06,
+      "loss": 5.0423,
+      "step": 404500
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 9.459902919258906e-06,
+      "loss": 5.0376,
+      "step": 405000
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 9.459219252068094e-06,
+      "loss": 5.0356,
+      "step": 405500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 9.458535584877282e-06,
+      "loss": 5.0261,
+      "step": 406000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 9.457851917686472e-06,
+      "loss": 5.0207,
+      "step": 406500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 9.45716825049566e-06,
+      "loss": 5.0168,
+      "step": 407000
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 9.456484583304848e-06,
+      "loss": 5.0124,
+      "step": 407500
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 9.455800916114036e-06,
+      "loss": 5.0092,
+      "step": 408000
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 9.455117248923224e-06,
+      "loss": 5.0001,
+      "step": 408500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 9.454433581732414e-06,
+      "loss": 4.9998,
+      "step": 409000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 9.453749914541602e-06,
+      "loss": 5.0025,
+      "step": 409500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 9.453066247350791e-06,
+      "loss": 4.9983,
+      "step": 410000
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 9.45238258015998e-06,
+      "loss": 4.987,
+      "step": 410500
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 9.451698912969167e-06,
+      "loss": 4.9847,
+      "step": 411000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 9.451015245778355e-06,
+      "loss": 4.9813,
+      "step": 411500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 9.450331578587545e-06,
+      "loss": 4.9697,
+      "step": 412000
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 9.449647911396733e-06,
+      "loss": 4.967,
+      "step": 412500
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 9.448964244205921e-06,
+      "loss": 4.9688,
+      "step": 413000
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 9.44828057701511e-06,
+      "loss": 4.9616,
+      "step": 413500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 9.447596909824297e-06,
+      "loss": 4.9597,
+      "step": 414000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 9.446913242633487e-06,
+      "loss": 4.9517,
+      "step": 414500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 9.446229575442675e-06,
+      "loss": 4.9499,
+      "step": 415000
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 9.445545908251865e-06,
+      "loss": 4.9453,
+      "step": 415500
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 9.444862241061053e-06,
+      "loss": 4.9431,
+      "step": 416000
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 9.444178573870241e-06,
+      "loss": 4.931,
+      "step": 416500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 9.443494906679429e-06,
+      "loss": 4.9305,
+      "step": 417000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 9.442811239488617e-06,
+      "loss": 4.9264,
+      "step": 417500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 9.442127572297807e-06,
+      "loss": 4.9231,
+      "step": 418000
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 9.441443905106995e-06,
+      "loss": 4.9224,
+      "step": 418500
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 9.440760237916183e-06,
+      "loss": 4.9133,
+      "step": 419000
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.440076570725371e-06,
+      "loss": 4.9129,
+      "step": 419500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 9.43939290353456e-06,
+      "loss": 4.9008,
+      "step": 420000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 9.438709236343749e-06,
+      "loss": 4.9002,
+      "step": 420500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 9.438025569152937e-06,
+      "loss": 4.897,
+      "step": 421000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 9.437341901962127e-06,
+      "loss": 4.8901,
+      "step": 421500
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 9.436658234771315e-06,
+      "loss": 4.8978,
+      "step": 422000
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 9.435974567580503e-06,
+      "loss": 4.8885,
+      "step": 422500
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 9.43529090038969e-06,
+      "loss": 4.8819,
+      "step": 423000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 9.434607233198879e-06,
+      "loss": 4.8666,
+      "step": 423500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 9.433923566008069e-06,
+      "loss": 4.8666,
+      "step": 424000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.433239898817257e-06,
+      "loss": 4.868,
+      "step": 424500
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 9.432556231626445e-06,
+      "loss": 4.8614,
+      "step": 425000
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 9.431872564435633e-06,
+      "loss": 4.8551,
+      "step": 425500
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.431188897244822e-06,
+      "loss": 4.8522,
+      "step": 426000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 9.43050523005401e-06,
+      "loss": 4.8455,
+      "step": 426500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 9.4298215628632e-06,
+      "loss": 4.8497,
+      "step": 427000
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 9.429137895672388e-06,
+      "loss": 4.8422,
+      "step": 427500
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 9.428454228481576e-06,
+      "loss": 4.8381,
+      "step": 428000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 9.427770561290764e-06,
+      "loss": 4.8332,
+      "step": 428500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 9.427086894099952e-06,
+      "loss": 4.828,
+      "step": 429000
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 9.42640322690914e-06,
+      "loss": 4.8257,
+      "step": 429500
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 9.42571955971833e-06,
+      "loss": 4.8209,
+      "step": 430000
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 9.425035892527518e-06,
+      "loss": 4.8202,
+      "step": 430500
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 9.424352225336706e-06,
+      "loss": 4.8045,
+      "step": 431000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 9.423668558145896e-06,
+      "loss": 4.8086,
+      "step": 431500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 9.422984890955084e-06,
+      "loss": 4.796,
+      "step": 432000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.422301223764272e-06,
+      "loss": 4.8023,
+      "step": 432500
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 9.421617556573462e-06,
+      "loss": 4.8,
+      "step": 433000
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 9.42093388938265e-06,
+      "loss": 4.788,
+      "step": 433500
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 9.420250222191838e-06,
+      "loss": 4.7906,
+      "step": 434000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 9.419566555001026e-06,
+      "loss": 4.7796,
+      "step": 434500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 9.418882887810214e-06,
+      "loss": 4.775,
+      "step": 435000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 9.418199220619402e-06,
+      "loss": 4.7699,
+      "step": 435500
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 9.417515553428592e-06,
+      "loss": 4.7769,
+      "step": 436000
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 9.41683188623778e-06,
+      "loss": 4.7597,
+      "step": 436500
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.416148219046968e-06,
+      "loss": 4.7565,
+      "step": 437000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 9.415464551856158e-06,
+      "loss": 4.7594,
+      "step": 437500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 9.414780884665346e-06,
+      "loss": 4.7578,
+      "step": 438000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 9.414097217474534e-06,
+      "loss": 4.7443,
+      "step": 438500
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 9.413413550283724e-06,
+      "loss": 4.7457,
+      "step": 439000
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 9.412729883092912e-06,
+      "loss": 4.7385,
+      "step": 439500
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 9.4120462159021e-06,
+      "loss": 4.7331,
+      "step": 440000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 9.411362548711288e-06,
+      "loss": 4.7294,
+      "step": 440500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 9.410678881520476e-06,
+      "loss": 4.7235,
+      "step": 441000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 9.409995214329664e-06,
+      "loss": 4.7231,
+      "step": 441500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 9.409311547138854e-06,
+      "loss": 4.7169,
+      "step": 442000
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 9.408627879948042e-06,
+      "loss": 4.714,
+      "step": 442500
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 9.407944212757231e-06,
+      "loss": 4.7057,
+      "step": 443000
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 9.40726054556642e-06,
+      "loss": 4.7064,
+      "step": 443500
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 9.406576878375607e-06,
+      "loss": 4.7007,
+      "step": 444000
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 9.405893211184796e-06,
+      "loss": 4.6979,
+      "step": 444500
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 9.405209543993985e-06,
+      "loss": 4.6917,
+      "step": 445000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 9.404525876803173e-06,
+      "loss": 4.6891,
+      "step": 445500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 9.403842209612361e-06,
+      "loss": 4.6861,
+      "step": 446000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.40315854242155e-06,
+      "loss": 4.6716,
+      "step": 446500
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 9.402474875230737e-06,
+      "loss": 4.6704,
+      "step": 447000
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 9.401791208039927e-06,
+      "loss": 4.675,
+      "step": 447500
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 9.401107540849115e-06,
+      "loss": 4.6661,
+      "step": 448000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 9.400423873658305e-06,
+      "loss": 4.6651,
+      "step": 448500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 9.399740206467493e-06,
+      "loss": 4.664,
+      "step": 449000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 9.399056539276681e-06,
+      "loss": 4.6533,
+      "step": 449500
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 9.39837287208587e-06,
+      "loss": 4.6466,
+      "step": 450000
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 9.397689204895057e-06,
+      "loss": 4.6398,
+      "step": 450500
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 9.397005537704247e-06,
+      "loss": 4.6468,
+      "step": 451000
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 9.396321870513435e-06,
+      "loss": 4.6372,
+      "step": 451500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 9.395638203322623e-06,
+      "loss": 4.6343,
+      "step": 452000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 9.394954536131811e-06,
+      "loss": 4.6259,
+      "step": 452500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 9.394270868941001e-06,
+      "loss": 4.6267,
+      "step": 453000
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 9.393587201750189e-06,
+      "loss": 4.6198,
+      "step": 453500
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 9.392903534559377e-06,
+      "loss": 4.6203,
+      "step": 454000
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 9.392219867368567e-06,
+      "loss": 4.6121,
+      "step": 454500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 9.391536200177755e-06,
+      "loss": 4.607,
+      "step": 455000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 9.390852532986943e-06,
+      "loss": 4.6163,
+      "step": 455500
+    },
+    {
+      "epoch": 6.22,
+      "step": 455738,
+      "total_flos": 1.0079877250909471e+18,
+      "train_loss": 5.8071160892662865,
+      "train_runtime": 197998.8518,
+      "train_samples_per_second": 2958.986,
+      "train_steps_per_second": 36.988
+    }
+  ],
+  "logging_steps": 500,
+  "max_steps": 7323500,
+  "num_train_epochs": 100,
+  "save_steps": 1000000,
+  "total_flos": 1.0079877250909471e+18,
+  "trial_name": null,
+  "trial_params": null
+}