diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7879 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 654693,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9961814163279585e-05,
+      "loss": 3.3994,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.992362832655917e-05,
+      "loss": 3.3296,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.988544248983875e-05,
+      "loss": 3.305,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.984725665311833e-05,
+      "loss": 3.2736,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.980914718807136e-05,
+      "loss": 3.2743,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.977096135135094e-05,
+      "loss": 3.2606,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9732775514630525e-05,
+      "loss": 3.2427,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.969458967791011e-05,
+      "loss": 3.2402,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.965640384118969e-05,
+      "loss": 3.2245,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.961821800446927e-05,
+      "loss": 3.2243,
+      "step": 5000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9580032167748855e-05,
+      "loss": 3.2103,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.954184633102844e-05,
+      "loss": 3.2114,
+      "step": 6000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.950366049430803e-05,
+      "loss": 3.1994,
+      "step": 6500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.946555102926105e-05,
+      "loss": 3.1978,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.942736519254062e-05,
+      "loss": 3.1957,
+      "step": 7500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.938917935582021e-05,
+      "loss": 3.1963,
+      "step": 8000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9350993519099795e-05,
+      "loss": 3.188,
+      "step": 8500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9312884054052814e-05,
+      "loss": 3.1749,
+      "step": 9000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.92746982173324e-05,
+      "loss": 3.1691,
+      "step": 9500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923651238061198e-05,
+      "loss": 3.1681,
+      "step": 10000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919832654389156e-05,
+      "loss": 3.1713,
+      "step": 10500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.916014070717115e-05,
+      "loss": 3.1747,
+      "step": 11000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9121954870450734e-05,
+      "loss": 3.1719,
+      "step": 11500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.908376903373032e-05,
+      "loss": 3.1707,
+      "step": 12000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.90455831970099e-05,
+      "loss": 3.1666,
+      "step": 12500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.900739736028948e-05,
+      "loss": 3.1708,
+      "step": 13000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.89692878952425e-05,
+      "loss": 3.1665,
+      "step": 13500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.893110205852209e-05,
+      "loss": 3.1547,
+      "step": 14000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8892916221801674e-05,
+      "loss": 3.1513,
+      "step": 14500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.885480675675469e-05,
+      "loss": 3.1282,
+      "step": 15000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8816620920034276e-05,
+      "loss": 3.1537,
+      "step": 15500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.877843508331386e-05,
+      "loss": 3.1399,
+      "step": 16000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.874024924659344e-05,
+      "loss": 3.1364,
+      "step": 16500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.870206340987303e-05,
+      "loss": 3.13,
+      "step": 17000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.866387757315261e-05,
+      "loss": 3.1401,
+      "step": 17500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8625691736432196e-05,
+      "loss": 3.1489,
+      "step": 18000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.858750589971178e-05,
+      "loss": 3.1303,
+      "step": 18500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.85493964346648e-05,
+      "loss": 3.1322,
+      "step": 19000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.851121059794439e-05,
+      "loss": 3.1365,
+      "step": 19500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.847302476122397e-05,
+      "loss": 3.1354,
+      "step": 20000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.843483892450355e-05,
+      "loss": 3.1341,
+      "step": 20500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8396653087783135e-05,
+      "loss": 3.1198,
+      "step": 21000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835846725106272e-05,
+      "loss": 3.1277,
+      "step": 21500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.832035778601574e-05,
+      "loss": 3.1273,
+      "step": 22000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.828217194929532e-05,
+      "loss": 3.1133,
+      "step": 22500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.82439861125749e-05,
+      "loss": 3.1255,
+      "step": 23000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8205800275854486e-05,
+      "loss": 3.1172,
+      "step": 23500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.816761443913407e-05,
+      "loss": 3.1118,
+      "step": 24000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.812942860241365e-05,
+      "loss": 3.1041,
+      "step": 24500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8091242765693233e-05,
+      "loss": 3.12,
+      "step": 25000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.805313330064626e-05,
+      "loss": 3.1123,
+      "step": 25500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.801494746392584e-05,
+      "loss": 3.1124,
+      "step": 26000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7976761627205425e-05,
+      "loss": 3.117,
+      "step": 26500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.793857579048501e-05,
+      "loss": 3.1003,
+      "step": 27000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.790038995376459e-05,
+      "loss": 3.0989,
+      "step": 27500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.786220411704417e-05,
+      "loss": 3.1008,
+      "step": 28000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7824018280323756e-05,
+      "loss": 3.1083,
+      "step": 28500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.778583244360334e-05,
+      "loss": 3.1067,
+      "step": 29000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7747722978556365e-05,
+      "loss": 3.1142,
+      "step": 29500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.770961351350939e-05,
+      "loss": 3.1092,
+      "step": 30000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7671427676788974e-05,
+      "loss": 3.0984,
+      "step": 30500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7633241840068556e-05,
+      "loss": 3.1071,
+      "step": 31000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7595132375021576e-05,
+      "loss": 3.0963,
+      "step": 31500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.755694653830116e-05,
+      "loss": 3.1143,
+      "step": 32000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751876070158074e-05,
+      "loss": 3.1029,
+      "step": 32500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.748057486486033e-05,
+      "loss": 3.0904,
+      "step": 33000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.744238902813991e-05,
+      "loss": 3.0907,
+      "step": 33500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7404203191419496e-05,
+      "loss": 3.0907,
+      "step": 34000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.736601735469908e-05,
+      "loss": 3.0945,
+      "step": 34500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732783151797866e-05,
+      "loss": 3.0867,
+      "step": 35000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.728972205293168e-05,
+      "loss": 3.0839,
+      "step": 35500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.725153621621127e-05,
+      "loss": 3.0922,
+      "step": 36000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7213350379490846e-05,
+      "loss": 3.0875,
+      "step": 36500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.717516454277043e-05,
+      "loss": 3.0865,
+      "step": 37000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.713697870605001e-05,
+      "loss": 3.0835,
+      "step": 37500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.709886924100304e-05,
+      "loss": 3.0853,
+      "step": 38000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.706068340428262e-05,
+      "loss": 3.09,
+      "step": 38500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.70224975675622e-05,
+      "loss": 3.0848,
+      "step": 39000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6984311730841786e-05,
+      "loss": 3.0963,
+      "step": 39500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.694612589412137e-05,
+      "loss": 3.089,
+      "step": 40000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.690794005740095e-05,
+      "loss": 3.0775,
+      "step": 40500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6869754220680533e-05,
+      "loss": 3.0778,
+      "step": 41000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6831568383960116e-05,
+      "loss": 3.0788,
+      "step": 41500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.679345891891314e-05,
+      "loss": 3.0818,
+      "step": 42000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6755273082192725e-05,
+      "loss": 3.0795,
+      "step": 42500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.671708724547231e-05,
+      "loss": 3.0686,
+      "step": 43000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667890140875189e-05,
+      "loss": 3.0786,
+      "step": 43500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.664071557203147e-05,
+      "loss": 3.0738,
+      "step": 44000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.660268247865794e-05,
+      "loss": 3.0893,
+      "step": 44500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.656449664193752e-05,
+      "loss": 3.0732,
+      "step": 45000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65263108052171e-05,
+      "loss": 3.0673,
+      "step": 45500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648812496849669e-05,
+      "loss": 3.0918,
+      "step": 46000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.645001550344971e-05,
+      "loss": 3.062,
+      "step": 46500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6411829666729294e-05,
+      "loss": 3.0743,
+      "step": 47000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6373643830008876e-05,
+      "loss": 3.0738,
+      "step": 47500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.633545799328846e-05,
+      "loss": 3.0697,
+      "step": 48000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.629727215656804e-05,
+      "loss": 3.0694,
+      "step": 48500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.625908631984763e-05,
+      "loss": 3.0745,
+      "step": 49000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6220900483127213e-05,
+      "loss": 3.0493,
+      "step": 49500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6182714646406796e-05,
+      "loss": 3.0638,
+      "step": 50000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6144605181359816e-05,
+      "loss": 3.0588,
+      "step": 50500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61064193446394e-05,
+      "loss": 3.0617,
+      "step": 51000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.606823350791898e-05,
+      "loss": 3.0757,
+      "step": 51500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.6030047671198564e-05,
+      "loss": 3.0552,
+      "step": 52000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5991861834478146e-05,
+      "loss": 3.0695,
+      "step": 52500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.5953752369431166e-05,
+      "loss": 3.0476,
+      "step": 53000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5915566532710755e-05,
+      "loss": 3.0548,
+      "step": 53500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.587738069599034e-05,
+      "loss": 3.0757,
+      "step": 54000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.583919485926992e-05,
+      "loss": 3.0554,
+      "step": 54500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.58010090225495e-05,
+      "loss": 3.0629,
+      "step": 55000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5762823185829086e-05,
+      "loss": 3.0532,
+      "step": 55500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.572463734910867e-05,
+      "loss": 3.0625,
+      "step": 56000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568645151238825e-05,
+      "loss": 3.0506,
+      "step": 56500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5648265675667834e-05,
+      "loss": 3.0557,
+      "step": 57000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.561015621062086e-05,
+      "loss": 3.0515,
+      "step": 57500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.557197037390044e-05,
+      "loss": 3.0627,
+      "step": 58000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5533784537180025e-05,
+      "loss": 3.0472,
+      "step": 58500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549559870045961e-05,
+      "loss": 3.0565,
+      "step": 59000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5457489235412634e-05,
+      "loss": 3.0726,
+      "step": 59500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.541930339869222e-05,
+      "loss": 3.0383,
+      "step": 60000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538119393364524e-05,
+      "loss": 3.0565,
+      "step": 60500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.534300809692482e-05,
+      "loss": 3.0541,
+      "step": 61000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.53048222602044e-05,
+      "loss": 3.0567,
+      "step": 61500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.5266636423483985e-05,
+      "loss": 3.0433,
+      "step": 62000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5228450586763574e-05,
+      "loss": 3.0521,
+      "step": 62500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5190264750043157e-05,
+      "loss": 3.047,
+      "step": 63000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.515207891332274e-05,
+      "loss": 3.0514,
+      "step": 63500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.511389307660232e-05,
+      "loss": 3.048,
+      "step": 64000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.507578361155534e-05,
+      "loss": 3.0481,
+      "step": 64500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.5037597774834924e-05,
+      "loss": 3.0448,
+      "step": 65000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.499941193811451e-05,
+      "loss": 3.0535,
+      "step": 65500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.496122610139409e-05,
+      "loss": 3.0493,
+      "step": 66000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.492311663634711e-05,
+      "loss": 3.0479,
+      "step": 66500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48849307996267e-05,
+      "loss": 3.0612,
+      "step": 67000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.484674496290628e-05,
+      "loss": 3.0458,
+      "step": 67500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4808559126185864e-05,
+      "loss": 3.0583,
+      "step": 68000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4770373289465446e-05,
+      "loss": 3.0447,
+      "step": 68500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.473218745274503e-05,
+      "loss": 3.0361,
+      "step": 69000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469400161602461e-05,
+      "loss": 3.054,
+      "step": 69500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4655815779304194e-05,
+      "loss": 3.0509,
+      "step": 70000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.461762994258378e-05,
+      "loss": 3.0346,
+      "step": 70500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.45795204775368e-05,
+      "loss": 3.043,
+      "step": 71000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.4541334640816386e-05,
+      "loss": 3.0324,
+      "step": 71500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.450314880409597e-05,
+      "loss": 3.0319,
+      "step": 72000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.446496296737555e-05,
+      "loss": 3.0301,
+      "step": 72500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.442685350232858e-05,
+      "loss": 3.0262,
+      "step": 73000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.438866766560816e-05,
+      "loss": 3.0368,
+      "step": 73500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.435048182888774e-05,
+      "loss": 3.0453,
+      "step": 74000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4312295992167325e-05,
+      "loss": 3.0286,
+      "step": 74500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.427411015544691e-05,
+      "loss": 3.0333,
+      "step": 75000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4236000690399934e-05,
+      "loss": 3.0387,
+      "step": 75500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.419781485367952e-05,
+      "loss": 3.0283,
+      "step": 76000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.41596290169591e-05,
+      "loss": 3.0406,
+      "step": 76500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.412144318023868e-05,
+      "loss": 3.0335,
+      "step": 77000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.40833337151917e-05,
+      "loss": 3.0422,
+      "step": 77500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.404522425014473e-05,
+      "loss": 3.0233,
+      "step": 78000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.400703841342431e-05,
+      "loss": 3.0304,
+      "step": 78500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3968852576703894e-05,
+      "loss": 3.0296,
+      "step": 79000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.393066673998347e-05,
+      "loss": 3.0396,
+      "step": 79500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.389248090326306e-05,
+      "loss": 3.0348,
+      "step": 80000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385429506654264e-05,
+      "loss": 3.0316,
+      "step": 80500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.381618560149566e-05,
+      "loss": 3.0281,
+      "step": 81000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.3777999764775244e-05,
+      "loss": 3.0348,
+      "step": 81500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3739813928054827e-05,
+      "loss": 3.023,
+      "step": 82000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.370162809133441e-05,
+      "loss": 3.0399,
+      "step": 82500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3663442254614e-05,
+      "loss": 3.0136,
+      "step": 83000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.362525641789358e-05,
+      "loss": 3.0187,
+      "step": 83500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.35871469528466e-05,
+      "loss": 3.0331,
+      "step": 84000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3548961116126183e-05,
+      "loss": 3.0218,
+      "step": 84500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3510775279405766e-05,
+      "loss": 3.0333,
+      "step": 85000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.347258944268535e-05,
+      "loss": 3.0185,
+      "step": 85500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.343440360596494e-05,
+      "loss": 3.0311,
+      "step": 86000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.339629414091796e-05,
+      "loss": 3.0337,
+      "step": 86500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.335810830419754e-05,
+      "loss": 3.014,
+      "step": 87000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.331992246747712e-05,
+      "loss": 3.0163,
+      "step": 87500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.3281736630756706e-05,
+      "loss": 3.0079,
+      "step": 88000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.324355079403629e-05,
+      "loss": 3.0058,
+      "step": 88500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.320536495731588e-05,
+      "loss": 3.0199,
+      "step": 89000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.316717912059546e-05,
+      "loss": 3.0348,
+      "step": 89500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312899328387504e-05,
+      "loss": 3.0128,
+      "step": 90000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.309088381882806e-05,
+      "loss": 3.0251,
+      "step": 90500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3052697982107645e-05,
+      "loss": 3.0263,
+      "step": 91000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3014512145387235e-05,
+      "loss": 3.036,
+      "step": 91500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297632630866682e-05,
+      "loss": 3.0192,
+      "step": 92000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.293821684361984e-05,
+      "loss": 3.0077,
+      "step": 92500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.290010737857286e-05,
+      "loss": 3.0209,
+      "step": 93000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2861921541852446e-05,
+      "loss": 3.0112,
+      "step": 93500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282373570513202e-05,
+      "loss": 3.0159,
+      "step": 94000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2785549868411604e-05,
+      "loss": 3.0155,
+      "step": 94500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.274736403169119e-05,
+      "loss": 3.0239,
+      "step": 95000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.270917819497077e-05,
+      "loss": 3.0066,
+      "step": 95500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.2671068729923796e-05,
+      "loss": 3.0196,
+      "step": 96000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.263288289320338e-05,
+      "loss": 3.0061,
+      "step": 96500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.259469705648296e-05,
+      "loss": 3.0178,
+      "step": 97000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2556511219762544e-05,
+      "loss": 3.0051,
+      "step": 97500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2518325383042127e-05,
+      "loss": 3.0177,
+      "step": 98000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.248021591799515e-05,
+      "loss": 3.0114,
+      "step": 98500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2442030081274736e-05,
+      "loss": 3.0175,
+      "step": 99000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.240384424455432e-05,
+      "loss": 3.0158,
+      "step": 99500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.23656584078339e-05,
+      "loss": 3.0142,
+      "step": 100000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2327472571113484e-05,
+      "loss": 3.0104,
+      "step": 100500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2289286734393066e-05,
+      "loss": 3.0149,
+      "step": 101000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.225110089767265e-05,
+      "loss": 3.0157,
+      "step": 101500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.221291506095224e-05,
+      "loss": 3.0264,
+      "step": 102000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.217472922423182e-05,
+      "loss": 3.0066,
+      "step": 102500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213661975918484e-05,
+      "loss": 3.0054,
+      "step": 103000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.209843392246442e-05,
+      "loss": 3.0055,
+      "step": 103500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2060248085744006e-05,
+      "loss": 3.0023,
+      "step": 104000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202206224902359e-05,
+      "loss": 3.0115,
+      "step": 104500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.1983952783976615e-05,
+      "loss": 3.0115,
+      "step": 105000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.19457669472562e-05,
+      "loss": 3.0077,
+      "step": 105500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.190758111053578e-05,
+      "loss": 3.0016,
+      "step": 106000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.186939527381536e-05,
+      "loss": 3.003,
+      "step": 106500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1831209437094945e-05,
+      "loss": 3.0059,
+      "step": 107000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.179302360037453e-05,
+      "loss": 3.0143,
+      "step": 107500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.175483776365412e-05,
+      "loss": 3.0031,
+      "step": 108000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.171672829860713e-05,
+      "loss": 3.0139,
+      "step": 108500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.167854246188671e-05,
+      "loss": 3.0138,
+      "step": 109000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16403566251663e-05,
+      "loss": 3.0069,
+      "step": 109500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1602170788445885e-05,
+      "loss": 3.0008,
+      "step": 110000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.156398495172547e-05,
+      "loss": 3.0142,
+      "step": 110500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.152579911500505e-05,
+      "loss": 3.0116,
+      "step": 111000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.148761327828463e-05,
+      "loss": 2.9916,
+      "step": 111500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1449427441564215e-05,
+      "loss": 2.9977,
+      "step": 112000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.141131797651724e-05,
+      "loss": 2.9955,
+      "step": 112500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1373132139796824e-05,
+      "loss": 2.9961,
+      "step": 113000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.133494630307641e-05,
+      "loss": 2.9986,
+      "step": 113500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.129676046635599e-05,
+      "loss": 2.9989,
+      "step": 114000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.125865100130901e-05,
+      "loss": 3.0018,
+      "step": 114500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.12204651645886e-05,
+      "loss": 3.0059,
+      "step": 115000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118227932786818e-05,
+      "loss": 2.9917,
+      "step": 115500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1144093491147764e-05,
+      "loss": 3.0036,
+      "step": 116000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.1105984026100784e-05,
+      "loss": 2.9926,
+      "step": 116500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1067798189380366e-05,
+      "loss": 3.0071,
+      "step": 117000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.102961235265995e-05,
+      "loss": 2.9953,
+      "step": 117500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099142651593954e-05,
+      "loss": 3.0102,
+      "step": 118000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.095324067921912e-05,
+      "loss": 3.0012,
+      "step": 118500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0915054842498703e-05,
+      "loss": 3.0026,
+      "step": 119000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0876869005778286e-05,
+      "loss": 3.0046,
+      "step": 119500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.083868316905787e-05,
+      "loss": 2.9866,
+      "step": 120000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.080049733233745e-05,
+      "loss": 3.0078,
+      "step": 120500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0762464238963915e-05,
+      "loss": 2.9826,
+      "step": 121000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.07242784022435e-05,
+      "loss": 3.0031,
+      "step": 121500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.068609256552308e-05,
+      "loss": 2.9888,
+      "step": 122000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.064790672880266e-05,
+      "loss": 2.999,
+      "step": 122500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0609720892082245e-05,
+      "loss": 3.0009,
+      "step": 123000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057153505536183e-05,
+      "loss": 2.9958,
+      "step": 123500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053334921864141e-05,
+      "loss": 2.9914,
+      "step": 124000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049523975359443e-05,
+      "loss": 2.9922,
+      "step": 124500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045705391687401e-05,
+      "loss": 2.9997,
+      "step": 125000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.04188680801536e-05,
+      "loss": 2.9918,
+      "step": 125500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0380682243433185e-05,
+      "loss": 2.9689,
+      "step": 126000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.034249640671277e-05,
+      "loss": 3.0014,
+      "step": 126500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030431056999235e-05,
+      "loss": 2.9918,
+      "step": 127000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026612473327193e-05,
+      "loss": 3.0096,
+      "step": 127500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0227938896551515e-05,
+      "loss": 3.0136,
+      "step": 128000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.018982943150454e-05,
+      "loss": 2.9966,
+      "step": 128500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0151643594784124e-05,
+      "loss": 2.9937,
+      "step": 129000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.011345775806371e-05,
+      "loss": 3.0051,
+      "step": 129500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.007527192134329e-05,
+      "loss": 2.9994,
+      "step": 130000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.003708608462287e-05,
+      "loss": 2.9839,
+      "step": 130500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.999897661957589e-05,
+      "loss": 2.9765,
+      "step": 131000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.996079078285548e-05,
+      "loss": 2.9881,
+      "step": 131500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.9922604946135064e-05,
+      "loss": 3.0009,
+      "step": 132000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.988441910941465e-05,
+      "loss": 2.9904,
+      "step": 132500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.984623327269423e-05,
+      "loss": 2.9738,
+      "step": 133000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.980804743597381e-05,
+      "loss": 2.9845,
+      "step": 133500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9769861599253394e-05,
+      "loss": 2.9975,
+      "step": 134000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.973167576253298e-05,
+      "loss": 2.9949,
+      "step": 134500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.969348992581256e-05,
+      "loss": 2.9835,
+      "step": 135000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.965545683243902e-05,
+      "loss": 2.9957,
+      "step": 135500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9617270995718606e-05,
+      "loss": 2.9989,
+      "step": 136000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.957908515899819e-05,
+      "loss": 2.992,
+      "step": 136500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.954089932227777e-05,
+      "loss": 2.9782,
+      "step": 137000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9502713485557354e-05,
+      "loss": 2.9886,
+      "step": 137500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9464527648836936e-05,
+      "loss": 2.9913,
+      "step": 138000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.942634181211652e-05,
+      "loss": 3.0059,
+      "step": 138500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9388232347069545e-05,
+      "loss": 2.975,
+      "step": 139000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935004651034913e-05,
+      "loss": 2.98,
+      "step": 139500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931186067362871e-05,
+      "loss": 2.9788,
+      "step": 140000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927367483690829e-05,
+      "loss": 2.9763,
+      "step": 140500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9235489000187876e-05,
+      "loss": 2.9751,
+      "step": 141000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.919730316346746e-05,
+      "loss": 2.9754,
+      "step": 141500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.915911732674704e-05,
+      "loss": 2.9683,
+      "step": 142000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9120931490026624e-05,
+      "loss": 2.9905,
+      "step": 142500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.908282202497965e-05,
+      "loss": 2.9867,
+      "step": 143000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.904471255993267e-05,
+      "loss": 2.9915,
+      "step": 143500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.900652672321225e-05,
+      "loss": 2.991,
+      "step": 144000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.896834088649184e-05,
+      "loss": 2.9662,
+      "step": 144500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.8930155049771425e-05,
+      "loss": 2.9803,
+      "step": 145000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.889196921305101e-05,
+      "loss": 2.9769,
+      "step": 145500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.885385974800403e-05,
+      "loss": 2.9862,
+      "step": 146000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.881567391128361e-05,
+      "loss": 2.9916,
+      "step": 146500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.877748807456319e-05,
+      "loss": 2.9849,
+      "step": 147000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.873930223784278e-05,
+      "loss": 2.9755,
+      "step": 147500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8701116401122364e-05,
+      "loss": 2.9675,
+      "step": 148000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.866293056440195e-05,
+      "loss": 2.9773,
+      "step": 148500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.862474472768153e-05,
+      "loss": 2.9688,
+      "step": 149000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.858655889096111e-05,
+      "loss": 2.9827,
+      "step": 149500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.854844942591413e-05,
+      "loss": 2.9715,
+      "step": 150000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851026358919372e-05,
+      "loss": 2.988,
+      "step": 150500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8472077752473304e-05,
+      "loss": 2.9666,
+      "step": 151000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.843389191575288e-05,
+      "loss": 2.9847,
+      "step": 151500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.839570607903246e-05,
+      "loss": 2.9882,
+      "step": 152000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.835759661398549e-05,
+      "loss": 2.9871,
+      "step": 152500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.831941077726507e-05,
+      "loss": 2.9724,
+      "step": 153000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8281224940544654e-05,
+      "loss": 2.978,
+      "step": 153500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8243039103824236e-05,
+      "loss": 2.9718,
+      "step": 154000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8204929638777256e-05,
+      "loss": 2.9806,
+      "step": 154500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8166743802056846e-05,
+      "loss": 2.9943,
+      "step": 155000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.812855796533643e-05,
+      "loss": 2.9699,
+      "step": 155500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.809037212861601e-05,
+      "loss": 2.9909,
+      "step": 156000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805226266356903e-05,
+      "loss": 2.9693,
+      "step": 156500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.801407682684861e-05,
+      "loss": 2.9685,
+      "step": 157000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7975890990128196e-05,
+      "loss": 2.9776,
+      "step": 157500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7937705153407785e-05,
+      "loss": 2.9818,
+      "step": 158000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7899595688360805e-05,
+      "loss": 2.99,
+      "step": 158500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.786140985164039e-05,
+      "loss": 2.9751,
+      "step": 159000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.782322401491997e-05,
+      "loss": 2.9857,
+      "step": 159500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.778503817819955e-05,
+      "loss": 2.9775,
+      "step": 160000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.774685234147914e-05,
+      "loss": 2.9628,
+      "step": 160500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7708666504758725e-05,
+      "loss": 2.9836,
+      "step": 161000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7670557039711744e-05,
+      "loss": 2.9889,
+      "step": 161500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763237120299133e-05,
+      "loss": 2.9796,
+      "step": 162000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.759418536627091e-05,
+      "loss": 2.9657,
+      "step": 162500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755599952955049e-05,
+      "loss": 2.9832,
+      "step": 163000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.751781369283008e-05,
+      "loss": 2.9673,
+      "step": 163500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.74797042277831e-05,
+      "loss": 2.9691,
+      "step": 164000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7441518391062684e-05,
+      "loss": 2.9799,
+      "step": 164500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7403332554342266e-05,
+      "loss": 2.9755,
+      "step": 165000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.736514671762185e-05,
+      "loss": 2.976,
+      "step": 165500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.732696088090143e-05,
+      "loss": 2.9623,
+      "step": 166000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.728885141585445e-05,
+      "loss": 2.9701,
+      "step": 166500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7250665579134034e-05,
+      "loss": 2.966,
+      "step": 167000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.721247974241362e-05,
+      "loss": 2.9803,
+      "step": 167500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7174293905693206e-05,
+      "loss": 2.9721,
+      "step": 168000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.713610806897279e-05,
+      "loss": 2.9754,
+      "step": 168500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.709792223225237e-05,
+      "loss": 2.9712,
+      "step": 169000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7059736395531954e-05,
+      "loss": 2.9778,
+      "step": 169500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7021626930484974e-05,
+      "loss": 2.9746,
+      "step": 170000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6983441093764556e-05,
+      "loss": 2.9752,
+      "step": 170500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.6945255257044146e-05,
+      "loss": 2.9721,
+      "step": 171000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.690706942032373e-05,
+      "loss": 2.9713,
+      "step": 171500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.686888358360331e-05,
+      "loss": 2.9573,
+      "step": 172000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6830697746882893e-05,
+      "loss": 2.96,
+      "step": 172500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6792511910162476e-05,
+      "loss": 2.9653,
+      "step": 173000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.675432607344206e-05,
+      "loss": 2.9597,
+      "step": 173500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6716216608395085e-05,
+      "loss": 2.9688,
+      "step": 174000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.667803077167467e-05,
+      "loss": 2.9517,
+      "step": 174500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.663984493495425e-05,
+      "loss": 2.9729,
+      "step": 175000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.660165909823383e-05,
+      "loss": 2.961,
+      "step": 175500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6563473261513416e-05,
+      "loss": 2.9715,
+      "step": 176000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6525287424793e-05,
+      "loss": 2.9807,
+      "step": 176500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6487177959746025e-05,
+      "loss": 2.9624,
+      "step": 177000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.644899212302561e-05,
+      "loss": 2.9659,
+      "step": 177500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.641088265797863e-05,
+      "loss": 2.963,
+      "step": 178000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.637269682125821e-05,
+      "loss": 2.9727,
+      "step": 178500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.633451098453779e-05,
+      "loss": 2.9773,
+      "step": 179000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6296325147817375e-05,
+      "loss": 2.9798,
+      "step": 179500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6258139311096964e-05,
+      "loss": 2.9687,
+      "step": 180000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.622002984604998e-05,
+      "loss": 2.9678,
+      "step": 180500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.618184400932956e-05,
+      "loss": 2.9678,
+      "step": 181000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.614365817260915e-05,
+      "loss": 2.9757,
+      "step": 181500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.610547233588873e-05,
+      "loss": 2.9703,
+      "step": 182000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6067286499168314e-05,
+      "loss": 2.9592,
+      "step": 182500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.60291006624479e-05,
+      "loss": 2.9723,
+      "step": 183000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.599091482572748e-05,
+      "loss": 2.9544,
+      "step": 183500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.595272898900706e-05,
+      "loss": 2.9463,
+      "step": 184000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5914543152286645e-05,
+      "loss": 2.9793,
+      "step": 184500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.587643368723967e-05,
+      "loss": 2.9551,
+      "step": 185000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5838247850519254e-05,
+      "loss": 2.9549,
+      "step": 185500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5800062013798837e-05,
+      "loss": 2.9715,
+      "step": 186000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.576187617707842e-05,
+      "loss": 2.9718,
+      "step": 186500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5723690340358e-05,
+      "loss": 2.9639,
+      "step": 187000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.5685504503637584e-05,
+      "loss": 2.9505,
+      "step": 187500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.564731866691717e-05,
+      "loss": 2.9474,
+      "step": 188000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.560913283019675e-05,
+      "loss": 2.9622,
+      "step": 188500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.557109973682321e-05,
+      "loss": 2.9518,
+      "step": 189000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5532913900102796e-05,
+      "loss": 2.962,
+      "step": 189500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5494728063382385e-05,
+      "loss": 2.9608,
+      "step": 190000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.545654222666197e-05,
+      "loss": 2.9591,
+      "step": 190500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.541835638994155e-05,
+      "loss": 2.9564,
+      "step": 191000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.538017055322113e-05,
+      "loss": 2.9511,
+      "step": 191500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5341984716500716e-05,
+      "loss": 2.9512,
+      "step": 192000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.53037988797803e-05,
+      "loss": 2.9628,
+      "step": 192500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.526561304305988e-05,
+      "loss": 2.9589,
+      "step": 193000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5227427206339464e-05,
+      "loss": 2.9587,
+      "step": 193500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5189241369619046e-05,
+      "loss": 2.9677,
+      "step": 194000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.515113190457207e-05,
+      "loss": 2.9532,
+      "step": 194500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.511294606785165e-05,
+      "loss": 2.9587,
+      "step": 195000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507476023113123e-05,
+      "loss": 2.9514,
+      "step": 195500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.5036574394410814e-05,
+      "loss": 2.9527,
+      "step": 196000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.4998388557690396e-05,
+      "loss": 2.97,
+      "step": 196500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496035546431686e-05,
+      "loss": 2.9486,
+      "step": 197000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.492216962759645e-05,
+      "loss": 2.948,
+      "step": 197500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488398379087603e-05,
+      "loss": 2.9586,
+      "step": 198000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4845797954155615e-05,
+      "loss": 2.9696,
+      "step": 198500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.48076121174352e-05,
+      "loss": 2.9685,
+      "step": 199000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.476942628071478e-05,
+      "loss": 2.9576,
+      "step": 199500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.473124044399436e-05,
+      "loss": 2.9533,
+      "step": 200000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.469313097894739e-05,
+      "loss": 2.9726,
+      "step": 200500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465494514222697e-05,
+      "loss": 2.9693,
+      "step": 201000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4616759305506554e-05,
+      "loss": 2.963,
+      "step": 201500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.457857346878614e-05,
+      "loss": 2.953,
+      "step": 202000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.454038763206572e-05,
+      "loss": 2.9575,
+      "step": 202500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.45022017953453e-05,
+      "loss": 2.9459,
+      "step": 203000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4464015958624885e-05,
+      "loss": 2.9511,
+      "step": 203500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.442583012190447e-05,
+      "loss": 2.9526,
+      "step": 204000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.438764428518405e-05,
+      "loss": 2.9534,
+      "step": 204500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4349534820137076e-05,
+      "loss": 2.942,
+      "step": 205000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.431134898341666e-05,
+      "loss": 2.9637,
+      "step": 205500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.427316314669624e-05,
+      "loss": 2.9601,
+      "step": 206000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4234977309975824e-05,
+      "loss": 2.9624,
+      "step": 206500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.419686784492885e-05,
+      "loss": 2.9559,
+      "step": 207000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415875837988187e-05,
+      "loss": 2.9614,
+      "step": 207500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.412057254316145e-05,
+      "loss": 2.9544,
+      "step": 208000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4082386706441035e-05,
+      "loss": 2.9587,
+      "step": 208500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.404420086972062e-05,
+      "loss": 2.9406,
+      "step": 209000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.40060150330002e-05,
+      "loss": 2.9392,
+      "step": 209500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.396782919627978e-05,
+      "loss": 2.9505,
+      "step": 210000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3929643359559366e-05,
+      "loss": 2.957,
+      "step": 210500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389145752283895e-05,
+      "loss": 2.9626,
+      "step": 211000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.385327168611853e-05,
+      "loss": 2.9442,
+      "step": 211500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3815085849398114e-05,
+      "loss": 2.9417,
+      "step": 212000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3776900012677696e-05,
+      "loss": 2.956,
+      "step": 212500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.373879054763072e-05,
+      "loss": 2.9417,
+      "step": 213000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3700604710910306e-05,
+      "loss": 2.9512,
+      "step": 213500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.366241887418989e-05,
+      "loss": 2.9493,
+      "step": 214000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.362423303746947e-05,
+      "loss": 2.9472,
+      "step": 214500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358604720074905e-05,
+      "loss": 2.9493,
+      "step": 215000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.354793773570208e-05,
+      "loss": 2.9581,
+      "step": 215500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.350975189898166e-05,
+      "loss": 2.9589,
+      "step": 216000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3471566062261245e-05,
+      "loss": 2.9538,
+      "step": 216500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.343338022554083e-05,
+      "loss": 2.9613,
+      "step": 217000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.339534713216729e-05,
+      "loss": 2.9632,
+      "step": 217500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3357161295446874e-05,
+      "loss": 2.9513,
+      "step": 218000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3318975458726456e-05,
+      "loss": 2.9354,
+      "step": 218500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.328078962200604e-05,
+      "loss": 2.9409,
+      "step": 219000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324260378528563e-05,
+      "loss": 2.934,
+      "step": 219500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.320441794856521e-05,
+      "loss": 2.9271,
+      "step": 220000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3166232111844794e-05,
+      "loss": 2.9353,
+      "step": 220500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3128046275124376e-05,
+      "loss": 2.9385,
+      "step": 221000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.308986043840396e-05,
+      "loss": 2.9259,
+      "step": 221500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.305167460168354e-05,
+      "loss": 2.9339,
+      "step": 222000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3013488764963124e-05,
+      "loss": 2.9257,
+      "step": 222500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.297530292824271e-05,
+      "loss": 2.9194,
+      "step": 223000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.293711709152228e-05,
+      "loss": 2.936,
+      "step": 223500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.289908399814875e-05,
+      "loss": 2.9356,
+      "step": 224000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2860898161428336e-05,
+      "loss": 2.9061,
+      "step": 224500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.282271232470792e-05,
+      "loss": 2.9541,
+      "step": 225000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.27845264879875e-05,
+      "loss": 2.9272,
+      "step": 225500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2746340651267083e-05,
+      "loss": 2.9489,
+      "step": 226000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2708154814546666e-05,
+      "loss": 2.9429,
+      "step": 226500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.266996897782625e-05,
+      "loss": 2.931,
+      "step": 227000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.263178314110583e-05,
+      "loss": 2.9354,
+      "step": 227500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.259367367605886e-05,
+      "loss": 2.9262,
+      "step": 228000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.255548783933844e-05,
+      "loss": 2.9413,
+      "step": 228500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.251730200261802e-05,
+      "loss": 2.9255,
+      "step": 229000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2479116165897606e-05,
+      "loss": 2.938,
+      "step": 229500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.244093032917719e-05,
+      "loss": 2.9349,
+      "step": 230000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2402820864130215e-05,
+      "loss": 2.9299,
+      "step": 230500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.23646350274098e-05,
+      "loss": 2.9369,
+      "step": 231000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.232644919068938e-05,
+      "loss": 2.9277,
+      "step": 231500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.228826335396896e-05,
+      "loss": 2.9461,
+      "step": 232000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.225015388892199e-05,
+      "loss": 2.9482,
+      "step": 232500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.221196805220157e-05,
+      "loss": 2.9376,
+      "step": 233000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.2173782215481154e-05,
+      "loss": 2.9223,
+      "step": 233500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.213559637876074e-05,
+      "loss": 2.9348,
+      "step": 234000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.209741054204032e-05,
+      "loss": 2.9384,
+      "step": 234500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.20592247053199e-05,
+      "loss": 2.9315,
+      "step": 235000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.202111524027293e-05,
+      "loss": 2.9465,
+      "step": 235500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.198292940355251e-05,
+      "loss": 2.943,
+      "step": 236000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.1944743566832094e-05,
+      "loss": 2.9248,
+      "step": 236500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1906557730111676e-05,
+      "loss": 2.9476,
+      "step": 237000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.186837189339126e-05,
+      "loss": 2.9333,
+      "step": 237500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.183026242834428e-05,
+      "loss": 2.9279,
+      "step": 238000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.179207659162386e-05,
+      "loss": 2.9395,
+      "step": 238500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1753890754903444e-05,
+      "loss": 2.9444,
+      "step": 239000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1715704918183027e-05,
+      "loss": 2.9345,
+      "step": 239500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.167751908146261e-05,
+      "loss": 2.9302,
+      "step": 240000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.163933324474219e-05,
+      "loss": 2.9331,
+      "step": 240500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1601147408021774e-05,
+      "loss": 2.9322,
+      "step": 241000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.15630379429748e-05,
+      "loss": 2.938,
+      "step": 241500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1524852106254384e-05,
+      "loss": 2.9388,
+      "step": 242000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.1486666269533966e-05,
+      "loss": 2.9075,
+      "step": 242500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.144848043281355e-05,
+      "loss": 2.9184,
+      "step": 243000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1410370967766575e-05,
+      "loss": 2.9298,
+      "step": 243500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1372261502719595e-05,
+      "loss": 2.9295,
+      "step": 244000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.133407566599918e-05,
+      "loss": 2.9326,
+      "step": 244500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.129588982927876e-05,
+      "loss": 2.9216,
+      "step": 245000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.125770399255834e-05,
+      "loss": 2.9305,
+      "step": 245500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.121951815583793e-05,
+      "loss": 2.9247,
+      "step": 246000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1181332319117515e-05,
+      "loss": 2.9358,
+      "step": 246500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.11431464823971e-05,
+      "loss": 2.9264,
+      "step": 247000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.110496064567668e-05,
+      "loss": 2.9357,
+      "step": 247500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.10668511806297e-05,
+      "loss": 2.9111,
+      "step": 248000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.102866534390928e-05,
+      "loss": 2.9462,
+      "step": 248500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.099055587886231e-05,
+      "loss": 2.934,
+      "step": 249000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.095237004214189e-05,
+      "loss": 2.9237,
+      "step": 249500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0914184205421474e-05,
+      "loss": 2.9271,
+      "step": 250000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.087599836870106e-05,
+      "loss": 2.9421,
+      "step": 250500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.083781253198064e-05,
+      "loss": 2.9248,
+      "step": 251000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.079962669526022e-05,
+      "loss": 2.9299,
+      "step": 251500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.076151723021324e-05,
+      "loss": 2.9146,
+      "step": 252000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0723331393492824e-05,
+      "loss": 2.932,
+      "step": 252500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.068514555677241e-05,
+      "loss": 2.928,
+      "step": 253000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0646959720051996e-05,
+      "loss": 2.9297,
+      "step": 253500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.060877388333158e-05,
+      "loss": 2.9249,
+      "step": 254000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.057058804661116e-05,
+      "loss": 2.9362,
+      "step": 254500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0532402209890744e-05,
+      "loss": 2.9206,
+      "step": 255000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0494292744843767e-05,
+      "loss": 2.9183,
+      "step": 255500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.045610690812335e-05,
+      "loss": 2.9425,
+      "step": 256000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0417921071402932e-05,
+      "loss": 2.9311,
+      "step": 256500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0379735234682515e-05,
+      "loss": 2.9201,
+      "step": 257000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0341549397962098e-05,
+      "loss": 2.9353,
+      "step": 257500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0303363561241684e-05,
+      "loss": 2.934,
+      "step": 258000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0265330467868147e-05,
+      "loss": 2.9289,
+      "step": 258500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.022714463114773e-05,
+      "loss": 2.9142,
+      "step": 259000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0188958794427312e-05,
+      "loss": 2.9197,
+      "step": 259500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0150772957706895e-05,
+      "loss": 2.9232,
+      "step": 260000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.011258712098648e-05,
+      "loss": 2.9131,
+      "step": 260500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0074401284266064e-05,
+      "loss": 2.9209,
+      "step": 261000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0036215447545646e-05,
+      "loss": 2.937,
+      "step": 261500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.999802961082523e-05,
+      "loss": 2.9382,
+      "step": 262000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.995984377410481e-05,
+      "loss": 2.9268,
+      "step": 262500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9921657937384394e-05,
+      "loss": 2.9215,
+      "step": 263000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.988347210066398e-05,
+      "loss": 2.9335,
+      "step": 263500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9845286263943563e-05,
+      "loss": 2.9219,
+      "step": 264000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9807176798896586e-05,
+      "loss": 2.933,
+      "step": 264500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.976899096217617e-05,
+      "loss": 2.9231,
+      "step": 265000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.973080512545575e-05,
+      "loss": 2.905,
+      "step": 265500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9692619288735334e-05,
+      "loss": 2.933,
+      "step": 266000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9654433452014913e-05,
+      "loss": 2.9156,
+      "step": 266500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9616323986967936e-05,
+      "loss": 2.9238,
+      "step": 267000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.957821452192096e-05,
+      "loss": 2.9303,
+      "step": 267500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9540028685200545e-05,
+      "loss": 2.9368,
+      "step": 268000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9501842848480128e-05,
+      "loss": 2.9084,
+      "step": 268500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.946365701175971e-05,
+      "loss": 2.9127,
+      "step": 269000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9425471175039293e-05,
+      "loss": 2.9288,
+      "step": 269500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9387285338318876e-05,
+      "loss": 2.931,
+      "step": 270000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9349099501598458e-05,
+      "loss": 2.9236,
+      "step": 270500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9310913664878044e-05,
+      "loss": 2.9233,
+      "step": 271000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9272804199831067e-05,
+      "loss": 2.9109,
+      "step": 271500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.923469473478409e-05,
+      "loss": 2.928,
+      "step": 272000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9196508898063673e-05,
+      "loss": 2.9268,
+      "step": 272500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9158323061343256e-05,
+      "loss": 2.9147,
+      "step": 273000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.912013722462284e-05,
+      "loss": 2.9333,
+      "step": 273500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9081951387902424e-05,
+      "loss": 2.9311,
+      "step": 274000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9043841922855447e-05,
+      "loss": 2.9258,
+      "step": 274500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.900565608613503e-05,
+      "loss": 2.9322,
+      "step": 275000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8967470249414612e-05,
+      "loss": 2.9357,
+      "step": 275500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8929284412694195e-05,
+      "loss": 2.9125,
+      "step": 276000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.889109857597378e-05,
+      "loss": 2.9333,
+      "step": 276500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8852912739253364e-05,
+      "loss": 2.9326,
+      "step": 277000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8814726902532946e-05,
+      "loss": 2.9078,
+      "step": 277500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.877661743748597e-05,
+      "loss": 2.9337,
+      "step": 278000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8738431600765552e-05,
+      "loss": 2.9273,
+      "step": 278500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8700245764045135e-05,
+      "loss": 2.9157,
+      "step": 279000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.866205992732472e-05,
+      "loss": 2.9309,
+      "step": 279500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8623874090604303e-05,
+      "loss": 2.9329,
+      "step": 280000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8585688253883886e-05,
+      "loss": 2.9198,
+      "step": 280500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8547502417163462e-05,
+      "loss": 2.9368,
+      "step": 281000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8509316580443048e-05,
+      "loss": 2.9159,
+      "step": 281500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.847113074372263e-05,
+      "loss": 2.9144,
+      "step": 282000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8433021278675653e-05,
+      "loss": 2.9233,
+      "step": 282500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8394835441955236e-05,
+      "loss": 2.9155,
+      "step": 283000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.835664960523482e-05,
+      "loss": 2.914,
+      "step": 283500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.83184637685144e-05,
+      "loss": 2.9127,
+      "step": 284000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8280354303467428e-05,
+      "loss": 2.9158,
+      "step": 284500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.824216846674701e-05,
+      "loss": 2.9214,
+      "step": 285000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8204059001700033e-05,
+      "loss": 2.9325,
+      "step": 285500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8165873164979616e-05,
+      "loss": 2.9182,
+      "step": 286000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.81276873282592e-05,
+      "loss": 2.9233,
+      "step": 286500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8089501491538785e-05,
+      "loss": 2.923,
+      "step": 287000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8051315654818367e-05,
+      "loss": 2.9185,
+      "step": 287500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.801312981809795e-05,
+      "loss": 2.9361,
+      "step": 288000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7974943981377533e-05,
+      "loss": 2.9301,
+      "step": 288500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.7936758144657115e-05,
+      "loss": 2.9221,
+      "step": 289000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7898572307936698e-05,
+      "loss": 2.9119,
+      "step": 289500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7860386471216284e-05,
+      "loss": 2.9074,
+      "step": 290000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7822277006169307e-05,
+      "loss": 2.9194,
+      "step": 290500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.778409116944889e-05,
+      "loss": 2.926,
+      "step": 291000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7745905332728472e-05,
+      "loss": 2.914,
+      "step": 291500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7707719496008055e-05,
+      "loss": 2.9312,
+      "step": 292000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.766961003096108e-05,
+      "loss": 2.9159,
+      "step": 292500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7631424194240664e-05,
+      "loss": 2.9304,
+      "step": 293000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7593238357520246e-05,
+      "loss": 2.9204,
+      "step": 293500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.755512889247327e-05,
+      "loss": 2.9239,
+      "step": 294000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7516943055752852e-05,
+      "loss": 2.9218,
+      "step": 294500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7478757219032435e-05,
+      "loss": 2.9261,
+      "step": 295000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7440571382312014e-05,
+      "loss": 2.9211,
+      "step": 295500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7402385545591597e-05,
+      "loss": 2.9135,
+      "step": 296000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.736419970887118e-05,
+      "loss": 2.9291,
+      "step": 296500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7326013872150762e-05,
+      "loss": 2.9185,
+      "step": 297000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7287828035430348e-05,
+      "loss": 2.9062,
+      "step": 297500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.724964219870993e-05,
+      "loss": 2.9161,
+      "step": 298000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7211456361989513e-05,
+      "loss": 2.8969,
+      "step": 298500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7173346896942536e-05,
+      "loss": 2.9183,
+      "step": 299000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.713516106022212e-05,
+      "loss": 2.9488,
+      "step": 299500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7097051595175145e-05,
+      "loss": 2.9202,
+      "step": 300000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7058865758454728e-05,
+      "loss": 2.9093,
+      "step": 300500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.702067992173431e-05,
+      "loss": 2.921,
+      "step": 301000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6982494085013893e-05,
+      "loss": 2.9097,
+      "step": 301500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6944308248293476e-05,
+      "loss": 2.9005,
+      "step": 302000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6906122411573058e-05,
+      "loss": 2.9131,
+      "step": 302500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.686793657485264e-05,
+      "loss": 2.92,
+      "step": 303000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6829750738132227e-05,
+      "loss": 2.9241,
+      "step": 303500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.679164127308525e-05,
+      "loss": 2.9092,
+      "step": 304000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6753455436364833e-05,
+      "loss": 2.9069,
+      "step": 304500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6715269599644415e-05,
+      "loss": 2.9111,
+      "step": 305000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.667716013459744e-05,
+      "loss": 2.9046,
+      "step": 305500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6638974297877024e-05,
+      "loss": 2.9099,
+      "step": 306000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6600788461156607e-05,
+      "loss": 2.9146,
+      "step": 306500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.656260262443619e-05,
+      "loss": 2.9114,
+      "step": 307000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6524416787715772e-05,
+      "loss": 2.901,
+      "step": 307500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6486230950995355e-05,
+      "loss": 2.9266,
+      "step": 308000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6448045114274937e-05,
+      "loss": 2.9179,
+      "step": 308500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.640985927755452e-05,
+      "loss": 2.9222,
+      "step": 309000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6371673440834106e-05,
+      "loss": 2.9223,
+      "step": 309500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6333487604113682e-05,
+      "loss": 2.9074,
+      "step": 310000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6295378139066705e-05,
+      "loss": 2.9166,
+      "step": 310500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.625719230234629e-05,
+      "loss": 2.9198,
+      "step": 311000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6219006465625874e-05,
+      "loss": 2.905,
+      "step": 311500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6180820628905456e-05,
+      "loss": 2.9282,
+      "step": 312000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.614263479218504e-05,
+      "loss": 2.9163,
+      "step": 312500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.610444895546462e-05,
+      "loss": 2.9062,
+      "step": 313000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6066263118744204e-05,
+      "loss": 2.905,
+      "step": 313500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.602807728202379e-05,
+      "loss": 2.9101,
+      "step": 314000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5989967816976813e-05,
+      "loss": 2.9177,
+      "step": 314500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5951781980256396e-05,
+      "loss": 2.8948,
+      "step": 315000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.591359614353598e-05,
+      "loss": 2.9131,
+      "step": 315500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.587541030681556e-05,
+      "loss": 2.9116,
+      "step": 316000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5837224470095144e-05,
+      "loss": 2.9166,
+      "step": 316500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.579903863337473e-05,
+      "loss": 2.9167,
+      "step": 317000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5760929168327753e-05,
+      "loss": 2.9145,
+      "step": 317500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5722743331607335e-05,
+      "loss": 2.9055,
+      "step": 318000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5684557494886918e-05,
+      "loss": 2.9237,
+      "step": 318500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.564644802983994e-05,
+      "loss": 2.9188,
+      "step": 319000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5608262193119527e-05,
+      "loss": 2.9144,
+      "step": 319500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.557007635639911e-05,
+      "loss": 2.9132,
+      "step": 320000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5531890519678692e-05,
+      "loss": 2.9213,
+      "step": 320500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5493704682958275e-05,
+      "loss": 2.9261,
+      "step": 321000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5455518846237857e-05,
+      "loss": 2.9196,
+      "step": 321500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.541733300951744e-05,
+      "loss": 2.9098,
+      "step": 322000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5379147172797023e-05,
+      "loss": 2.9167,
+      "step": 322500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.534096133607661e-05,
+      "loss": 2.904,
+      "step": 323000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5302851871029632e-05,
+      "loss": 2.9123,
+      "step": 323500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5264666034309214e-05,
+      "loss": 2.9143,
+      "step": 324000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5226480197588794e-05,
+      "loss": 2.9077,
+      "step": 324500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5188294360868376e-05,
+      "loss": 2.9074,
+      "step": 325000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.515010852414796e-05,
+      "loss": 2.9127,
+      "step": 325500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5111999059100982e-05,
+      "loss": 2.9249,
+      "step": 326000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5073813222380565e-05,
+      "loss": 2.9186,
+      "step": 326500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5035627385660147e-05,
+      "loss": 2.9118,
+      "step": 327000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4997441548939733e-05,
+      "loss": 2.913,
+      "step": 327500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4959332083892756e-05,
+      "loss": 2.9089,
+      "step": 328000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.492114624717234e-05,
+      "loss": 2.9115,
+      "step": 328500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.488296041045192e-05,
+      "loss": 2.9055,
+      "step": 329000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4844774573731504e-05,
+      "loss": 2.9102,
+      "step": 329500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.480666510868453e-05,
+      "loss": 2.9125,
+      "step": 330000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4768555643637554e-05,
+      "loss": 2.9097,
+      "step": 330500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4730369806917136e-05,
+      "loss": 2.9227,
+      "step": 331000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.469218397019672e-05,
+      "loss": 2.8987,
+      "step": 331500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.46539981334763e-05,
+      "loss": 2.9095,
+      "step": 332000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4615812296755884e-05,
+      "loss": 2.9199,
+      "step": 332500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.457762646003547e-05,
+      "loss": 2.9084,
+      "step": 333000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4539440623315053e-05,
+      "loss": 2.8994,
+      "step": 333500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4501254786594635e-05,
+      "loss": 2.8996,
+      "step": 334000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.446314532154766e-05,
+      "loss": 2.9066,
+      "step": 334500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.442495948482724e-05,
+      "loss": 2.9143,
+      "step": 335000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4386773648106824e-05,
+      "loss": 2.9236,
+      "step": 335500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4348587811386406e-05,
+      "loss": 2.9171,
+      "step": 336000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.431040197466599e-05,
+      "loss": 2.9095,
+      "step": 336500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.427221613794557e-05,
+      "loss": 2.8998,
+      "step": 337000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4234030301225154e-05,
+      "loss": 2.9093,
+      "step": 337500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.419592083617818e-05,
+      "loss": 2.9058,
+      "step": 338000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4157734999457763e-05,
+      "loss": 2.8831,
+      "step": 338500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4119549162737346e-05,
+      "loss": 2.9209,
+      "step": 339000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.408136332601693e-05,
+      "loss": 2.889,
+      "step": 339500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.404325386096995e-05,
+      "loss": 2.8755,
+      "step": 340000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4005068024249534e-05,
+      "loss": 2.9066,
+      "step": 340500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.396688218752912e-05,
+      "loss": 2.9012,
+      "step": 341000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3928696350808703e-05,
+      "loss": 2.8974,
+      "step": 341500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3890510514088285e-05,
+      "loss": 2.9178,
+      "step": 342000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3852324677367865e-05,
+      "loss": 2.9024,
+      "step": 342500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3814138840647447e-05,
+      "loss": 2.9003,
+      "step": 343000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3775953003927033e-05,
+      "loss": 2.8987,
+      "step": 343500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3737843538880056e-05,
+      "loss": 2.9069,
+      "step": 344000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.369965770215964e-05,
+      "loss": 2.9123,
+      "step": 344500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.366147186543922e-05,
+      "loss": 2.9179,
+      "step": 345000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3623286028718804e-05,
+      "loss": 2.9179,
+      "step": 345500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3585100191998387e-05,
+      "loss": 2.9118,
+      "step": 346000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3546990726951413e-05,
+      "loss": 2.9015,
+      "step": 346500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3508804890230996e-05,
+      "loss": 2.9076,
+      "step": 347000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.347061905351058e-05,
+      "loss": 2.898,
+      "step": 347500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.343243321679016e-05,
+      "loss": 2.9125,
+      "step": 348000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3394247380069744e-05,
+      "loss": 2.8941,
+      "step": 348500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3356061543349326e-05,
+      "loss": 2.9063,
+      "step": 349000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.331795207830235e-05,
+      "loss": 2.8873,
+      "step": 349500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3279766241581932e-05,
+      "loss": 2.9007,
+      "step": 350000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3241580404861515e-05,
+      "loss": 2.9173,
+      "step": 350500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3203394568141097e-05,
+      "loss": 2.9069,
+      "step": 351000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3165208731420683e-05,
+      "loss": 2.9106,
+      "step": 351500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3127099266373706e-05,
+      "loss": 2.8918,
+      "step": 352000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.308891342965329e-05,
+      "loss": 2.8999,
+      "step": 352500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.305072759293287e-05,
+      "loss": 2.9054,
+      "step": 353000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3012541756212454e-05,
+      "loss": 2.9144,
+      "step": 353500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2974355919492037e-05,
+      "loss": 2.9171,
+      "step": 354000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2936246454445063e-05,
+      "loss": 2.8971,
+      "step": 354500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2898060617724646e-05,
+      "loss": 2.9032,
+      "step": 355000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.285995115267767e-05,
+      "loss": 2.9077,
+      "step": 355500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.282176531595725e-05,
+      "loss": 2.8971,
+      "step": 356000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2783579479236834e-05,
+      "loss": 2.9031,
+      "step": 356500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2745393642516417e-05,
+      "loss": 2.9073,
+      "step": 357000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2707207805796e-05,
+      "loss": 2.9042,
+      "step": 357500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2669021969075582e-05,
+      "loss": 2.8942,
+      "step": 358000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2630836132355165e-05,
+      "loss": 2.9205,
+      "step": 358500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2592650295634747e-05,
+      "loss": 2.9234,
+      "step": 359000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2554464458914333e-05,
+      "loss": 2.9042,
+      "step": 359500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2516354993867356e-05,
+      "loss": 2.9034,
+      "step": 360000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.247816915714694e-05,
+      "loss": 2.9083,
+      "step": 360500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.243998332042652e-05,
+      "loss": 2.8971,
+      "step": 361000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2401797483706104e-05,
+      "loss": 2.9104,
+      "step": 361500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2363611646985687e-05,
+      "loss": 2.9089,
+      "step": 362000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2325425810265273e-05,
+      "loss": 2.8966,
+      "step": 362500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2287316345218296e-05,
+      "loss": 2.9039,
+      "step": 363000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.224913050849788e-05,
+      "loss": 2.8973,
+      "step": 363500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2210944671777458e-05,
+      "loss": 2.8917,
+      "step": 364000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.217275883505704e-05,
+      "loss": 2.8913,
+      "step": 364500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2134649370010067e-05,
+      "loss": 2.9039,
+      "step": 365000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.209646353328965e-05,
+      "loss": 2.9069,
+      "step": 365500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2058277696569232e-05,
+      "loss": 2.9,
+      "step": 366000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2020091859848815e-05,
+      "loss": 2.9026,
+      "step": 366500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1981906023128397e-05,
+      "loss": 2.9028,
+      "step": 367000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1943796558081424e-05,
+      "loss": 2.9162,
+      "step": 367500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1905610721361006e-05,
+      "loss": 2.9066,
+      "step": 368000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.186742488464059e-05,
+      "loss": 2.9054,
+      "step": 368500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1829239047920172e-05,
+      "loss": 2.8995,
+      "step": 369000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1791053211199754e-05,
+      "loss": 2.9059,
+      "step": 369500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1752943746152777e-05,
+      "loss": 2.899,
+      "step": 370000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1714757909432363e-05,
+      "loss": 2.9038,
+      "step": 370500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1676572072711943e-05,
+      "loss": 2.9063,
+      "step": 371000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1638386235991525e-05,
+      "loss": 2.9067,
+      "step": 371500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.160027677094455e-05,
+      "loss": 2.9117,
+      "step": 372000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1562090934224134e-05,
+      "loss": 2.9134,
+      "step": 372500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1523905097503717e-05,
+      "loss": 2.899,
+      "step": 373000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.148579563245674e-05,
+      "loss": 2.8934,
+      "step": 373500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1447609795736323e-05,
+      "loss": 2.8826,
+      "step": 374000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1409423959015905e-05,
+      "loss": 2.8987,
+      "step": 374500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1371238122295488e-05,
+      "loss": 2.9168,
+      "step": 375000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1333052285575074e-05,
+      "loss": 2.9087,
+      "step": 375500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1294866448854657e-05,
+      "loss": 2.8964,
+      "step": 376000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.125668061213424e-05,
+      "loss": 2.8985,
+      "step": 376500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1218494775413822e-05,
+      "loss": 2.8997,
+      "step": 377000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1180385310366845e-05,
+      "loss": 2.9051,
+      "step": 377500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1142199473646427e-05,
+      "loss": 2.8842,
+      "step": 378000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.110401363692601e-05,
+      "loss": 2.9088,
+      "step": 378500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1065827800205593e-05,
+      "loss": 2.8956,
+      "step": 379000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1027641963485175e-05,
+      "loss": 2.8839,
+      "step": 379500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.098960887011164e-05,
+      "loss": 2.9021,
+      "step": 380000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0951423033391225e-05,
+      "loss": 2.8932,
+      "step": 380500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0913237196670808e-05,
+      "loss": 2.9065,
+      "step": 381000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.087505135995039e-05,
+      "loss": 2.9073,
+      "step": 381500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0836865523229973e-05,
+      "loss": 2.8905,
+      "step": 382000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0798679686509555e-05,
+      "loss": 2.9029,
+      "step": 382500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0760570221462582e-05,
+      "loss": 2.906,
+      "step": 383000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0722384384742164e-05,
+      "loss": 2.9054,
+      "step": 383500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0684198548021747e-05,
+      "loss": 2.8993,
+      "step": 384000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.064601271130133e-05,
+      "loss": 2.8958,
+      "step": 384500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0607826874580912e-05,
+      "loss": 2.9028,
+      "step": 385000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.056964103786049e-05,
+      "loss": 2.8852,
+      "step": 385500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0531455201140078e-05,
+      "loss": 2.9006,
+      "step": 386000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.049326936441966e-05,
+      "loss": 2.9022,
+      "step": 386500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0455083527699243e-05,
+      "loss": 2.8909,
+      "step": 387000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0416897690978825e-05,
+      "loss": 2.9045,
+      "step": 387500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.037886459760529e-05,
+      "loss": 2.8952,
+      "step": 388000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0340678760884875e-05,
+      "loss": 2.9026,
+      "step": 388500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0302492924164458e-05,
+      "loss": 2.9002,
+      "step": 389000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.026430708744404e-05,
+      "loss": 2.903,
+      "step": 389500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0226121250723623e-05,
+      "loss": 2.8986,
+      "step": 390000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0187935414003205e-05,
+      "loss": 2.8944,
+      "step": 390500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0149749577282788e-05,
+      "loss": 2.8973,
+      "step": 391000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0111563740562374e-05,
+      "loss": 2.9159,
+      "step": 391500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0073377903841957e-05,
+      "loss": 2.8995,
+      "step": 392000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0035192067121536e-05,
+      "loss": 2.8984,
+      "step": 392500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.999708260207456e-05,
+      "loss": 2.9002,
+      "step": 393000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.995889676535414e-05,
+      "loss": 2.8747,
+      "step": 393500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9920710928633728e-05,
+      "loss": 2.9051,
+      "step": 394000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.988252509191331e-05,
+      "loss": 2.8871,
+      "step": 394500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9844339255192893e-05,
+      "loss": 2.8948,
+      "step": 395000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9806229790145916e-05,
+      "loss": 2.9111,
+      "step": 395500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.97680439534255e-05,
+      "loss": 2.9075,
+      "step": 396000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.972985811670508e-05,
+      "loss": 2.8882,
+      "step": 396500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9691672279984667e-05,
+      "loss": 2.8866,
+      "step": 397000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.965356281493769e-05,
+      "loss": 2.8902,
+      "step": 397500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9615376978217273e-05,
+      "loss": 2.9079,
+      "step": 398000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9577191141496855e-05,
+      "loss": 2.8912,
+      "step": 398500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9539005304776438e-05,
+      "loss": 2.8829,
+      "step": 399000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9500819468056024e-05,
+      "loss": 2.9002,
+      "step": 399500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9462710003009044e-05,
+      "loss": 2.8936,
+      "step": 400000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9424524166288626e-05,
+      "loss": 2.889,
+      "step": 400500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.938633832956821e-05,
+      "loss": 2.895,
+      "step": 401000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.934815249284779e-05,
+      "loss": 2.8811,
+      "step": 401500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9309966656127378e-05,
+      "loss": 2.8853,
+      "step": 402000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.927178081940696e-05,
+      "loss": 2.9085,
+      "step": 402500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9233671354359983e-05,
+      "loss": 2.889,
+      "step": 403000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9195485517639566e-05,
+      "loss": 2.8878,
+      "step": 403500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.915729968091915e-05,
+      "loss": 2.9038,
+      "step": 404000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9119190215872175e-05,
+      "loss": 2.8864,
+      "step": 404500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9081004379151758e-05,
+      "loss": 2.8855,
+      "step": 405000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.904281854243134e-05,
+      "loss": 2.89,
+      "step": 405500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9004632705710923e-05,
+      "loss": 2.8858,
+      "step": 406000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8966446868990505e-05,
+      "loss": 2.8963,
+      "step": 406500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8928261032270088e-05,
+      "loss": 2.8985,
+      "step": 407000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.889015156722311e-05,
+      "loss": 2.9008,
+      "step": 407500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8851965730502694e-05,
+      "loss": 2.8938,
+      "step": 408000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8813779893782276e-05,
+      "loss": 2.8844,
+      "step": 408500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.877559405706186e-05,
+      "loss": 2.8951,
+      "step": 409000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.873740822034144e-05,
+      "loss": 2.8872,
+      "step": 409500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8699222383621028e-05,
+      "loss": 2.9104,
+      "step": 410000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.866103654690061e-05,
+      "loss": 2.9076,
+      "step": 410500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8622850710180193e-05,
+      "loss": 2.8942,
+      "step": 411000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8584664873459775e-05,
+      "loss": 2.8934,
+      "step": 411500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.85465554084128e-05,
+      "loss": 2.8946,
+      "step": 412000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8508445943365825e-05,
+      "loss": 2.8877,
+      "step": 412500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8470260106645408e-05,
+      "loss": 2.8834,
+      "step": 413000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.843207426992499e-05,
+      "loss": 2.8947,
+      "step": 413500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.839388843320457e-05,
+      "loss": 2.9052,
+      "step": 414000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8355702596484152e-05,
+      "loss": 2.8853,
+      "step": 414500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8317516759763738e-05,
+      "loss": 2.8927,
+      "step": 415000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.827933092304332e-05,
+      "loss": 2.891,
+      "step": 415500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8241145086322903e-05,
+      "loss": 2.8862,
+      "step": 416000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8203035621275926e-05,
+      "loss": 2.9023,
+      "step": 416500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.816492615622895e-05,
+      "loss": 2.8902,
+      "step": 417000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8126740319508536e-05,
+      "loss": 2.8958,
+      "step": 417500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8088554482788118e-05,
+      "loss": 2.9001,
+      "step": 418000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.80503686460677e-05,
+      "loss": 2.8917,
+      "step": 418500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8012182809347283e-05,
+      "loss": 2.8993,
+      "step": 419000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7973996972626866e-05,
+      "loss": 2.8722,
+      "step": 419500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.793581113590645e-05,
+      "loss": 2.8857,
+      "step": 420000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.789762529918603e-05,
+      "loss": 2.8907,
+      "step": 420500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7859439462465617e-05,
+      "loss": 2.8874,
+      "step": 421000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7821329997418637e-05,
+      "loss": 2.8826,
+      "step": 421500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.778314416069822e-05,
+      "loss": 2.8933,
+      "step": 422000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7744958323977802e-05,
+      "loss": 2.8928,
+      "step": 422500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7706772487257385e-05,
+      "loss": 2.9001,
+      "step": 423000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.766866302221041e-05,
+      "loss": 2.8968,
+      "step": 423500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7630477185489994e-05,
+      "loss": 2.8797,
+      "step": 424000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7592291348769577e-05,
+      "loss": 2.8971,
+      "step": 424500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.755410551204916e-05,
+      "loss": 2.8913,
+      "step": 425000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7515996047002182e-05,
+      "loss": 2.8795,
+      "step": 425500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7477810210281768e-05,
+      "loss": 2.8839,
+      "step": 426000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.743962437356135e-05,
+      "loss": 2.8956,
+      "step": 426500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7401438536840933e-05,
+      "loss": 2.9027,
+      "step": 427000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7363252700120516e-05,
+      "loss": 2.8974,
+      "step": 427500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.732514323507354e-05,
+      "loss": 2.8945,
+      "step": 428000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7286957398353122e-05,
+      "loss": 2.9098,
+      "step": 428500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7248771561632704e-05,
+      "loss": 2.883,
+      "step": 429000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7210585724912287e-05,
+      "loss": 2.8708,
+      "step": 429500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.717255263153875e-05,
+      "loss": 2.8872,
+      "step": 430000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7134366794818337e-05,
+      "loss": 2.8892,
+      "step": 430500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.709618095809792e-05,
+      "loss": 2.8899,
+      "step": 431000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7057995121377502e-05,
+      "loss": 2.9005,
+      "step": 431500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7019809284657084e-05,
+      "loss": 2.9046,
+      "step": 432000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6981623447936667e-05,
+      "loss": 2.8906,
+      "step": 432500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.694343761121625e-05,
+      "loss": 2.8844,
+      "step": 433000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6905328146169276e-05,
+      "loss": 2.886,
+      "step": 433500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.686714230944886e-05,
+      "loss": 2.8801,
+      "step": 434000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.682895647272844e-05,
+      "loss": 2.8866,
+      "step": 434500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6790770636008024e-05,
+      "loss": 2.8973,
+      "step": 435000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6752584799287603e-05,
+      "loss": 2.8827,
+      "step": 435500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.671439896256719e-05,
+      "loss": 2.8807,
+      "step": 436000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6676213125846772e-05,
+      "loss": 2.8896,
+      "step": 436500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6638027289126354e-05,
+      "loss": 2.9127,
+      "step": 437000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6599841452405937e-05,
+      "loss": 2.8906,
+      "step": 437500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.656165561568552e-05,
+      "loss": 2.8825,
+      "step": 438000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6523469778965102e-05,
+      "loss": 2.8864,
+      "step": 438500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.648536031391813e-05,
+      "loss": 2.8807,
+      "step": 439000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.644717447719771e-05,
+      "loss": 2.876,
+      "step": 439500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6408988640477294e-05,
+      "loss": 2.8779,
+      "step": 440000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6370802803756877e-05,
+      "loss": 2.8779,
+      "step": 440500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.63326933387099e-05,
+      "loss": 2.8612,
+      "step": 441000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6294507501989482e-05,
+      "loss": 2.877,
+      "step": 441500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6256321665269068e-05,
+      "loss": 2.8748,
+      "step": 442000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.621813582854865e-05,
+      "loss": 2.8853,
+      "step": 442500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.617994999182823e-05,
+      "loss": 2.8812,
+      "step": 443000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6141764155107813e-05,
+      "loss": 2.8962,
+      "step": 443500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6103578318387395e-05,
+      "loss": 2.8679,
+      "step": 444000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.606539248166698e-05,
+      "loss": 2.888,
+      "step": 444500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6027283016620004e-05,
+      "loss": 2.887,
+      "step": 445000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5989097179899587e-05,
+      "loss": 2.862,
+      "step": 445500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.595091134317917e-05,
+      "loss": 2.8708,
+      "step": 446000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5912725506458752e-05,
+      "loss": 2.8747,
+      "step": 446500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5874539669738335e-05,
+      "loss": 2.8728,
+      "step": 447000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5836506576364802e-05,
+      "loss": 2.8993,
+      "step": 447500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5798320739644385e-05,
+      "loss": 2.8729,
+      "step": 448000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5760134902923967e-05,
+      "loss": 2.8753,
+      "step": 448500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.572194906620355e-05,
+      "loss": 2.8661,
+      "step": 449000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5683763229483132e-05,
+      "loss": 2.884,
+      "step": 449500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5645653764436155e-05,
+      "loss": 2.8685,
+      "step": 450000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.560754429938918e-05,
+      "loss": 2.8903,
+      "step": 450500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.556935846266876e-05,
+      "loss": 2.8896,
+      "step": 451000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5531172625948344e-05,
+      "loss": 2.8597,
+      "step": 451500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.549298678922793e-05,
+      "loss": 2.8674,
+      "step": 452000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5454800952507512e-05,
+      "loss": 2.8695,
+      "step": 452500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5416615115787095e-05,
+      "loss": 2.8777,
+      "step": 453000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5378429279066678e-05,
+      "loss": 2.8771,
+      "step": 453500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.534024344234626e-05,
+      "loss": 2.8726,
+      "step": 454000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5302057605625843e-05,
+      "loss": 2.8752,
+      "step": 454500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.526394814057887e-05,
+      "loss": 2.8843,
+      "step": 455000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5225762303858452e-05,
+      "loss": 2.8742,
+      "step": 455500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5187576467138035e-05,
+      "loss": 2.8562,
+      "step": 456000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5149390630417617e-05,
+      "loss": 2.8864,
+      "step": 456500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5111204793697198e-05,
+      "loss": 2.8841,
+      "step": 457000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5073171700323663e-05,
+      "loss": 2.8735,
+      "step": 457500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5034985863603246e-05,
+      "loss": 2.9038,
+      "step": 458000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4996800026882829e-05,
+      "loss": 2.8835,
+      "step": 458500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4958614190162413e-05,
+      "loss": 2.8823,
+      "step": 459000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4920428353441996e-05,
+      "loss": 2.87,
+      "step": 459500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4882242516721578e-05,
+      "loss": 2.881,
+      "step": 460000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.484405668000116e-05,
+      "loss": 2.8769,
+      "step": 460500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4805870843280745e-05,
+      "loss": 2.8718,
+      "step": 461000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4767761378233768e-05,
+      "loss": 2.8665,
+      "step": 461500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4729575541513352e-05,
+      "loss": 2.8801,
+      "step": 462000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4691389704792935e-05,
+      "loss": 2.8907,
+      "step": 462500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4653203868072518e-05,
+      "loss": 2.8818,
+      "step": 463000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4615018031352102e-05,
+      "loss": 2.881,
+      "step": 463500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4576832194631685e-05,
+      "loss": 2.875,
+      "step": 464000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4538722729584706e-05,
+      "loss": 2.8751,
+      "step": 464500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4500536892864289e-05,
+      "loss": 2.8926,
+      "step": 465000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4462427427817313e-05,
+      "loss": 2.8695,
+      "step": 465500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4424241591096896e-05,
+      "loss": 2.8736,
+      "step": 466000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4386055754376479e-05,
+      "loss": 2.8801,
+      "step": 466500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4347869917656063e-05,
+      "loss": 2.8698,
+      "step": 467000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4309684080935646e-05,
+      "loss": 2.8834,
+      "step": 467500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4271498244215228e-05,
+      "loss": 2.873,
+      "step": 468000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.423331240749481e-05,
+      "loss": 2.8704,
+      "step": 468500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4195126570774395e-05,
+      "loss": 2.8837,
+      "step": 469000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4156940734053978e-05,
+      "loss": 2.8705,
+      "step": 469500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4118831269007002e-05,
+      "loss": 2.87,
+      "step": 470000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4080645432286585e-05,
+      "loss": 2.8712,
+      "step": 470500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4042459595566168e-05,
+      "loss": 2.8737,
+      "step": 471000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4004273758845749e-05,
+      "loss": 2.8653,
+      "step": 471500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3966087922125331e-05,
+      "loss": 2.8871,
+      "step": 472000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3927902085404914e-05,
+      "loss": 2.876,
+      "step": 472500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3889792620357939e-05,
+      "loss": 2.8805,
+      "step": 473000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3851606783637521e-05,
+      "loss": 2.8858,
+      "step": 473500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3813420946917106e-05,
+      "loss": 2.8757,
+      "step": 474000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3775235110196688e-05,
+      "loss": 2.8945,
+      "step": 474500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3737125645149711e-05,
+      "loss": 2.8875,
+      "step": 475000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3698939808429296e-05,
+      "loss": 2.8816,
+      "step": 475500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3660753971708878e-05,
+      "loss": 2.8713,
+      "step": 476000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.362256813498846e-05,
+      "loss": 2.8725,
+      "step": 476500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3584382298268045e-05,
+      "loss": 2.8841,
+      "step": 477000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3546272833221068e-05,
+      "loss": 2.8908,
+      "step": 477500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3508086996500653e-05,
+      "loss": 2.8792,
+      "step": 478000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3469901159780232e-05,
+      "loss": 2.8701,
+      "step": 478500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3431715323059816e-05,
+      "loss": 2.8802,
+      "step": 479000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.339360585801284e-05,
+      "loss": 2.8664,
+      "step": 479500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3355420021292422e-05,
+      "loss": 2.8688,
+      "step": 480000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3317234184572006e-05,
+      "loss": 2.8774,
+      "step": 480500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3279048347851589e-05,
+      "loss": 2.8815,
+      "step": 481000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3240862511131171e-05,
+      "loss": 2.8793,
+      "step": 481500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3202676674410756e-05,
+      "loss": 2.8776,
+      "step": 482000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3164567209363779e-05,
+      "loss": 2.8635,
+      "step": 482500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3126457744316804e-05,
+      "loss": 2.8859,
+      "step": 483000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3088271907596386e-05,
+      "loss": 2.8863,
+      "step": 483500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3050086070875969e-05,
+      "loss": 2.8826,
+      "step": 484000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3011900234155553e-05,
+      "loss": 2.8585,
+      "step": 484500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2973714397435136e-05,
+      "loss": 2.8644,
+      "step": 485000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2935604932388157e-05,
+      "loss": 2.8759,
+      "step": 485500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.289741909566774e-05,
+      "loss": 2.8763,
+      "step": 486000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2859233258947322e-05,
+      "loss": 2.8676,
+      "step": 486500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2821047422226907e-05,
+      "loss": 2.8728,
+      "step": 487000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.278286158550649e-05,
+      "loss": 2.8687,
+      "step": 487500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2744675748786072e-05,
+      "loss": 2.8648,
+      "step": 488000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2706489912065656e-05,
+      "loss": 2.8762,
+      "step": 488500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2668304075345239e-05,
+      "loss": 2.8739,
+      "step": 489000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2630194610298262e-05,
+      "loss": 2.8765,
+      "step": 489500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2592008773577846e-05,
+      "loss": 2.872,
+      "step": 490000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2553822936857429e-05,
+      "loss": 2.8714,
+      "step": 490500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2515637100137011e-05,
+      "loss": 2.8745,
+      "step": 491000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2477451263416594e-05,
+      "loss": 2.8844,
+      "step": 491500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2439265426696177e-05,
+      "loss": 2.8768,
+      "step": 492000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2401155961649201e-05,
+      "loss": 2.8847,
+      "step": 492500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2362970124928784e-05,
+      "loss": 2.8752,
+      "step": 493000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2324784288208367e-05,
+      "loss": 2.8798,
+      "step": 493500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2286598451487951e-05,
+      "loss": 2.8663,
+      "step": 494000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2248488986440974e-05,
+      "loss": 2.882,
+      "step": 494500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2210303149720557e-05,
+      "loss": 2.8649,
+      "step": 495000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.217219368467358e-05,
+      "loss": 2.8778,
+      "step": 495500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2134007847953164e-05,
+      "loss": 2.8846,
+      "step": 496000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2095822011232747e-05,
+      "loss": 2.8669,
+      "step": 496500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.205763617451233e-05,
+      "loss": 2.8801,
+      "step": 497000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2019450337791912e-05,
+      "loss": 2.8864,
+      "step": 497500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1981340872744937e-05,
+      "loss": 2.8755,
+      "step": 498000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.194323140769796e-05,
+      "loss": 2.8852,
+      "step": 498500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1905045570977542e-05,
+      "loss": 2.8752,
+      "step": 499000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1866859734257125e-05,
+      "loss": 2.8749,
+      "step": 499500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.182867389753671e-05,
+      "loss": 2.8631,
+      "step": 500000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1790488060816292e-05,
+      "loss": 2.8735,
+      "step": 500500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1752302224095875e-05,
+      "loss": 2.8639,
+      "step": 501000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1714116387375459e-05,
+      "loss": 2.8724,
+      "step": 501500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.167593055065504e-05,
+      "loss": 2.8733,
+      "step": 502000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1637744713934622e-05,
+      "loss": 2.868,
+      "step": 502500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1599635248887647e-05,
+      "loss": 2.8647,
+      "step": 503000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.156144941216723e-05,
+      "loss": 2.8838,
+      "step": 503500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1523263575446812e-05,
+      "loss": 2.8687,
+      "step": 504000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1485077738726397e-05,
+      "loss": 2.88,
+      "step": 504500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.144689190200598e-05,
+      "loss": 2.864,
+      "step": 505000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1408706065285562e-05,
+      "loss": 2.8706,
+      "step": 505500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1370520228565145e-05,
+      "loss": 2.8707,
+      "step": 506000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1332334391844727e-05,
+      "loss": 2.8737,
+      "step": 506500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1294148555124311e-05,
+      "loss": 2.873,
+      "step": 507000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1256039090077335e-05,
+      "loss": 2.8766,
+      "step": 507500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1217853253356917e-05,
+      "loss": 2.8735,
+      "step": 508000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1179667416636501e-05,
+      "loss": 2.8855,
+      "step": 508500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1141557951589523e-05,
+      "loss": 2.8702,
+      "step": 509000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1103372114869107e-05,
+      "loss": 2.8656,
+      "step": 509500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.106518627814869e-05,
+      "loss": 2.8778,
+      "step": 510000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1027000441428272e-05,
+      "loss": 2.8759,
+      "step": 510500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0988814604707857e-05,
+      "loss": 2.8693,
+      "step": 511000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.095062876798744e-05,
+      "loss": 2.8658,
+      "step": 511500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0912442931267022e-05,
+      "loss": 2.878,
+      "step": 512000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0874333466220045e-05,
+      "loss": 2.8662,
+      "step": 512500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0836147629499628e-05,
+      "loss": 2.8782,
+      "step": 513000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0797961792779212e-05,
+      "loss": 2.8764,
+      "step": 513500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0759775956058795e-05,
+      "loss": 2.8712,
+      "step": 514000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0721590119338377e-05,
+      "loss": 2.8644,
+      "step": 514500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.068340428261796e-05,
+      "loss": 2.8682,
+      "step": 515000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0645218445897544e-05,
+      "loss": 2.8714,
+      "step": 515500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0607032609177127e-05,
+      "loss": 2.8737,
+      "step": 516000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.056892314413015e-05,
+      "loss": 2.864,
+      "step": 516500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0530737307409732e-05,
+      "loss": 2.8838,
+      "step": 517000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0492627842362757e-05,
+      "loss": 2.8732,
+      "step": 517500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.045444200564234e-05,
+      "loss": 2.8677,
+      "step": 518000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0416256168921922e-05,
+      "loss": 2.8737,
+      "step": 518500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0378070332201507e-05,
+      "loss": 2.8911,
+      "step": 519000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.033988449548109e-05,
+      "loss": 2.8778,
+      "step": 519500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.030169865876067e-05,
+      "loss": 2.8886,
+      "step": 520000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0263512822040255e-05,
+      "loss": 2.8779,
+      "step": 520500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0225326985319837e-05,
+      "loss": 2.8831,
+      "step": 521000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0187217520272862e-05,
+      "loss": 2.8716,
+      "step": 521500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0149031683552445e-05,
+      "loss": 2.8692,
+      "step": 522000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0110845846832027e-05,
+      "loss": 2.8724,
+      "step": 522500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.007266001011161e-05,
+      "loss": 2.8732,
+      "step": 523000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0034550545064633e-05,
+      "loss": 2.8661,
+      "step": 523500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.996364708344216e-06,
+      "loss": 2.8756,
+      "step": 524000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.9581788716238e-06,
+      "loss": 2.8812,
+      "step": 524500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.919993034903382e-06,
+      "loss": 2.8799,
+      "step": 525000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.881807198182965e-06,
+      "loss": 2.8569,
+      "step": 525500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.84362136146255e-06,
+      "loss": 2.8747,
+      "step": 526000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.805435524742132e-06,
+      "loss": 2.8755,
+      "step": 526500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.767326059695155e-06,
+      "loss": 2.8676,
+      "step": 527000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.729140222974738e-06,
+      "loss": 2.8787,
+      "step": 527500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.69095438625432e-06,
+      "loss": 2.8893,
+      "step": 528000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.652768549533905e-06,
+      "loss": 2.8709,
+      "step": 528500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.614582712813487e-06,
+      "loss": 2.8712,
+      "step": 529000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.57639687609307e-06,
+      "loss": 2.8716,
+      "step": 529500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.538211039372654e-06,
+      "loss": 2.8617,
+      "step": 530000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.500025202652237e-06,
+      "loss": 2.8726,
+      "step": 530500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.46191573760526e-06,
+      "loss": 2.8771,
+      "step": 531000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.423729900884843e-06,
+      "loss": 2.8594,
+      "step": 531500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.385544064164425e-06,
+      "loss": 2.874,
+      "step": 532000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.347358227444008e-06,
+      "loss": 2.8649,
+      "step": 532500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.309248762397033e-06,
+      "loss": 2.867,
+      "step": 533000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.271062925676615e-06,
+      "loss": 2.8719,
+      "step": 533500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.2328770889562e-06,
+      "loss": 2.878,
+      "step": 534000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.19469125223578e-06,
+      "loss": 2.863,
+      "step": 534500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.156505415515363e-06,
+      "loss": 2.8588,
+      "step": 535000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.118319578794947e-06,
+      "loss": 2.857,
+      "step": 535500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.08013374207453e-06,
+      "loss": 2.8779,
+      "step": 536000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.042024277027555e-06,
+      "loss": 2.8762,
+      "step": 536500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.003991183654018e-06,
+      "loss": 2.8794,
+      "step": 537000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.965805346933603e-06,
+      "loss": 2.8494,
+      "step": 537500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.927619510213183e-06,
+      "loss": 2.8637,
+      "step": 538000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.889433673492766e-06,
+      "loss": 2.8686,
+      "step": 538500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.85124783677235e-06,
+      "loss": 2.8768,
+      "step": 539000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.813062000051933e-06,
+      "loss": 2.8712,
+      "step": 539500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.774876163331516e-06,
+      "loss": 2.8793,
+      "step": 540000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.7366903266111e-06,
+      "loss": 2.8766,
+      "step": 540500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.698504489890683e-06,
+      "loss": 2.8603,
+      "step": 541000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.660318653170263e-06,
+      "loss": 2.8675,
+      "step": 541500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.622132816449848e-06,
+      "loss": 2.8774,
+      "step": 542000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.58394697972943e-06,
+      "loss": 2.8673,
+      "step": 542500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.545761143009013e-06,
+      "loss": 2.8881,
+      "step": 543000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.507651677962038e-06,
+      "loss": 2.8799,
+      "step": 543500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.46946584124162e-06,
+      "loss": 2.886,
+      "step": 544000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.431280004521205e-06,
+      "loss": 2.8751,
+      "step": 544500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.393170539474226e-06,
+      "loss": 2.8702,
+      "step": 545000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.35498470275381e-06,
+      "loss": 2.868,
+      "step": 545500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.316798866033393e-06,
+      "loss": 2.8763,
+      "step": 546000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.278613029312976e-06,
+      "loss": 2.8677,
+      "step": 546500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.240427192592558e-06,
+      "loss": 2.8656,
+      "step": 547000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.202241355872143e-06,
+      "loss": 2.8698,
+      "step": 547500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.164055519151725e-06,
+      "loss": 2.8886,
+      "step": 548000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.125869682431308e-06,
+      "loss": 2.8753,
+      "step": 548500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.087760217384331e-06,
+      "loss": 2.8525,
+      "step": 549000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.049574380663914e-06,
+      "loss": 2.8711,
+      "step": 549500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.011388543943498e-06,
+      "loss": 2.8686,
+      "step": 550000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.97320270722308e-06,
+      "loss": 2.8752,
+      "step": 550500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.935093242176105e-06,
+      "loss": 2.8837,
+      "step": 551000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.896907405455688e-06,
+      "loss": 2.8557,
+      "step": 551500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.85872156873527e-06,
+      "loss": 2.854,
+      "step": 552000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.820535732014853e-06,
+      "loss": 2.8626,
+      "step": 552500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.782349895294436e-06,
+      "loss": 2.8747,
+      "step": 553000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.74424043024746e-06,
+      "loss": 2.8841,
+      "step": 553500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.706130965200484e-06,
+      "loss": 2.8718,
+      "step": 554000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.667945128480066e-06,
+      "loss": 2.882,
+      "step": 554500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.62975929175965e-06,
+      "loss": 2.8677,
+      "step": 555000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.591573455039233e-06,
+      "loss": 2.8756,
+      "step": 555500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.553463989992255e-06,
+      "loss": 2.8572,
+      "step": 556000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.515278153271839e-06,
+      "loss": 2.8874,
+      "step": 556500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.477092316551422e-06,
+      "loss": 2.8611,
+      "step": 557000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.438906479831005e-06,
+      "loss": 2.8718,
+      "step": 557500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.400720643110588e-06,
+      "loss": 2.8793,
+      "step": 558000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.362534806390171e-06,
+      "loss": 2.8439,
+      "step": 558500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.3243489696697544e-06,
+      "loss": 2.882,
+      "step": 559000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.286163132949336e-06,
+      "loss": 2.8533,
+      "step": 559500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.24797729622892e-06,
+      "loss": 2.8671,
+      "step": 560000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.209791459508502e-06,
+      "loss": 2.86,
+      "step": 560500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.171681994461526e-06,
+      "loss": 2.8854,
+      "step": 561000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.13349615774111e-06,
+      "loss": 2.8684,
+      "step": 561500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.095310321020693e-06,
+      "loss": 2.8834,
+      "step": 562000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.057124484300276e-06,
+      "loss": 2.8763,
+      "step": 562500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.0189386475798575e-06,
+      "loss": 2.8665,
+      "step": 563000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.980752810859441e-06,
+      "loss": 2.8797,
+      "step": 563500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.942566974139024e-06,
+      "loss": 2.8663,
+      "step": 564000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.904457509092048e-06,
+      "loss": 2.8597,
+      "step": 564500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.866271672371631e-06,
+      "loss": 2.8752,
+      "step": 565000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8280858356512145e-06,
+      "loss": 2.869,
+      "step": 565500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.789976370604238e-06,
+      "loss": 2.8643,
+      "step": 566000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.75179053388382e-06,
+      "loss": 2.8726,
+      "step": 566500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.713604697163403e-06,
+      "loss": 2.8791,
+      "step": 567000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.675418860442986e-06,
+      "loss": 2.8717,
+      "step": 567500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.63723302372257e-06,
+      "loss": 2.8687,
+      "step": 568000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.599047187002152e-06,
+      "loss": 2.8728,
+      "step": 568500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.560861350281736e-06,
+      "loss": 2.8876,
+      "step": 569000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.522675513561318e-06,
+      "loss": 2.8868,
+      "step": 569500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.484566048514343e-06,
+      "loss": 2.8505,
+      "step": 570000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.446380211793925e-06,
+      "loss": 2.8609,
+      "step": 570500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.4081943750735076e-06,
+      "loss": 2.8777,
+      "step": 571000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.370008538353091e-06,
+      "loss": 2.8744,
+      "step": 571500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.331822701632674e-06,
+      "loss": 2.8532,
+      "step": 572000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.293713236585698e-06,
+      "loss": 2.8628,
+      "step": 572500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.255527399865281e-06,
+      "loss": 2.8738,
+      "step": 573000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.217341563144864e-06,
+      "loss": 2.8693,
+      "step": 573500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.179155726424446e-06,
+      "loss": 2.858,
+      "step": 574000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.141046261377471e-06,
+      "loss": 2.8565,
+      "step": 574500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.102860424657053e-06,
+      "loss": 2.8584,
+      "step": 575000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.064674587936636e-06,
+      "loss": 2.8608,
+      "step": 575500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.02648875121622e-06,
+      "loss": 2.8597,
+      "step": 576000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9883029144958015e-06,
+      "loss": 2.8669,
+      "step": 576500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.9501934494488255e-06,
+      "loss": 2.879,
+      "step": 577000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.912007612728409e-06,
+      "loss": 2.8657,
+      "step": 577500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.873821776007992e-06,
+      "loss": 2.8567,
+      "step": 578000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.835635939287574e-06,
+      "loss": 2.8667,
+      "step": 578500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.797450102567158e-06,
+      "loss": 2.8672,
+      "step": 579000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.7593406375201816e-06,
+      "loss": 2.8713,
+      "step": 579500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.721231172473205e-06,
+      "loss": 2.8541,
+      "step": 580000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.683045335752788e-06,
+      "loss": 2.8715,
+      "step": 580500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6448594990323716e-06,
+      "loss": 2.8735,
+      "step": 581000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.606673662311954e-06,
+      "loss": 2.8524,
+      "step": 581500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.568487825591537e-06,
+      "loss": 2.8709,
+      "step": 582000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.53030198887112e-06,
+      "loss": 2.8849,
+      "step": 582500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.492116152150703e-06,
+      "loss": 2.8748,
+      "step": 583000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.453930315430286e-06,
+      "loss": 2.859,
+      "step": 583500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.415744478709869e-06,
+      "loss": 2.8575,
+      "step": 584000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.377635013662893e-06,
+      "loss": 2.8672,
+      "step": 584500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.3394491769424755e-06,
+      "loss": 2.8736,
+      "step": 585000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.301263340222058e-06,
+      "loss": 2.8841,
+      "step": 585500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.263077503501642e-06,
+      "loss": 2.8585,
+      "step": 586000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.224891666781224e-06,
+      "loss": 2.8619,
+      "step": 586500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.186782201734248e-06,
+      "loss": 2.8692,
+      "step": 587000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.148596365013831e-06,
+      "loss": 2.8741,
+      "step": 587500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.110410528293414e-06,
+      "loss": 2.8699,
+      "step": 588000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.072224691572997e-06,
+      "loss": 2.8906,
+      "step": 588500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.03403885485258e-06,
+      "loss": 2.868,
+      "step": 589000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.995853018132163e-06,
+      "loss": 2.879,
+      "step": 589500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.9576671814117455e-06,
+      "loss": 2.8647,
+      "step": 590000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.91955771636477e-06,
+      "loss": 2.8694,
+      "step": 590500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.881371879644352e-06,
+      "loss": 2.8565,
+      "step": 591000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8431860429239355e-06,
+      "loss": 2.8694,
+      "step": 591500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.805000206203519e-06,
+      "loss": 2.8645,
+      "step": 592000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.766814369483102e-06,
+      "loss": 2.8634,
+      "step": 592500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.728704904436125e-06,
+      "loss": 2.8727,
+      "step": 593000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.690519067715708e-06,
+      "loss": 2.8699,
+      "step": 593500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.652333230995292e-06,
+      "loss": 2.8762,
+      "step": 594000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.614147394274873e-06,
+      "loss": 2.8448,
+      "step": 594500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.576037929227898e-06,
+      "loss": 2.8731,
+      "step": 595000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.537852092507481e-06,
+      "loss": 2.8604,
+      "step": 595500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.499666255787064e-06,
+      "loss": 2.8649,
+      "step": 596000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.461480419066647e-06,
+      "loss": 2.8689,
+      "step": 596500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.423370954019671e-06,
+      "loss": 2.8563,
+      "step": 597000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3851851172992534e-06,
+      "loss": 2.848,
+      "step": 597500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.346999280578836e-06,
+      "loss": 2.8708,
+      "step": 598000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3088134438584195e-06,
+      "loss": 2.8628,
+      "step": 598500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.270627607138002e-06,
+      "loss": 2.8624,
+      "step": 599000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.232441770417586e-06,
+      "loss": 2.8615,
+      "step": 599500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.194332305370609e-06,
+      "loss": 2.8647,
+      "step": 600000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.156222840323633e-06,
+      "loss": 2.8823,
+      "step": 600500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.118037003603216e-06,
+      "loss": 2.876,
+      "step": 601000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.079851166882799e-06,
+      "loss": 2.8604,
+      "step": 601500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.041665330162381e-06,
+      "loss": 2.8604,
+      "step": 602000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.003479493441965e-06,
+      "loss": 2.8738,
+      "step": 602500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.965293656721547e-06,
+      "loss": 2.8762,
+      "step": 603000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.92710782000113e-06,
+      "loss": 2.8724,
+      "step": 603500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8889219832807135e-06,
+      "loss": 2.8547,
+      "step": 604000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.850812518233737e-06,
+      "loss": 2.8582,
+      "step": 604500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.8126266815133204e-06,
+      "loss": 2.8469,
+      "step": 605000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.774440844792903e-06,
+      "loss": 2.8556,
+      "step": 605500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.736255008072486e-06,
+      "loss": 2.8612,
+      "step": 606000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.698069171352069e-06,
+      "loss": 2.8545,
+      "step": 606500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6599597063050926e-06,
+      "loss": 2.8754,
+      "step": 607000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6217738695846757e-06,
+      "loss": 2.8665,
+      "step": 607500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5835880328642587e-06,
+      "loss": 2.8622,
+      "step": 608000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5454021961438418e-06,
+      "loss": 2.8632,
+      "step": 608500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5072163594234244e-06,
+      "loss": 2.8667,
+      "step": 609000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4690305227030074e-06,
+      "loss": 2.864,
+      "step": 609500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.4308446859825904e-06,
+      "loss": 2.8696,
+      "step": 610000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.392735220935615e-06,
+      "loss": 2.8652,
+      "step": 610500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.354549384215197e-06,
+      "loss": 2.8669,
+      "step": 611000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.31636354749478e-06,
+      "loss": 2.8779,
+      "step": 611500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.2781777107743635e-06,
+      "loss": 2.8698,
+      "step": 612000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2400682457273866e-06,
+      "loss": 2.8609,
+      "step": 612500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2018824090069696e-06,
+      "loss": 2.8599,
+      "step": 613000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.163696572286553e-06,
+      "loss": 2.8513,
+      "step": 613500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.125510735566136e-06,
+      "loss": 2.8609,
+      "step": 614000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0873248988457187e-06,
+      "loss": 2.869,
+      "step": 614500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0491390621253018e-06,
+      "loss": 2.868,
+      "step": 615000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.010953225404885e-06,
+      "loss": 2.8761,
+      "step": 615500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9727673886844674e-06,
+      "loss": 2.8649,
+      "step": 616000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.9346579236374914e-06,
+      "loss": 2.8623,
+      "step": 616500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8965484585905153e-06,
+      "loss": 2.8666,
+      "step": 617000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.858362621870098e-06,
+      "loss": 2.865,
+      "step": 617500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.820176785149681e-06,
+      "loss": 2.8652,
+      "step": 618000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.781990948429264e-06,
+      "loss": 2.8627,
+      "step": 618500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.743805111708847e-06,
+      "loss": 2.8776,
+      "step": 619000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.7056192749884297e-06,
+      "loss": 2.8645,
+      "step": 619500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.667433438268013e-06,
+      "loss": 2.8597,
+      "step": 620000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6292476015475957e-06,
+      "loss": 2.8744,
+      "step": 620500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5910617648271783e-06,
+      "loss": 2.8736,
+      "step": 621000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5529522997802027e-06,
+      "loss": 2.8753,
+      "step": 621500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5147664630597853e-06,
+      "loss": 2.8594,
+      "step": 622000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4765806263393684e-06,
+      "loss": 2.8478,
+      "step": 622500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4383947896189514e-06,
+      "loss": 2.8637,
+      "step": 623000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.4002853245719753e-06,
+      "loss": 2.8586,
+      "step": 623500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.362175859524999e-06,
+      "loss": 2.8623,
+      "step": 624000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.3239900228045815e-06,
+      "loss": 2.8752,
+      "step": 624500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.285804186084165e-06,
+      "loss": 2.846,
+      "step": 625000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2476183493637476e-06,
+      "loss": 2.8619,
+      "step": 625500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2094325126433306e-06,
+      "loss": 2.8637,
+      "step": 626000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1712466759229136e-06,
+      "loss": 2.8661,
+      "step": 626500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1330608392024967e-06,
+      "loss": 2.8736,
+      "step": 627000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0948750024820793e-06,
+      "loss": 2.8538,
+      "step": 627500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0566891657616627e-06,
+      "loss": 2.8636,
+      "step": 628000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.0185033290412454e-06,
+      "loss": 2.8624,
+      "step": 628500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9803174923208284e-06,
+      "loss": 2.8594,
+      "step": 629000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9421316556004114e-06,
+      "loss": 2.8726,
+      "step": 629500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.903945818879994e-06,
+      "loss": 2.8627,
+      "step": 630000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.865836353833018e-06,
+      "loss": 2.8626,
+      "step": 630500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8276505171126008e-06,
+      "loss": 2.8598,
+      "step": 631000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.789541052065625e-06,
+      "loss": 2.8711,
+      "step": 631500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7513552153452076e-06,
+      "loss": 2.8812,
+      "step": 632000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.7131693786247908e-06,
+      "loss": 2.8665,
+      "step": 632500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6749835419043737e-06,
+      "loss": 2.8603,
+      "step": 633000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6367977051839567e-06,
+      "loss": 2.8574,
+      "step": 633500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5986882401369804e-06,
+      "loss": 2.8613,
+      "step": 634000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.560578775090004e-06,
+      "loss": 2.8771,
+      "step": 634500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.522392938369587e-06,
+      "loss": 2.8468,
+      "step": 635000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.48420710164917e-06,
+      "loss": 2.8653,
+      "step": 635500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4460212649287528e-06,
+      "loss": 2.8745,
+      "step": 636000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.4078354282083359e-06,
+      "loss": 2.8668,
+      "step": 636500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.369649591487919e-06,
+      "loss": 2.8697,
+      "step": 637000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3314637547675017e-06,
+      "loss": 2.8665,
+      "step": 637500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2932779180470848e-06,
+      "loss": 2.8545,
+      "step": 638000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2550920813266676e-06,
+      "loss": 2.8481,
+      "step": 638500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2169062446062506e-06,
+      "loss": 2.8443,
+      "step": 639000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1787967795592744e-06,
+      "loss": 2.8786,
+      "step": 639500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1406109428388572e-06,
+      "loss": 2.8738,
+      "step": 640000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1024251061184402e-06,
+      "loss": 2.8769,
+      "step": 640500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0642392693980233e-06,
+      "loss": 2.8594,
+      "step": 641000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.026053432677606e-06,
+      "loss": 2.8775,
+      "step": 641500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.878675959571891e-07,
+      "loss": 2.8724,
+      "step": 642000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.49758130910213e-07,
+      "loss": 2.8658,
+      "step": 642500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 9.115722941897959e-07,
+      "loss": 2.8642,
+      "step": 643000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.733864574693788e-07,
+      "loss": 2.8659,
+      "step": 643500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.352006207489619e-07,
+      "loss": 2.8654,
+      "step": 644000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.970147840285448e-07,
+      "loss": 2.854,
+      "step": 644500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.588289473081276e-07,
+      "loss": 2.8683,
+      "step": 645000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.206431105877107e-07,
+      "loss": 2.8637,
+      "step": 645500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.825336455407344e-07,
+      "loss": 2.8665,
+      "step": 646000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.443478088203173e-07,
+      "loss": 2.8712,
+      "step": 646500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.061619720999003e-07,
+      "loss": 2.8707,
+      "step": 647000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.679761353794832e-07,
+      "loss": 2.8581,
+      "step": 647500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.29866670332507e-07,
+      "loss": 2.8803,
+      "step": 648000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.9168083361209e-07,
+      "loss": 2.8674,
+      "step": 648500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.534949968916729e-07,
+      "loss": 2.8539,
+      "step": 649000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.153091601712559e-07,
+      "loss": 2.8571,
+      "step": 649500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.771233234508388e-07,
+      "loss": 2.8651,
+      "step": 650000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.3893748673042174e-07,
+      "loss": 2.8751,
+      "step": 650500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.007516500100047e-07,
+      "loss": 2.8613,
+      "step": 651000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.6256581328958765e-07,
+      "loss": 2.8795,
+      "step": 651500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.244563482426114e-07,
+      "loss": 2.8593,
+      "step": 652000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.8627051152219437e-07,
+      "loss": 2.8592,
+      "step": 652500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.4808467480177733e-07,
+      "loss": 2.8647,
+      "step": 653000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.0989883808136027e-07,
+      "loss": 2.864,
+      "step": 653500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.18657447078249e-08,
+      "loss": 2.8679,
+      "step": 654000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.367990798740784e-08,
+      "loss": 2.8587,
+      "step": 654500
+    },
+    {
+      "epoch": 3.0,
+      "step": 654693,
+      "total_flos": 2.737055461146624e+18,
+      "train_loss": 2.9363299626561012,
+      "train_runtime": 989713.3533,
+      "train_samples_per_second": 5.292,
+      "train_steps_per_second": 0.661
+    }
+  ],
+  "max_steps": 654693,
+  "num_train_epochs": 3,
+  "total_flos": 2.737055461146624e+18,
+  "trial_name": null,
+  "trial_params": null
+}