diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,8221 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "global_step": 683205,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.996340776194554e-05,
+      "loss": 1.7875,
+      "step": 500
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.9926815523891076e-05,
+      "loss": 1.7396,
+      "step": 1000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.989022328583661e-05,
+      "loss": 1.7118,
+      "step": 1500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.985363104778214e-05,
+      "loss": 1.7065,
+      "step": 2000
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.981703880972768e-05,
+      "loss": 1.698,
+      "step": 2500
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.978044657167322e-05,
+      "loss": 1.6919,
+      "step": 3000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.974385433361876e-05,
+      "loss": 1.6989,
+      "step": 3500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.970726209556429e-05,
+      "loss": 1.6784,
+      "step": 4000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.967066985750983e-05,
+      "loss": 1.6807,
+      "step": 4500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.9634077619455364e-05,
+      "loss": 1.6871,
+      "step": 5000
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.95974853814009e-05,
+      "loss": 1.6708,
+      "step": 5500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.956089314334644e-05,
+      "loss": 1.6594,
+      "step": 6000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.952430090529197e-05,
+      "loss": 1.6701,
+      "step": 6500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.948770866723751e-05,
+      "loss": 1.6653,
+      "step": 7000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.9451116429183045e-05,
+      "loss": 1.6653,
+      "step": 7500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.941452419112858e-05,
+      "loss": 1.6667,
+      "step": 8000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.937793195307411e-05,
+      "loss": 1.6589,
+      "step": 8500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.934133971501965e-05,
+      "loss": 1.6673,
+      "step": 9000
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.930474747696519e-05,
+      "loss": 1.6612,
+      "step": 9500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.9268155238910726e-05,
+      "loss": 1.649,
+      "step": 10000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.923156300085626e-05,
+      "loss": 1.6415,
+      "step": 10500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.919497076280179e-05,
+      "loss": 1.6449,
+      "step": 11000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.915837852474733e-05,
+      "loss": 1.6472,
+      "step": 11500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9121786286692866e-05,
+      "loss": 1.6556,
+      "step": 12000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.9085194048638406e-05,
+      "loss": 1.644,
+      "step": 12500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9048601810583947e-05,
+      "loss": 1.6387,
+      "step": 13000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.901200957252948e-05,
+      "loss": 1.6533,
+      "step": 13500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8975417334475013e-05,
+      "loss": 1.6425,
+      "step": 14000
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.893882509642055e-05,
+      "loss": 1.6375,
+      "step": 14500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.890223285836609e-05,
+      "loss": 1.652,
+      "step": 15000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.886564062031162e-05,
+      "loss": 1.6382,
+      "step": 15500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.882904838225716e-05,
+      "loss": 1.63,
+      "step": 16000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.8792456144202694e-05,
+      "loss": 1.6422,
+      "step": 16500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.875586390614823e-05,
+      "loss": 1.6405,
+      "step": 17000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.871927166809377e-05,
+      "loss": 1.6394,
+      "step": 17500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.86826794300393e-05,
+      "loss": 1.6402,
+      "step": 18000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.864608719198484e-05,
+      "loss": 1.6419,
+      "step": 18500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.8609494953930375e-05,
+      "loss": 1.6315,
+      "step": 19000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8572902715875915e-05,
+      "loss": 1.6376,
+      "step": 19500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.853631047782144e-05,
+      "loss": 1.6371,
+      "step": 20000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.849971823976698e-05,
+      "loss": 1.6185,
+      "step": 20500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8463126001712516e-05,
+      "loss": 1.6192,
+      "step": 21000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.8426533763658056e-05,
+      "loss": 1.6314,
+      "step": 21500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8389941525603596e-05,
+      "loss": 1.6257,
+      "step": 22000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.835334928754913e-05,
+      "loss": 1.6355,
+      "step": 22500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.831675704949466e-05,
+      "loss": 1.6286,
+      "step": 23000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.8280164811440196e-05,
+      "loss": 1.6391,
+      "step": 23500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.824357257338574e-05,
+      "loss": 1.6253,
+      "step": 24000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.820698033533127e-05,
+      "loss": 1.6261,
+      "step": 24500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.817038809727681e-05,
+      "loss": 1.6318,
+      "step": 25000
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.8133795859222344e-05,
+      "loss": 1.6289,
+      "step": 25500
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 4.809720362116788e-05,
+      "loss": 1.613,
+      "step": 26000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.806061138311342e-05,
+      "loss": 1.6307,
+      "step": 26500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.802401914505895e-05,
+      "loss": 1.6251,
+      "step": 27000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.798742690700449e-05,
+      "loss": 1.6164,
+      "step": 27500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.7950834668950025e-05,
+      "loss": 1.6239,
+      "step": 28000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7914242430895565e-05,
+      "loss": 1.6148,
+      "step": 28500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.787765019284109e-05,
+      "loss": 1.6191,
+      "step": 29000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.784105795478663e-05,
+      "loss": 1.6163,
+      "step": 29500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7804465716732165e-05,
+      "loss": 1.6258,
+      "step": 30000
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.7767873478677705e-05,
+      "loss": 1.6233,
+      "step": 30500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7731281240623246e-05,
+      "loss": 1.608,
+      "step": 31000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.769468900256878e-05,
+      "loss": 1.6132,
+      "step": 31500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.765809676451431e-05,
+      "loss": 1.6137,
+      "step": 32000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7621504526459846e-05,
+      "loss": 1.6208,
+      "step": 32500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.7584912288405386e-05,
+      "loss": 1.6132,
+      "step": 33000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.754832005035092e-05,
+      "loss": 1.6134,
+      "step": 33500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.751172781229646e-05,
+      "loss": 1.6111,
+      "step": 34000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.7475135574242e-05,
+      "loss": 1.6185,
+      "step": 34500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.743854333618753e-05,
+      "loss": 1.5992,
+      "step": 35000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.740195109813307e-05,
+      "loss": 1.618,
+      "step": 35500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.73653588600786e-05,
+      "loss": 1.6062,
+      "step": 36000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.732876662202414e-05,
+      "loss": 1.5988,
+      "step": 36500
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7292174383969674e-05,
+      "loss": 1.6068,
+      "step": 37000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.7255582145915214e-05,
+      "loss": 1.613,
+      "step": 37500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.721898990786075e-05,
+      "loss": 1.6041,
+      "step": 38000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.718239766980628e-05,
+      "loss": 1.6077,
+      "step": 38500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.714580543175182e-05,
+      "loss": 1.5987,
+      "step": 39000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.7109213193697355e-05,
+      "loss": 1.6093,
+      "step": 39500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.7072620955642895e-05,
+      "loss": 1.5935,
+      "step": 40000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.703602871758843e-05,
+      "loss": 1.6035,
+      "step": 40500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.699943647953396e-05,
+      "loss": 1.607,
+      "step": 41000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6962844241479495e-05,
+      "loss": 1.6108,
+      "step": 41500
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.6926252003425036e-05,
+      "loss": 1.5912,
+      "step": 42000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.688965976537057e-05,
+      "loss": 1.5959,
+      "step": 42500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.685306752731611e-05,
+      "loss": 1.6109,
+      "step": 43000
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.681647528926165e-05,
+      "loss": 1.6021,
+      "step": 43500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.6779883051207176e-05,
+      "loss": 1.6028,
+      "step": 44000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6743290813152716e-05,
+      "loss": 1.5993,
+      "step": 44500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.670669857509825e-05,
+      "loss": 1.5972,
+      "step": 45000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.667010633704379e-05,
+      "loss": 1.6042,
+      "step": 45500
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6633514098989324e-05,
+      "loss": 1.5958,
+      "step": 46000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.6596921860934864e-05,
+      "loss": 1.6014,
+      "step": 46500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.65603296228804e-05,
+      "loss": 1.5949,
+      "step": 47000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.652373738482593e-05,
+      "loss": 1.602,
+      "step": 47500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.648714514677147e-05,
+      "loss": 1.601,
+      "step": 48000
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.6450552908717004e-05,
+      "loss": 1.6017,
+      "step": 48500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6413960670662545e-05,
+      "loss": 1.5913,
+      "step": 49000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.637736843260808e-05,
+      "loss": 1.6067,
+      "step": 49500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.634077619455361e-05,
+      "loss": 1.6002,
+      "step": 50000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6304183956499145e-05,
+      "loss": 1.5893,
+      "step": 50500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.6267591718444685e-05,
+      "loss": 1.6048,
+      "step": 51000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.623099948039022e-05,
+      "loss": 1.6024,
+      "step": 51500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.619440724233576e-05,
+      "loss": 1.6034,
+      "step": 52000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.61578150042813e-05,
+      "loss": 1.5978,
+      "step": 52500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.612122276622683e-05,
+      "loss": 1.6012,
+      "step": 53000
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.6084630528172366e-05,
+      "loss": 1.5923,
+      "step": 53500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.60480382901179e-05,
+      "loss": 1.6001,
+      "step": 54000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.601144605206344e-05,
+      "loss": 1.6076,
+      "step": 54500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.597485381400897e-05,
+      "loss": 1.5937,
+      "step": 55000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.593826157595451e-05,
+      "loss": 1.5978,
+      "step": 55500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.590166933790005e-05,
+      "loss": 1.5854,
+      "step": 56000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.586507709984558e-05,
+      "loss": 1.5891,
+      "step": 56500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.582848486179112e-05,
+      "loss": 1.579,
+      "step": 57000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5791892623736654e-05,
+      "loss": 1.5941,
+      "step": 57500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.5755300385682194e-05,
+      "loss": 1.5754,
+      "step": 58000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.571870814762773e-05,
+      "loss": 1.595,
+      "step": 58500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.568211590957326e-05,
+      "loss": 1.5915,
+      "step": 59000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5645523671518794e-05,
+      "loss": 1.5749,
+      "step": 59500
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.5608931433464335e-05,
+      "loss": 1.5878,
+      "step": 60000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5572339195409875e-05,
+      "loss": 1.5955,
+      "step": 60500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.553574695735541e-05,
+      "loss": 1.5799,
+      "step": 61000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.549915471930095e-05,
+      "loss": 1.5876,
+      "step": 61500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.546256248124648e-05,
+      "loss": 1.5931,
+      "step": 62000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.5425970243192015e-05,
+      "loss": 1.5785,
+      "step": 62500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.538937800513755e-05,
+      "loss": 1.5797,
+      "step": 63000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.535278576708309e-05,
+      "loss": 1.5822,
+      "step": 63500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.531619352902862e-05,
+      "loss": 1.5798,
+      "step": 64000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.527960129097416e-05,
+      "loss": 1.5917,
+      "step": 64500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5243009052919696e-05,
+      "loss": 1.5884,
+      "step": 65000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.520641681486523e-05,
+      "loss": 1.5937,
+      "step": 65500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.516982457681077e-05,
+      "loss": 1.5757,
+      "step": 66000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.51332323387563e-05,
+      "loss": 1.5893,
+      "step": 66500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.5096640100701844e-05,
+      "loss": 1.5926,
+      "step": 67000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.506004786264738e-05,
+      "loss": 1.5903,
+      "step": 67500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.502345562459292e-05,
+      "loss": 1.5884,
+      "step": 68000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4986863386538444e-05,
+      "loss": 1.5711,
+      "step": 68500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.4950271148483984e-05,
+      "loss": 1.586,
+      "step": 69000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4913678910429524e-05,
+      "loss": 1.5826,
+      "step": 69500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.487708667237506e-05,
+      "loss": 1.5879,
+      "step": 70000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.48404944343206e-05,
+      "loss": 1.5824,
+      "step": 70500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.480390219626613e-05,
+      "loss": 1.5747,
+      "step": 71000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.4767309958211665e-05,
+      "loss": 1.5566,
+      "step": 71500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.47307177201572e-05,
+      "loss": 1.5786,
+      "step": 72000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.469412548210274e-05,
+      "loss": 1.5556,
+      "step": 72500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.465753324404827e-05,
+      "loss": 1.5717,
+      "step": 73000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.462094100599381e-05,
+      "loss": 1.5763,
+      "step": 73500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.4584348767939346e-05,
+      "loss": 1.5723,
+      "step": 74000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.454775652988488e-05,
+      "loss": 1.5621,
+      "step": 74500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.451116429183042e-05,
+      "loss": 1.5774,
+      "step": 75000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.447457205377595e-05,
+      "loss": 1.5675,
+      "step": 75500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.443797981572149e-05,
+      "loss": 1.5687,
+      "step": 76000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4401387577667027e-05,
+      "loss": 1.5625,
+      "step": 76500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.436479533961257e-05,
+      "loss": 1.5657,
+      "step": 77000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.43282031015581e-05,
+      "loss": 1.5849,
+      "step": 77500
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4291610863503634e-05,
+      "loss": 1.5824,
+      "step": 78000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 4.4255018625449174e-05,
+      "loss": 1.5763,
+      "step": 78500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.421842638739471e-05,
+      "loss": 1.5691,
+      "step": 79000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.418183414934025e-05,
+      "loss": 1.5786,
+      "step": 79500
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.414524191128578e-05,
+      "loss": 1.567,
+      "step": 80000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.4108649673231314e-05,
+      "loss": 1.586,
+      "step": 80500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.407205743517685e-05,
+      "loss": 1.5548,
+      "step": 81000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.403546519712239e-05,
+      "loss": 1.5772,
+      "step": 81500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.399887295906793e-05,
+      "loss": 1.5669,
+      "step": 82000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.396228072101346e-05,
+      "loss": 1.5724,
+      "step": 82500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.3925688482959e-05,
+      "loss": 1.5664,
+      "step": 83000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.388909624490453e-05,
+      "loss": 1.562,
+      "step": 83500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.385250400685007e-05,
+      "loss": 1.5765,
+      "step": 84000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.38159117687956e-05,
+      "loss": 1.5822,
+      "step": 84500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.377931953074114e-05,
+      "loss": 1.5759,
+      "step": 85000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3742727292686676e-05,
+      "loss": 1.561,
+      "step": 85500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.3706135054632216e-05,
+      "loss": 1.5691,
+      "step": 86000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.366954281657775e-05,
+      "loss": 1.5768,
+      "step": 86500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.363295057852328e-05,
+      "loss": 1.5721,
+      "step": 87000
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.359635834046882e-05,
+      "loss": 1.5655,
+      "step": 87500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.355976610241436e-05,
+      "loss": 1.5636,
+      "step": 88000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.35231738643599e-05,
+      "loss": 1.5688,
+      "step": 88500
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.348658162630543e-05,
+      "loss": 1.564,
+      "step": 89000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.3449989388250964e-05,
+      "loss": 1.5676,
+      "step": 89500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.34133971501965e-05,
+      "loss": 1.5664,
+      "step": 90000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.337680491214204e-05,
+      "loss": 1.5682,
+      "step": 90500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.334021267408758e-05,
+      "loss": 1.5615,
+      "step": 91000
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.330362043603311e-05,
+      "loss": 1.56,
+      "step": 91500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.326702819797865e-05,
+      "loss": 1.5638,
+      "step": 92000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.323043595992418e-05,
+      "loss": 1.5556,
+      "step": 92500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.319384372186972e-05,
+      "loss": 1.5681,
+      "step": 93000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.315725148381525e-05,
+      "loss": 1.5625,
+      "step": 93500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.312065924576079e-05,
+      "loss": 1.5622,
+      "step": 94000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.3084067007706326e-05,
+      "loss": 1.5591,
+      "step": 94500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.3047474769651866e-05,
+      "loss": 1.5588,
+      "step": 95000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.30108825315974e-05,
+      "loss": 1.5519,
+      "step": 95500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.297429029354293e-05,
+      "loss": 1.554,
+      "step": 96000
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.293769805548847e-05,
+      "loss": 1.5645,
+      "step": 96500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2901105817434006e-05,
+      "loss": 1.5534,
+      "step": 97000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2864513579379547e-05,
+      "loss": 1.5692,
+      "step": 97500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.282792134132508e-05,
+      "loss": 1.5591,
+      "step": 98000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2791329103270613e-05,
+      "loss": 1.5715,
+      "step": 98500
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.2754736865216154e-05,
+      "loss": 1.5574,
+      "step": 99000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.271814462716169e-05,
+      "loss": 1.561,
+      "step": 99500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.268155238910723e-05,
+      "loss": 1.5613,
+      "step": 100000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.264496015105276e-05,
+      "loss": 1.5654,
+      "step": 100500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.26083679129983e-05,
+      "loss": 1.5536,
+      "step": 101000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.2571775674943834e-05,
+      "loss": 1.5496,
+      "step": 101500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.253518343688937e-05,
+      "loss": 1.5497,
+      "step": 102000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.24985911988349e-05,
+      "loss": 1.5441,
+      "step": 102500
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.246199896078044e-05,
+      "loss": 1.5639,
+      "step": 103000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.242540672272598e-05,
+      "loss": 1.5557,
+      "step": 103500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.2388814484671515e-05,
+      "loss": 1.5551,
+      "step": 104000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.235222224661705e-05,
+      "loss": 1.5641,
+      "step": 104500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.231563000856258e-05,
+      "loss": 1.5574,
+      "step": 105000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.227903777050812e-05,
+      "loss": 1.5336,
+      "step": 105500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2242445532453656e-05,
+      "loss": 1.545,
+      "step": 106000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.2205853294399196e-05,
+      "loss": 1.5489,
+      "step": 106500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.216926105634473e-05,
+      "loss": 1.5452,
+      "step": 107000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.213266881829026e-05,
+      "loss": 1.5658,
+      "step": 107500
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.20960765802358e-05,
+      "loss": 1.5526,
+      "step": 108000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.2059484342181337e-05,
+      "loss": 1.5434,
+      "step": 108500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.202289210412688e-05,
+      "loss": 1.554,
+      "step": 109000
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.198629986607241e-05,
+      "loss": 1.5452,
+      "step": 109500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.194970762801795e-05,
+      "loss": 1.5456,
+      "step": 110000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.1913115389963484e-05,
+      "loss": 1.5394,
+      "step": 110500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.187652315190902e-05,
+      "loss": 1.552,
+      "step": 111000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.183993091385455e-05,
+      "loss": 1.5573,
+      "step": 111500
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.180333867580009e-05,
+      "loss": 1.5588,
+      "step": 112000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.176674643774563e-05,
+      "loss": 1.5438,
+      "step": 112500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1730154199691165e-05,
+      "loss": 1.5493,
+      "step": 113000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.16935619616367e-05,
+      "loss": 1.5468,
+      "step": 113500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.165696972358223e-05,
+      "loss": 1.5592,
+      "step": 114000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.162037748552777e-05,
+      "loss": 1.5441,
+      "step": 114500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.1583785247473305e-05,
+      "loss": 1.5499,
+      "step": 115000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.1547193009418846e-05,
+      "loss": 1.5453,
+      "step": 115500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.151060077136438e-05,
+      "loss": 1.5467,
+      "step": 116000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.147400853330992e-05,
+      "loss": 1.551,
+      "step": 116500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.143741629525545e-05,
+      "loss": 1.5446,
+      "step": 117000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1400824057200986e-05,
+      "loss": 1.5563,
+      "step": 117500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1364231819146526e-05,
+      "loss": 1.5519,
+      "step": 118000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.132763958109206e-05,
+      "loss": 1.5402,
+      "step": 118500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.12910473430376e-05,
+      "loss": 1.5404,
+      "step": 119000
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.1254455104983133e-05,
+      "loss": 1.5427,
+      "step": 119500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.121786286692867e-05,
+      "loss": 1.5476,
+      "step": 120000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.118127062887421e-05,
+      "loss": 1.5531,
+      "step": 120500
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.114467839081974e-05,
+      "loss": 1.5506,
+      "step": 121000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.110808615276528e-05,
+      "loss": 1.5361,
+      "step": 121500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.1071493914710814e-05,
+      "loss": 1.5424,
+      "step": 122000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.103490167665635e-05,
+      "loss": 1.5373,
+      "step": 122500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.099830943860188e-05,
+      "loss": 1.5339,
+      "step": 123000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.096171720054742e-05,
+      "loss": 1.5425,
+      "step": 123500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.0925124962492955e-05,
+      "loss": 1.5455,
+      "step": 124000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0888532724438495e-05,
+      "loss": 1.5454,
+      "step": 124500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.0851940486384035e-05,
+      "loss": 1.5468,
+      "step": 125000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.081534824832957e-05,
+      "loss": 1.5421,
+      "step": 125500
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.07787560102751e-05,
+      "loss": 1.5461,
+      "step": 126000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0742163772220636e-05,
+      "loss": 1.5463,
+      "step": 126500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0705571534166176e-05,
+      "loss": 1.5325,
+      "step": 127000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.066897929611171e-05,
+      "loss": 1.5417,
+      "step": 127500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.063238705805725e-05,
+      "loss": 1.5408,
+      "step": 128000
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.059579482000278e-05,
+      "loss": 1.5389,
+      "step": 128500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0559202581948316e-05,
+      "loss": 1.5448,
+      "step": 129000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.052261034389386e-05,
+      "loss": 1.5242,
+      "step": 129500
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.048601810583939e-05,
+      "loss": 1.5329,
+      "step": 130000
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044942586778493e-05,
+      "loss": 1.5452,
+      "step": 130500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0412833629730464e-05,
+      "loss": 1.5321,
+      "step": 131000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0376241391676004e-05,
+      "loss": 1.5371,
+      "step": 131500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.033964915362153e-05,
+      "loss": 1.5361,
+      "step": 132000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030305691556707e-05,
+      "loss": 1.5405,
+      "step": 132500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0266464677512604e-05,
+      "loss": 1.5341,
+      "step": 133000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0229872439458145e-05,
+      "loss": 1.5274,
+      "step": 133500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0193280201403685e-05,
+      "loss": 1.537,
+      "step": 134000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.015668796334922e-05,
+      "loss": 1.5251,
+      "step": 134500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.012009572529475e-05,
+      "loss": 1.5368,
+      "step": 135000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.0083503487240285e-05,
+      "loss": 1.5465,
+      "step": 135500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.0046911249185825e-05,
+      "loss": 1.5424,
+      "step": 136000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.001031901113136e-05,
+      "loss": 1.5382,
+      "step": 136500
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.99737267730769e-05,
+      "loss": 1.53,
+      "step": 137000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.993713453502243e-05,
+      "loss": 1.5277,
+      "step": 137500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9900542296967966e-05,
+      "loss": 1.5407,
+      "step": 138000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.9863950058913506e-05,
+      "loss": 1.5283,
+      "step": 138500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.982735782085904e-05,
+      "loss": 1.5339,
+      "step": 139000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.979076558280458e-05,
+      "loss": 1.5417,
+      "step": 139500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.975417334475011e-05,
+      "loss": 1.5307,
+      "step": 140000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.9717581106695653e-05,
+      "loss": 1.5313,
+      "step": 140500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.968098886864118e-05,
+      "loss": 1.5328,
+      "step": 141000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.964439663058672e-05,
+      "loss": 1.5253,
+      "step": 141500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.960780439253226e-05,
+      "loss": 1.5312,
+      "step": 142000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9571212154477794e-05,
+      "loss": 1.5283,
+      "step": 142500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9534619916423334e-05,
+      "loss": 1.5348,
+      "step": 143000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.949802767836887e-05,
+      "loss": 1.5302,
+      "step": 143500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.94614354403144e-05,
+      "loss": 1.5343,
+      "step": 144000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.9424843202259935e-05,
+      "loss": 1.5379,
+      "step": 144500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.9388250964205475e-05,
+      "loss": 1.5368,
+      "step": 145000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.935165872615101e-05,
+      "loss": 1.5287,
+      "step": 145500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.931506648809655e-05,
+      "loss": 1.5237,
+      "step": 146000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.927847425004209e-05,
+      "loss": 1.5289,
+      "step": 146500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9241882011987615e-05,
+      "loss": 1.535,
+      "step": 147000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.9205289773933156e-05,
+      "loss": 1.5277,
+      "step": 147500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.916869753587869e-05,
+      "loss": 1.5355,
+      "step": 148000
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.913210529782423e-05,
+      "loss": 1.5292,
+      "step": 148500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 3.909551305976976e-05,
+      "loss": 1.5256,
+      "step": 149000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.90589208217153e-05,
+      "loss": 1.5148,
+      "step": 149500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.9022328583660836e-05,
+      "loss": 1.528,
+      "step": 150000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.898573634560637e-05,
+      "loss": 1.5253,
+      "step": 150500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 3.894914410755191e-05,
+      "loss": 1.5328,
+      "step": 151000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8912551869497444e-05,
+      "loss": 1.526,
+      "step": 151500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8875959631442984e-05,
+      "loss": 1.5255,
+      "step": 152000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.883936739338852e-05,
+      "loss": 1.5191,
+      "step": 152500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.880277515533405e-05,
+      "loss": 1.5236,
+      "step": 153000
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 3.8766182917279584e-05,
+      "loss": 1.5324,
+      "step": 153500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.8729590679225124e-05,
+      "loss": 1.5193,
+      "step": 154000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.869299844117066e-05,
+      "loss": 1.533,
+      "step": 154500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.86564062031162e-05,
+      "loss": 1.5139,
+      "step": 155000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 3.861981396506174e-05,
+      "loss": 1.5097,
+      "step": 155500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8583221727007265e-05,
+      "loss": 1.5298,
+      "step": 156000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.8546629488952805e-05,
+      "loss": 1.5302,
+      "step": 156500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.851003725089834e-05,
+      "loss": 1.5199,
+      "step": 157000
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.847344501284388e-05,
+      "loss": 1.5328,
+      "step": 157500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 3.843685277478941e-05,
+      "loss": 1.5239,
+      "step": 158000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.840026053673495e-05,
+      "loss": 1.5224,
+      "step": 158500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.8363668298680486e-05,
+      "loss": 1.5227,
+      "step": 159000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.832707606062602e-05,
+      "loss": 1.519,
+      "step": 159500
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.829048382257156e-05,
+      "loss": 1.5164,
+      "step": 160000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 3.825389158451709e-05,
+      "loss": 1.5264,
+      "step": 160500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.821729934646263e-05,
+      "loss": 1.5119,
+      "step": 161000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.818070710840817e-05,
+      "loss": 1.5022,
+      "step": 161500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.81441148703537e-05,
+      "loss": 1.5245,
+      "step": 162000
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.8107522632299234e-05,
+      "loss": 1.5295,
+      "step": 162500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8070930394244774e-05,
+      "loss": 1.5189,
+      "step": 163000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8034338156190314e-05,
+      "loss": 1.5065,
+      "step": 163500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799774591813585e-05,
+      "loss": 1.5261,
+      "step": 164000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796115368008139e-05,
+      "loss": 1.5244,
+      "step": 164500
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.792456144202692e-05,
+      "loss": 1.5189,
+      "step": 165000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7887969203972455e-05,
+      "loss": 1.5254,
+      "step": 165500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.785137696591799e-05,
+      "loss": 1.5157,
+      "step": 166000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781478472786353e-05,
+      "loss": 1.5174,
+      "step": 166500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.777819248980906e-05,
+      "loss": 1.5191,
+      "step": 167000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.77416002517546e-05,
+      "loss": 1.5164,
+      "step": 167500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.7705008013700135e-05,
+      "loss": 1.5216,
+      "step": 168000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.766841577564567e-05,
+      "loss": 1.5164,
+      "step": 168500
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.763182353759121e-05,
+      "loss": 1.5141,
+      "step": 169000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.759523129953674e-05,
+      "loss": 1.5132,
+      "step": 169500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.755863906148228e-05,
+      "loss": 1.5111,
+      "step": 170000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.7522046823427816e-05,
+      "loss": 1.5128,
+      "step": 170500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.748545458537335e-05,
+      "loss": 1.5241,
+      "step": 171000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.744886234731888e-05,
+      "loss": 1.4987,
+      "step": 171500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.741227010926442e-05,
+      "loss": 1.5239,
+      "step": 172000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.7375677871209964e-05,
+      "loss": 1.5105,
+      "step": 172500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.73390856331555e-05,
+      "loss": 1.5114,
+      "step": 173000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.730249339510104e-05,
+      "loss": 1.5154,
+      "step": 173500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.726590115704657e-05,
+      "loss": 1.5179,
+      "step": 174000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.7229308918992104e-05,
+      "loss": 1.5113,
+      "step": 174500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.719271668093764e-05,
+      "loss": 1.5114,
+      "step": 175000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.715612444288318e-05,
+      "loss": 1.5118,
+      "step": 175500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 3.711953220482871e-05,
+      "loss": 1.5257,
+      "step": 176000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.708293996677425e-05,
+      "loss": 1.5052,
+      "step": 176500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.7046347728719785e-05,
+      "loss": 1.5256,
+      "step": 177000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.700975549066532e-05,
+      "loss": 1.5031,
+      "step": 177500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.697316325261086e-05,
+      "loss": 1.5137,
+      "step": 178000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 3.693657101455639e-05,
+      "loss": 1.5036,
+      "step": 178500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.689997877650193e-05,
+      "loss": 1.5165,
+      "step": 179000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6863386538447466e-05,
+      "loss": 1.5188,
+      "step": 179500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.6826794300393006e-05,
+      "loss": 1.4965,
+      "step": 180000
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.679020206233854e-05,
+      "loss": 1.4962,
+      "step": 180500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 3.675360982428407e-05,
+      "loss": 1.5142,
+      "step": 181000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.671701758622961e-05,
+      "loss": 1.5141,
+      "step": 181500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.6680425348175146e-05,
+      "loss": 1.5157,
+      "step": 182000
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.664383311012069e-05,
+      "loss": 1.5099,
+      "step": 182500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 3.660724087206622e-05,
+      "loss": 1.5176,
+      "step": 183000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.6570648634011754e-05,
+      "loss": 1.5051,
+      "step": 183500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.653405639595729e-05,
+      "loss": 1.5002,
+      "step": 184000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.649746415790283e-05,
+      "loss": 1.5082,
+      "step": 184500
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.646087191984837e-05,
+      "loss": 1.5042,
+      "step": 185000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 3.64242796817939e-05,
+      "loss": 1.5008,
+      "step": 185500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.6387687443739434e-05,
+      "loss": 1.5092,
+      "step": 186000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.635109520568497e-05,
+      "loss": 1.5057,
+      "step": 186500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.631450296763051e-05,
+      "loss": 1.5127,
+      "step": 187000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 3.627791072957604e-05,
+      "loss": 1.513,
+      "step": 187500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.624131849152158e-05,
+      "loss": 1.5124,
+      "step": 188000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6204726253467115e-05,
+      "loss": 1.513,
+      "step": 188500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.6168134015412655e-05,
+      "loss": 1.5058,
+      "step": 189000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.613154177735819e-05,
+      "loss": 1.5117,
+      "step": 189500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 3.609494953930372e-05,
+      "loss": 1.4978,
+      "step": 190000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.605835730124926e-05,
+      "loss": 1.5079,
+      "step": 190500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.6021765063194796e-05,
+      "loss": 1.5046,
+      "step": 191000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.5985172825140336e-05,
+      "loss": 1.5125,
+      "step": 191500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 3.594858058708587e-05,
+      "loss": 1.5092,
+      "step": 192000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.59119883490314e-05,
+      "loss": 1.5076,
+      "step": 192500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.5875396110976937e-05,
+      "loss": 1.51,
+      "step": 193000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.583880387292248e-05,
+      "loss": 1.5126,
+      "step": 193500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.580221163486802e-05,
+      "loss": 1.5077,
+      "step": 194000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 3.576561939681355e-05,
+      "loss": 1.5012,
+      "step": 194500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.572902715875909e-05,
+      "loss": 1.5062,
+      "step": 195000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.569243492070462e-05,
+      "loss": 1.5043,
+      "step": 195500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.565584268265016e-05,
+      "loss": 1.5013,
+      "step": 196000
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 3.561925044459569e-05,
+      "loss": 1.4952,
+      "step": 196500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.558265820654123e-05,
+      "loss": 1.4998,
+      "step": 197000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5546065968486765e-05,
+      "loss": 1.4978,
+      "step": 197500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.5509473730432305e-05,
+      "loss": 1.4972,
+      "step": 198000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.547288149237784e-05,
+      "loss": 1.4966,
+      "step": 198500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.543628925432337e-05,
+      "loss": 1.497,
+      "step": 199000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539969701626891e-05,
+      "loss": 1.5182,
+      "step": 199500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5363104778214445e-05,
+      "loss": 1.5102,
+      "step": 200000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5326512540159986e-05,
+      "loss": 1.5105,
+      "step": 200500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528992030210552e-05,
+      "loss": 1.4905,
+      "step": 201000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.525332806405105e-05,
+      "loss": 1.4966,
+      "step": 201500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.521673582599659e-05,
+      "loss": 1.5061,
+      "step": 202000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5180143587942126e-05,
+      "loss": 1.4983,
+      "step": 202500
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.5143551349887667e-05,
+      "loss": 1.5121,
+      "step": 203000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.51069591118332e-05,
+      "loss": 1.5049,
+      "step": 203500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.507036687377874e-05,
+      "loss": 1.502,
+      "step": 204000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.503377463572427e-05,
+      "loss": 1.4914,
+      "step": 204500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.499718239766981e-05,
+      "loss": 1.4958,
+      "step": 205000
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.496059015961534e-05,
+      "loss": 1.4919,
+      "step": 205500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 3.492399792156088e-05,
+      "loss": 1.4943,
+      "step": 206000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.488740568350642e-05,
+      "loss": 1.4962,
+      "step": 206500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.4850813445451954e-05,
+      "loss": 1.5059,
+      "step": 207000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.481422120739749e-05,
+      "loss": 1.5004,
+      "step": 207500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.477762896934302e-05,
+      "loss": 1.5044,
+      "step": 208000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.474103673128856e-05,
+      "loss": 1.5077,
+      "step": 208500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4704444493234095e-05,
+      "loss": 1.4931,
+      "step": 209000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4667852255179635e-05,
+      "loss": 1.5043,
+      "step": 209500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.463126001712517e-05,
+      "loss": 1.4922,
+      "step": 210000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.45946677790707e-05,
+      "loss": 1.4931,
+      "step": 210500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.455807554101624e-05,
+      "loss": 1.4972,
+      "step": 211000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4521483302961776e-05,
+      "loss": 1.4916,
+      "step": 211500
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.4484891064907316e-05,
+      "loss": 1.4966,
+      "step": 212000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.444829882685285e-05,
+      "loss": 1.4955,
+      "step": 212500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.441170658879839e-05,
+      "loss": 1.4976,
+      "step": 213000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.437511435074392e-05,
+      "loss": 1.4968,
+      "step": 213500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.4338522112689457e-05,
+      "loss": 1.5,
+      "step": 214000
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.430192987463499e-05,
+      "loss": 1.4804,
+      "step": 214500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 3.426533763658053e-05,
+      "loss": 1.4931,
+      "step": 215000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.422874539852607e-05,
+      "loss": 1.4924,
+      "step": 215500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.4192153160471604e-05,
+      "loss": 1.4876,
+      "step": 216000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.415556092241714e-05,
+      "loss": 1.4834,
+      "step": 216500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 3.411896868436267e-05,
+      "loss": 1.4859,
+      "step": 217000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.408237644630821e-05,
+      "loss": 1.4926,
+      "step": 217500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4045784208253744e-05,
+      "loss": 1.4856,
+      "step": 218000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4009191970199285e-05,
+      "loss": 1.4998,
+      "step": 218500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.397259973214482e-05,
+      "loss": 1.4946,
+      "step": 219000
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.393600749409035e-05,
+      "loss": 1.4883,
+      "step": 219500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.389941525603589e-05,
+      "loss": 1.4962,
+      "step": 220000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3862823017981425e-05,
+      "loss": 1.4869,
+      "step": 220500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3826230779926966e-05,
+      "loss": 1.5042,
+      "step": 221000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.37896385418725e-05,
+      "loss": 1.4868,
+      "step": 221500
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.375304630381804e-05,
+      "loss": 1.4902,
+      "step": 222000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.371645406576357e-05,
+      "loss": 1.494,
+      "step": 222500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3679861827709106e-05,
+      "loss": 1.4982,
+      "step": 223000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.3643269589654646e-05,
+      "loss": 1.4951,
+      "step": 223500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.360667735160018e-05,
+      "loss": 1.4826,
+      "step": 224000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.357008511354572e-05,
+      "loss": 1.4886,
+      "step": 224500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3533492875491253e-05,
+      "loss": 1.4923,
+      "step": 225000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.349690063743679e-05,
+      "loss": 1.5034,
+      "step": 225500
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.346030839938232e-05,
+      "loss": 1.4953,
+      "step": 226000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.342371616132786e-05,
+      "loss": 1.4864,
+      "step": 226500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3387123923273394e-05,
+      "loss": 1.4831,
+      "step": 227000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3350531685218934e-05,
+      "loss": 1.4883,
+      "step": 227500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3313939447164474e-05,
+      "loss": 1.4822,
+      "step": 228000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.327734720911001e-05,
+      "loss": 1.475,
+      "step": 228500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.324075497105554e-05,
+      "loss": 1.4794,
+      "step": 229000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3204162733001075e-05,
+      "loss": 1.4772,
+      "step": 229500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.3167570494946615e-05,
+      "loss": 1.4831,
+      "step": 230000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.313097825689215e-05,
+      "loss": 1.4867,
+      "step": 230500
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 3.309438601883769e-05,
+      "loss": 1.4798,
+      "step": 231000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.305779378078322e-05,
+      "loss": 1.4777,
+      "step": 231500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.3021201542728756e-05,
+      "loss": 1.4872,
+      "step": 232000
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.2984609304674296e-05,
+      "loss": 1.4875,
+      "step": 232500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 3.294801706661983e-05,
+      "loss": 1.4736,
+      "step": 233000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.291142482856537e-05,
+      "loss": 1.4773,
+      "step": 233500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.28748325905109e-05,
+      "loss": 1.488,
+      "step": 234000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.2838240352456436e-05,
+      "loss": 1.4796,
+      "step": 234500
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.280164811440197e-05,
+      "loss": 1.483,
+      "step": 235000
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 3.276505587634751e-05,
+      "loss": 1.4805,
+      "step": 235500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2728463638293043e-05,
+      "loss": 1.4837,
+      "step": 236000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2691871400238584e-05,
+      "loss": 1.4773,
+      "step": 236500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.2655279162184124e-05,
+      "loss": 1.491,
+      "step": 237000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 3.261868692412966e-05,
+      "loss": 1.4761,
+      "step": 237500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.258209468607519e-05,
+      "loss": 1.4824,
+      "step": 238000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2545502448020724e-05,
+      "loss": 1.4767,
+      "step": 238500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.2508910209966265e-05,
+      "loss": 1.4718,
+      "step": 239000
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.24723179719118e-05,
+      "loss": 1.4887,
+      "step": 239500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 3.243572573385734e-05,
+      "loss": 1.476,
+      "step": 240000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.239913349580287e-05,
+      "loss": 1.4795,
+      "step": 240500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2362541257748405e-05,
+      "loss": 1.4674,
+      "step": 241000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.2325949019693945e-05,
+      "loss": 1.481,
+      "step": 241500
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.228935678163948e-05,
+      "loss": 1.49,
+      "step": 242000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 3.225276454358502e-05,
+      "loss": 1.4811,
+      "step": 242500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.221617230553055e-05,
+      "loss": 1.4741,
+      "step": 243000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.217958006747609e-05,
+      "loss": 1.4684,
+      "step": 243500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.214298782942162e-05,
+      "loss": 1.4863,
+      "step": 244000
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 3.210639559136716e-05,
+      "loss": 1.4699,
+      "step": 244500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.20698033533127e-05,
+      "loss": 1.4665,
+      "step": 245000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.203321111525823e-05,
+      "loss": 1.4577,
+      "step": 245500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.1996618877203773e-05,
+      "loss": 1.4852,
+      "step": 246000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.196002663914931e-05,
+      "loss": 1.4776,
+      "step": 246500
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 3.192343440109484e-05,
+      "loss": 1.4705,
+      "step": 247000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1886842163040374e-05,
+      "loss": 1.4617,
+      "step": 247500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.1850249924985914e-05,
+      "loss": 1.4691,
+      "step": 248000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.181365768693145e-05,
+      "loss": 1.4705,
+      "step": 248500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 3.177706544887699e-05,
+      "loss": 1.4794,
+      "step": 249000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.174047321082252e-05,
+      "loss": 1.476,
+      "step": 249500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1703880972768055e-05,
+      "loss": 1.4817,
+      "step": 250000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.1667288734713595e-05,
+      "loss": 1.4653,
+      "step": 250500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.163069649665913e-05,
+      "loss": 1.4639,
+      "step": 251000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 3.159410425860467e-05,
+      "loss": 1.4806,
+      "step": 251500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.15575120205502e-05,
+      "loss": 1.4766,
+      "step": 252000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.152091978249574e-05,
+      "loss": 1.4691,
+      "step": 252500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.148432754444127e-05,
+      "loss": 1.4793,
+      "step": 253000
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 3.144773530638681e-05,
+      "loss": 1.4678,
+      "step": 253500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.141114306833235e-05,
+      "loss": 1.475,
+      "step": 254000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.137455083027788e-05,
+      "loss": 1.4716,
+      "step": 254500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.133795859222342e-05,
+      "loss": 1.4663,
+      "step": 255000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.1301366354168956e-05,
+      "loss": 1.4693,
+      "step": 255500
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 3.126477411611449e-05,
+      "loss": 1.4735,
+      "step": 256000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.122818187806002e-05,
+      "loss": 1.4784,
+      "step": 256500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.1191589640005563e-05,
+      "loss": 1.4716,
+      "step": 257000
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.11549974019511e-05,
+      "loss": 1.4619,
+      "step": 257500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 3.111840516389664e-05,
+      "loss": 1.4704,
+      "step": 258000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.108181292584218e-05,
+      "loss": 1.4796,
+      "step": 258500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1045220687787704e-05,
+      "loss": 1.4783,
+      "step": 259000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.1008628449733244e-05,
+      "loss": 1.4711,
+      "step": 259500
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.097203621167878e-05,
+      "loss": 1.4638,
+      "step": 260000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 3.093544397362432e-05,
+      "loss": 1.4655,
+      "step": 260500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.089885173556985e-05,
+      "loss": 1.4595,
+      "step": 261000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.086225949751539e-05,
+      "loss": 1.4673,
+      "step": 261500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0825667259460925e-05,
+      "loss": 1.4682,
+      "step": 262000
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.078907502140646e-05,
+      "loss": 1.4749,
+      "step": 262500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 3.0752482783352e-05,
+      "loss": 1.4524,
+      "step": 263000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.071589054529753e-05,
+      "loss": 1.4649,
+      "step": 263500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.067929830724307e-05,
+      "loss": 1.458,
+      "step": 264000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.0642706069188606e-05,
+      "loss": 1.4653,
+      "step": 264500
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 3.060611383113414e-05,
+      "loss": 1.4766,
+      "step": 265000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.056952159307967e-05,
+      "loss": 1.4627,
+      "step": 265500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.053292935502521e-05,
+      "loss": 1.4746,
+      "step": 266000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0496337116970753e-05,
+      "loss": 1.4701,
+      "step": 266500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0459744878916287e-05,
+      "loss": 1.4714,
+      "step": 267000
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 3.0423152640861824e-05,
+      "loss": 1.4657,
+      "step": 267500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0386560402807357e-05,
+      "loss": 1.4634,
+      "step": 268000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0349968164752897e-05,
+      "loss": 1.4654,
+      "step": 268500
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0313375926698427e-05,
+      "loss": 1.4651,
+      "step": 269000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 3.0276783688643967e-05,
+      "loss": 1.4647,
+      "step": 269500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.02401914505895e-05,
+      "loss": 1.4685,
+      "step": 270000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0203599212535038e-05,
+      "loss": 1.4768,
+      "step": 270500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0167006974480578e-05,
+      "loss": 1.47,
+      "step": 271000
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.013041473642611e-05,
+      "loss": 1.4728,
+      "step": 271500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 3.0093822498371648e-05,
+      "loss": 1.4801,
+      "step": 272000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0057230260317182e-05,
+      "loss": 1.4647,
+      "step": 272500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 3.0020638022262722e-05,
+      "loss": 1.4687,
+      "step": 273000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9984045784208252e-05,
+      "loss": 1.4618,
+      "step": 273500
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 2.9947453546153792e-05,
+      "loss": 1.4551,
+      "step": 274000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9910861308099326e-05,
+      "loss": 1.4703,
+      "step": 274500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9874269070044862e-05,
+      "loss": 1.4583,
+      "step": 275000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9837676831990403e-05,
+      "loss": 1.4737,
+      "step": 275500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9801084593935936e-05,
+      "loss": 1.477,
+      "step": 276000
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 2.9764492355881473e-05,
+      "loss": 1.4587,
+      "step": 276500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9727900117827006e-05,
+      "loss": 1.4587,
+      "step": 277000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9691307879772547e-05,
+      "loss": 1.468,
+      "step": 277500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9654715641718077e-05,
+      "loss": 1.4589,
+      "step": 278000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 2.9618123403663617e-05,
+      "loss": 1.4661,
+      "step": 278500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.958153116560915e-05,
+      "loss": 1.4504,
+      "step": 279000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9544938927554687e-05,
+      "loss": 1.4472,
+      "step": 279500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9508346689500227e-05,
+      "loss": 1.467,
+      "step": 280000
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.947175445144576e-05,
+      "loss": 1.4575,
+      "step": 280500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 2.9435162213391298e-05,
+      "loss": 1.4558,
+      "step": 281000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.939856997533683e-05,
+      "loss": 1.4492,
+      "step": 281500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.936197773728237e-05,
+      "loss": 1.4552,
+      "step": 282000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.93253854992279e-05,
+      "loss": 1.4646,
+      "step": 282500
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9288793261173442e-05,
+      "loss": 1.4627,
+      "step": 283000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 2.9252201023118975e-05,
+      "loss": 1.4544,
+      "step": 283500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9215608785064512e-05,
+      "loss": 1.4598,
+      "step": 284000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9179016547010052e-05,
+      "loss": 1.4671,
+      "step": 284500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9142424308955586e-05,
+      "loss": 1.4576,
+      "step": 285000
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 2.9105832070901123e-05,
+      "loss": 1.4607,
+      "step": 285500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9069239832846656e-05,
+      "loss": 1.4689,
+      "step": 286000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.9032647594792196e-05,
+      "loss": 1.451,
+      "step": 286500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.899605535673773e-05,
+      "loss": 1.4459,
+      "step": 287000
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8959463118683266e-05,
+      "loss": 1.4543,
+      "step": 287500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 2.8922870880628807e-05,
+      "loss": 1.4457,
+      "step": 288000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8886278642574337e-05,
+      "loss": 1.4578,
+      "step": 288500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8849686404519877e-05,
+      "loss": 1.4534,
+      "step": 289000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.881309416646541e-05,
+      "loss": 1.462,
+      "step": 289500
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 2.8776501928410947e-05,
+      "loss": 1.464,
+      "step": 290000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.873990969035648e-05,
+      "loss": 1.4558,
+      "step": 290500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.870331745230202e-05,
+      "loss": 1.4746,
+      "step": 291000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.8666725214247554e-05,
+      "loss": 1.4608,
+      "step": 291500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.863013297619309e-05,
+      "loss": 1.4552,
+      "step": 292000
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 2.859354073813863e-05,
+      "loss": 1.4459,
+      "step": 292500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.855694850008416e-05,
+      "loss": 1.4538,
+      "step": 293000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8520356262029702e-05,
+      "loss": 1.4621,
+      "step": 293500
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8483764023975235e-05,
+      "loss": 1.4639,
+      "step": 294000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 2.8447171785920772e-05,
+      "loss": 1.4555,
+      "step": 294500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8410579547866305e-05,
+      "loss": 1.4574,
+      "step": 295000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8373987309811846e-05,
+      "loss": 1.4561,
+      "step": 295500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.833739507175738e-05,
+      "loss": 1.4475,
+      "step": 296000
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8300802833702916e-05,
+      "loss": 1.4421,
+      "step": 296500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 2.8264210595648456e-05,
+      "loss": 1.4492,
+      "step": 297000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8227618357593986e-05,
+      "loss": 1.468,
+      "step": 297500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8191026119539526e-05,
+      "loss": 1.4536,
+      "step": 298000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.815443388148506e-05,
+      "loss": 1.4538,
+      "step": 298500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 2.8117841643430597e-05,
+      "loss": 1.4647,
+      "step": 299000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.808124940537613e-05,
+      "loss": 1.4547,
+      "step": 299500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.804465716732167e-05,
+      "loss": 1.4507,
+      "step": 300000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.8008064929267204e-05,
+      "loss": 1.4644,
+      "step": 300500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.797147269121274e-05,
+      "loss": 1.4516,
+      "step": 301000
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 2.793488045315828e-05,
+      "loss": 1.4417,
+      "step": 301500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7898288215103814e-05,
+      "loss": 1.4488,
+      "step": 302000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.786169597704935e-05,
+      "loss": 1.4614,
+      "step": 302500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7825103738994885e-05,
+      "loss": 1.4508,
+      "step": 303000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.778851150094042e-05,
+      "loss": 1.4504,
+      "step": 303500
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 2.7751919262885955e-05,
+      "loss": 1.4437,
+      "step": 304000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7715327024831495e-05,
+      "loss": 1.4503,
+      "step": 304500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.767873478677703e-05,
+      "loss": 1.456,
+      "step": 305000
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7642142548722565e-05,
+      "loss": 1.4557,
+      "step": 305500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 2.7605550310668106e-05,
+      "loss": 1.4514,
+      "step": 306000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.756895807261364e-05,
+      "loss": 1.4441,
+      "step": 306500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7532365834559176e-05,
+      "loss": 1.4497,
+      "step": 307000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.749577359650471e-05,
+      "loss": 1.4647,
+      "step": 307500
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.7459181358450246e-05,
+      "loss": 1.4527,
+      "step": 308000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 2.742258912039578e-05,
+      "loss": 1.4432,
+      "step": 308500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.738599688234132e-05,
+      "loss": 1.4486,
+      "step": 309000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.7349404644286857e-05,
+      "loss": 1.4461,
+      "step": 309500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.731281240623239e-05,
+      "loss": 1.4458,
+      "step": 310000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 2.727622016817793e-05,
+      "loss": 1.4462,
+      "step": 310500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7239627930123464e-05,
+      "loss": 1.4523,
+      "step": 311000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7203035692069e-05,
+      "loss": 1.453,
+      "step": 311500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7166443454014534e-05,
+      "loss": 1.4467,
+      "step": 312000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.712985121596007e-05,
+      "loss": 1.4449,
+      "step": 312500
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 2.7093258977905604e-05,
+      "loss": 1.4375,
+      "step": 313000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.7056666739851145e-05,
+      "loss": 1.4475,
+      "step": 313500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.702007450179668e-05,
+      "loss": 1.4335,
+      "step": 314000
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6983482263742215e-05,
+      "loss": 1.4456,
+      "step": 314500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 2.6946890025687755e-05,
+      "loss": 1.4434,
+      "step": 315000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.691029778763329e-05,
+      "loss": 1.4475,
+      "step": 315500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.6873705549578825e-05,
+      "loss": 1.4456,
+      "step": 316000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.683711331152436e-05,
+      "loss": 1.4445,
+      "step": 316500
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.68005210734699e-05,
+      "loss": 1.4565,
+      "step": 317000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 2.676392883541543e-05,
+      "loss": 1.4396,
+      "step": 317500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.672733659736097e-05,
+      "loss": 1.4454,
+      "step": 318000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.6690744359306506e-05,
+      "loss": 1.4381,
+      "step": 318500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.665415212125204e-05,
+      "loss": 1.4529,
+      "step": 319000
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 2.661755988319758e-05,
+      "loss": 1.448,
+      "step": 319500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6580967645143113e-05,
+      "loss": 1.4328,
+      "step": 320000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.654437540708865e-05,
+      "loss": 1.448,
+      "step": 320500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6507783169034184e-05,
+      "loss": 1.4541,
+      "step": 321000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6471190930979724e-05,
+      "loss": 1.4476,
+      "step": 321500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 2.6434598692925254e-05,
+      "loss": 1.4489,
+      "step": 322000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6398006454870794e-05,
+      "loss": 1.4618,
+      "step": 322500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.636141421681633e-05,
+      "loss": 1.4398,
+      "step": 323000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6324821978761864e-05,
+      "loss": 1.4411,
+      "step": 323500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6288229740707405e-05,
+      "loss": 1.4441,
+      "step": 324000
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 2.6251637502652938e-05,
+      "loss": 1.4395,
+      "step": 324500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.6215045264598475e-05,
+      "loss": 1.4371,
+      "step": 325000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.617845302654401e-05,
+      "loss": 1.4468,
+      "step": 325500
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.614186078848955e-05,
+      "loss": 1.4403,
+      "step": 326000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 2.610526855043508e-05,
+      "loss": 1.4377,
+      "step": 326500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.606867631238062e-05,
+      "loss": 1.4597,
+      "step": 327000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.6032084074326156e-05,
+      "loss": 1.4503,
+      "step": 327500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.599549183627169e-05,
+      "loss": 1.4433,
+      "step": 328000
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.595889959821723e-05,
+      "loss": 1.4418,
+      "step": 328500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 2.5922307360162763e-05,
+      "loss": 1.445,
+      "step": 329000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.58857151221083e-05,
+      "loss": 1.4398,
+      "step": 329500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5849122884053833e-05,
+      "loss": 1.4406,
+      "step": 330000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.5812530645999373e-05,
+      "loss": 1.4306,
+      "step": 330500
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 2.577593840794491e-05,
+      "loss": 1.446,
+      "step": 331000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5739346169890444e-05,
+      "loss": 1.4414,
+      "step": 331500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5702753931835984e-05,
+      "loss": 1.4427,
+      "step": 332000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5666161693781514e-05,
+      "loss": 1.4269,
+      "step": 332500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5629569455727054e-05,
+      "loss": 1.4273,
+      "step": 333000
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 2.5592977217672588e-05,
+      "loss": 1.4398,
+      "step": 333500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5556384979618124e-05,
+      "loss": 1.4393,
+      "step": 334000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5519792741563658e-05,
+      "loss": 1.4275,
+      "step": 334500
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5483200503509198e-05,
+      "loss": 1.4423,
+      "step": 335000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 2.5446608265454735e-05,
+      "loss": 1.4364,
+      "step": 335500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.541001602740027e-05,
+      "loss": 1.4414,
+      "step": 336000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.537342378934581e-05,
+      "loss": 1.4501,
+      "step": 336500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.533683155129134e-05,
+      "loss": 1.4437,
+      "step": 337000
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.530023931323688e-05,
+      "loss": 1.4224,
+      "step": 337500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 2.5263647075182412e-05,
+      "loss": 1.4348,
+      "step": 338000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.522705483712795e-05,
+      "loss": 1.4394,
+      "step": 338500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5190462599073483e-05,
+      "loss": 1.4336,
+      "step": 339000
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.5153870361019023e-05,
+      "loss": 1.4446,
+      "step": 339500
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 2.511727812296456e-05,
+      "loss": 1.4356,
+      "step": 340000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5080685884910093e-05,
+      "loss": 1.4399,
+      "step": 340500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5044093646855633e-05,
+      "loss": 1.4292,
+      "step": 341000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.5007501408801163e-05,
+      "loss": 1.4266,
+      "step": 341500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.4970909170746704e-05,
+      "loss": 1.4426,
+      "step": 342000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 2.493431693269224e-05,
+      "loss": 1.4315,
+      "step": 342500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4897724694637774e-05,
+      "loss": 1.4309,
+      "step": 343000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.486113245658331e-05,
+      "loss": 1.4369,
+      "step": 343500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4824540218528848e-05,
+      "loss": 1.4459,
+      "step": 344000
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.478794798047438e-05,
+      "loss": 1.4316,
+      "step": 344500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 2.4751355742419918e-05,
+      "loss": 1.432,
+      "step": 345000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4714763504365455e-05,
+      "loss": 1.4253,
+      "step": 345500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.467817126631099e-05,
+      "loss": 1.4421,
+      "step": 346000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.464157902825653e-05,
+      "loss": 1.4383,
+      "step": 346500
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 2.4604986790202065e-05,
+      "loss": 1.4369,
+      "step": 347000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.45683945521476e-05,
+      "loss": 1.443,
+      "step": 347500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4531802314093136e-05,
+      "loss": 1.4279,
+      "step": 348000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4495210076038672e-05,
+      "loss": 1.4255,
+      "step": 348500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4458617837984206e-05,
+      "loss": 1.433,
+      "step": 349000
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 2.4422025599929743e-05,
+      "loss": 1.4421,
+      "step": 349500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.438543336187528e-05,
+      "loss": 1.4499,
+      "step": 350000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4348841123820816e-05,
+      "loss": 1.4263,
+      "step": 350500
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.4312248885766353e-05,
+      "loss": 1.4236,
+      "step": 351000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 2.427565664771189e-05,
+      "loss": 1.441,
+      "step": 351500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4239064409657423e-05,
+      "loss": 1.4418,
+      "step": 352000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.420247217160296e-05,
+      "loss": 1.4189,
+      "step": 352500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4165879933548497e-05,
+      "loss": 1.4225,
+      "step": 353000
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.412928769549403e-05,
+      "loss": 1.427,
+      "step": 353500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 2.4092695457439567e-05,
+      "loss": 1.4288,
+      "step": 354000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.4056103219385104e-05,
+      "loss": 1.4174,
+      "step": 354500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.401951098133064e-05,
+      "loss": 1.4329,
+      "step": 355000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3982918743276178e-05,
+      "loss": 1.4336,
+      "step": 355500
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 2.3946326505221715e-05,
+      "loss": 1.4388,
+      "step": 356000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3909734267167248e-05,
+      "loss": 1.4281,
+      "step": 356500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3873142029112785e-05,
+      "loss": 1.4196,
+      "step": 357000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3836549791058322e-05,
+      "loss": 1.4283,
+      "step": 357500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.379995755300386e-05,
+      "loss": 1.4332,
+      "step": 358000
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 2.3763365314949392e-05,
+      "loss": 1.4292,
+      "step": 358500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3726773076894932e-05,
+      "loss": 1.4226,
+      "step": 359000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3690180838840466e-05,
+      "loss": 1.4364,
+      "step": 359500
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.3653588600786003e-05,
+      "loss": 1.4235,
+      "step": 360000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 2.361699636273154e-05,
+      "loss": 1.4254,
+      "step": 360500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3580404124677073e-05,
+      "loss": 1.4348,
+      "step": 361000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.354381188662261e-05,
+      "loss": 1.4257,
+      "step": 361500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3507219648568147e-05,
+      "loss": 1.4319,
+      "step": 362000
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3470627410513683e-05,
+      "loss": 1.4242,
+      "step": 362500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 2.3434035172459217e-05,
+      "loss": 1.4264,
+      "step": 363000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3397442934404757e-05,
+      "loss": 1.427,
+      "step": 363500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.336085069635029e-05,
+      "loss": 1.4261,
+      "step": 364000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3324258458295827e-05,
+      "loss": 1.4274,
+      "step": 364500
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.3287666220241364e-05,
+      "loss": 1.4225,
+      "step": 365000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 2.32510739821869e-05,
+      "loss": 1.4236,
+      "step": 365500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3214481744132435e-05,
+      "loss": 1.4243,
+      "step": 366000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.317788950607797e-05,
+      "loss": 1.4178,
+      "step": 366500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3141297268023508e-05,
+      "loss": 1.4244,
+      "step": 367000
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 2.3104705029969045e-05,
+      "loss": 1.4273,
+      "step": 367500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3068112791914582e-05,
+      "loss": 1.4276,
+      "step": 368000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.3031520553860115e-05,
+      "loss": 1.425,
+      "step": 368500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2994928315805652e-05,
+      "loss": 1.4162,
+      "step": 369000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.295833607775119e-05,
+      "loss": 1.4085,
+      "step": 369500
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 2.2921743839696726e-05,
+      "loss": 1.4235,
+      "step": 370000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.288515160164226e-05,
+      "loss": 1.426,
+      "step": 370500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2848559363587796e-05,
+      "loss": 1.4236,
+      "step": 371000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.2811967125533333e-05,
+      "loss": 1.4346,
+      "step": 371500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 2.277537488747887e-05,
+      "loss": 1.4163,
+      "step": 372000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2738782649424407e-05,
+      "loss": 1.427,
+      "step": 372500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2702190411369943e-05,
+      "loss": 1.4308,
+      "step": 373000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2665598173315477e-05,
+      "loss": 1.4172,
+      "step": 373500
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.2629005935261014e-05,
+      "loss": 1.4327,
+      "step": 374000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 2.259241369720655e-05,
+      "loss": 1.437,
+      "step": 374500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2555821459152084e-05,
+      "loss": 1.421,
+      "step": 375000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.251922922109762e-05,
+      "loss": 1.4336,
+      "step": 375500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2482636983043158e-05,
+      "loss": 1.4263,
+      "step": 376000
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 2.2446044744988695e-05,
+      "loss": 1.4289,
+      "step": 376500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.240945250693423e-05,
+      "loss": 1.4185,
+      "step": 377000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2372860268879768e-05,
+      "loss": 1.4224,
+      "step": 377500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.23362680308253e-05,
+      "loss": 1.4195,
+      "step": 378000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.229967579277084e-05,
+      "loss": 1.4169,
+      "step": 378500
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 2.2263083554716375e-05,
+      "loss": 1.4122,
+      "step": 379000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.222649131666191e-05,
+      "loss": 1.4241,
+      "step": 379500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2189899078607446e-05,
+      "loss": 1.4257,
+      "step": 380000
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.2153306840552986e-05,
+      "loss": 1.4324,
+      "step": 380500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 2.211671460249852e-05,
+      "loss": 1.4189,
+      "step": 381000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2080122364444056e-05,
+      "loss": 1.4134,
+      "step": 381500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2043530126389593e-05,
+      "loss": 1.4238,
+      "step": 382000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.2006937888335126e-05,
+      "loss": 1.4123,
+      "step": 382500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.1970345650280663e-05,
+      "loss": 1.4131,
+      "step": 383000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 2.19337534122262e-05,
+      "loss": 1.4179,
+      "step": 383500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1897161174171734e-05,
+      "loss": 1.4139,
+      "step": 384000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.186056893611727e-05,
+      "loss": 1.4309,
+      "step": 384500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.182397669806281e-05,
+      "loss": 1.4202,
+      "step": 385000
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.1787384460008344e-05,
+      "loss": 1.4096,
+      "step": 385500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 2.175079222195388e-05,
+      "loss": 1.4156,
+      "step": 386000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1714199983899418e-05,
+      "loss": 1.4273,
+      "step": 386500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.167760774584495e-05,
+      "loss": 1.42,
+      "step": 387000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1641015507790488e-05,
+      "loss": 1.419,
+      "step": 387500
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 2.1604423269736025e-05,
+      "loss": 1.4246,
+      "step": 388000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1567831031681558e-05,
+      "loss": 1.4296,
+      "step": 388500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.15312387936271e-05,
+      "loss": 1.4141,
+      "step": 389000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1494646555572635e-05,
+      "loss": 1.426,
+      "step": 389500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.145805431751817e-05,
+      "loss": 1.4242,
+      "step": 390000
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 2.1421462079463706e-05,
+      "loss": 1.4317,
+      "step": 390500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1384869841409242e-05,
+      "loss": 1.4178,
+      "step": 391000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1348277603354776e-05,
+      "loss": 1.4106,
+      "step": 391500
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.1311685365300313e-05,
+      "loss": 1.4157,
+      "step": 392000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 2.127509312724585e-05,
+      "loss": 1.414,
+      "step": 392500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1238500889191383e-05,
+      "loss": 1.4123,
+      "step": 393000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1201908651136923e-05,
+      "loss": 1.4141,
+      "step": 393500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.116531641308246e-05,
+      "loss": 1.4296,
+      "step": 394000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.1128724175027994e-05,
+      "loss": 1.4074,
+      "step": 394500
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 2.109213193697353e-05,
+      "loss": 1.4068,
+      "step": 395000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.1055539698919067e-05,
+      "loss": 1.4139,
+      "step": 395500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.10189474608646e-05,
+      "loss": 1.4258,
+      "step": 396000
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0982355222810138e-05,
+      "loss": 1.4275,
+      "step": 396500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 2.0945762984755674e-05,
+      "loss": 1.4266,
+      "step": 397000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.090917074670121e-05,
+      "loss": 1.4043,
+      "step": 397500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0872578508646748e-05,
+      "loss": 1.4095,
+      "step": 398000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0835986270592285e-05,
+      "loss": 1.4199,
+      "step": 398500
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0799394032537818e-05,
+      "loss": 1.4112,
+      "step": 399000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 2.0762801794483355e-05,
+      "loss": 1.411,
+      "step": 399500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0726209556428892e-05,
+      "loss": 1.4181,
+      "step": 400000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0689617318374425e-05,
+      "loss": 1.4036,
+      "step": 400500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.0653025080319962e-05,
+      "loss": 1.4165,
+      "step": 401000
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 2.06164328422655e-05,
+      "loss": 1.4157,
+      "step": 401500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0579840604211036e-05,
+      "loss": 1.4258,
+      "step": 402000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0543248366156573e-05,
+      "loss": 1.4104,
+      "step": 402500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.050665612810211e-05,
+      "loss": 1.4068,
+      "step": 403000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.0470063890047643e-05,
+      "loss": 1.4088,
+      "step": 403500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 2.043347165199318e-05,
+      "loss": 1.4253,
+      "step": 404000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0396879413938717e-05,
+      "loss": 1.4201,
+      "step": 404500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.036028717588425e-05,
+      "loss": 1.4153,
+      "step": 405000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0323694937829787e-05,
+      "loss": 1.4145,
+      "step": 405500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.0287102699775324e-05,
+      "loss": 1.4058,
+      "step": 406000
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 2.025051046172086e-05,
+      "loss": 1.4129,
+      "step": 406500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0213918223666398e-05,
+      "loss": 1.3999,
+      "step": 407000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0177325985611934e-05,
+      "loss": 1.4136,
+      "step": 407500
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0140733747557468e-05,
+      "loss": 1.418,
+      "step": 408000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 2.0104141509503005e-05,
+      "loss": 1.3983,
+      "step": 408500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.006754927144854e-05,
+      "loss": 1.4243,
+      "step": 409000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 2.0030957033394075e-05,
+      "loss": 1.4031,
+      "step": 409500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9994364795339612e-05,
+      "loss": 1.4128,
+      "step": 410000
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9957772557285152e-05,
+      "loss": 1.4104,
+      "step": 410500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.9921180319230685e-05,
+      "loss": 1.3994,
+      "step": 411000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9884588081176222e-05,
+      "loss": 1.4002,
+      "step": 411500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.984799584312176e-05,
+      "loss": 1.4071,
+      "step": 412000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.9811403605067293e-05,
+      "loss": 1.3979,
+      "step": 412500
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.977481136701283e-05,
+      "loss": 1.4067,
+      "step": 413000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9738219128958366e-05,
+      "loss": 1.4121,
+      "step": 413500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9701626890903903e-05,
+      "loss": 1.4007,
+      "step": 414000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9665034652849437e-05,
+      "loss": 1.4051,
+      "step": 414500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.9628442414794977e-05,
+      "loss": 1.3998,
+      "step": 415000
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.959185017674051e-05,
+      "loss": 1.402,
+      "step": 415500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9555257938686047e-05,
+      "loss": 1.4112,
+      "step": 416000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9518665700631584e-05,
+      "loss": 1.4155,
+      "step": 416500
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9482073462577117e-05,
+      "loss": 1.4048,
+      "step": 417000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.9445481224522654e-05,
+      "loss": 1.4107,
+      "step": 417500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.940888898646819e-05,
+      "loss": 1.4015,
+      "step": 418000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9372296748413728e-05,
+      "loss": 1.4005,
+      "step": 418500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9335704510359265e-05,
+      "loss": 1.4099,
+      "step": 419000
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.92991122723048e-05,
+      "loss": 1.3981,
+      "step": 419500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.9262520034250335e-05,
+      "loss": 1.4044,
+      "step": 420000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9225927796195872e-05,
+      "loss": 1.4103,
+      "step": 420500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.918933555814141e-05,
+      "loss": 1.4028,
+      "step": 421000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.9152743320086945e-05,
+      "loss": 1.3967,
+      "step": 421500
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.911615108203248e-05,
+      "loss": 1.4172,
+      "step": 422000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9079558843978016e-05,
+      "loss": 1.4028,
+      "step": 422500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.9042966605923553e-05,
+      "loss": 1.4035,
+      "step": 423000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.900637436786909e-05,
+      "loss": 1.4115,
+      "step": 423500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8969782129814626e-05,
+      "loss": 1.4128,
+      "step": 424000
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.893318989176016e-05,
+      "loss": 1.3945,
+      "step": 424500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8896597653705697e-05,
+      "loss": 1.3949,
+      "step": 425000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8860005415651233e-05,
+      "loss": 1.4021,
+      "step": 425500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.882341317759677e-05,
+      "loss": 1.4071,
+      "step": 426000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8786820939542304e-05,
+      "loss": 1.4107,
+      "step": 426500
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.875022870148784e-05,
+      "loss": 1.3958,
+      "step": 427000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8713636463433377e-05,
+      "loss": 1.4203,
+      "step": 427500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8677044225378914e-05,
+      "loss": 1.41,
+      "step": 428000
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.864045198732445e-05,
+      "loss": 1.4073,
+      "step": 428500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8603859749269988e-05,
+      "loss": 1.4093,
+      "step": 429000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.856726751121552e-05,
+      "loss": 1.4112,
+      "step": 429500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8530675273161058e-05,
+      "loss": 1.3969,
+      "step": 430000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8494083035106595e-05,
+      "loss": 1.4068,
+      "step": 430500
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.845749079705213e-05,
+      "loss": 1.4025,
+      "step": 431000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.8420898558997665e-05,
+      "loss": 1.4085,
+      "step": 431500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8384306320943202e-05,
+      "loss": 1.4004,
+      "step": 432000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.834771408288874e-05,
+      "loss": 1.3999,
+      "step": 432500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8311121844834276e-05,
+      "loss": 1.4,
+      "step": 433000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8274529606779813e-05,
+      "loss": 1.3927,
+      "step": 433500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8237937368725346e-05,
+      "loss": 1.4053,
+      "step": 434000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8201345130670883e-05,
+      "loss": 1.3966,
+      "step": 434500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.816475289261642e-05,
+      "loss": 1.3804,
+      "step": 435000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8128160654561953e-05,
+      "loss": 1.4008,
+      "step": 435500
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.809156841650749e-05,
+      "loss": 1.4011,
+      "step": 436000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.805497617845303e-05,
+      "loss": 1.4068,
+      "step": 436500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8018383940398564e-05,
+      "loss": 1.4059,
+      "step": 437000
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.79817917023441e-05,
+      "loss": 1.4002,
+      "step": 437500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7945199464289637e-05,
+      "loss": 1.4019,
+      "step": 438000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.790860722623517e-05,
+      "loss": 1.4028,
+      "step": 438500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7872014988180708e-05,
+      "loss": 1.3993,
+      "step": 439000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7835422750126244e-05,
+      "loss": 1.3996,
+      "step": 439500
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7798830512071778e-05,
+      "loss": 1.4043,
+      "step": 440000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.7762238274017318e-05,
+      "loss": 1.4078,
+      "step": 440500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7725646035962855e-05,
+      "loss": 1.3996,
+      "step": 441000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.768905379790839e-05,
+      "loss": 1.3994,
+      "step": 441500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7652461559853925e-05,
+      "loss": 1.3899,
+      "step": 442000
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.7615869321799462e-05,
+      "loss": 1.3915,
+      "step": 442500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7579277083744996e-05,
+      "loss": 1.3932,
+      "step": 443000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7542684845690532e-05,
+      "loss": 1.4088,
+      "step": 443500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.750609260763607e-05,
+      "loss": 1.3854,
+      "step": 444000
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7469500369581603e-05,
+      "loss": 1.3995,
+      "step": 444500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.7432908131527143e-05,
+      "loss": 1.3955,
+      "step": 445000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.739631589347268e-05,
+      "loss": 1.4035,
+      "step": 445500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7359723655418213e-05,
+      "loss": 1.3868,
+      "step": 446000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.732313141736375e-05,
+      "loss": 1.3891,
+      "step": 446500
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.7286539179309287e-05,
+      "loss": 1.3975,
+      "step": 447000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.724994694125482e-05,
+      "loss": 1.3937,
+      "step": 447500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7213354703200357e-05,
+      "loss": 1.4002,
+      "step": 448000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7176762465145894e-05,
+      "loss": 1.4023,
+      "step": 448500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7140170227091427e-05,
+      "loss": 1.4124,
+      "step": 449000
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.7103577989036968e-05,
+      "loss": 1.3899,
+      "step": 449500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7066985750982504e-05,
+      "loss": 1.393,
+      "step": 450000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.7030393512928038e-05,
+      "loss": 1.3891,
+      "step": 450500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6993801274873575e-05,
+      "loss": 1.3943,
+      "step": 451000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.695720903681911e-05,
+      "loss": 1.3819,
+      "step": 451500
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.6920616798764645e-05,
+      "loss": 1.395,
+      "step": 452000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6884024560710182e-05,
+      "loss": 1.3971,
+      "step": 452500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.684743232265572e-05,
+      "loss": 1.3918,
+      "step": 453000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6810840084601256e-05,
+      "loss": 1.401,
+      "step": 453500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.6774247846546792e-05,
+      "loss": 1.3783,
+      "step": 454000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.673765560849233e-05,
+      "loss": 1.3959,
+      "step": 454500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6701063370437863e-05,
+      "loss": 1.3907,
+      "step": 455000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.66644711323834e-05,
+      "loss": 1.3914,
+      "step": 455500
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.6627878894328936e-05,
+      "loss": 1.3918,
+      "step": 456000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.659128665627447e-05,
+      "loss": 1.3777,
+      "step": 456500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6554694418220007e-05,
+      "loss": 1.3793,
+      "step": 457000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6518102180165543e-05,
+      "loss": 1.3961,
+      "step": 457500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.648150994211108e-05,
+      "loss": 1.3891,
+      "step": 458000
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.6444917704056617e-05,
+      "loss": 1.3831,
+      "step": 458500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6408325466002154e-05,
+      "loss": 1.3866,
+      "step": 459000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6371733227947687e-05,
+      "loss": 1.3961,
+      "step": 459500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6335140989893224e-05,
+      "loss": 1.3891,
+      "step": 460000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.629854875183876e-05,
+      "loss": 1.375,
+      "step": 460500
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.6261956513784295e-05,
+      "loss": 1.3786,
+      "step": 461000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.622536427572983e-05,
+      "loss": 1.3825,
+      "step": 461500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.618877203767537e-05,
+      "loss": 1.3741,
+      "step": 462000
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6152179799620905e-05,
+      "loss": 1.3948,
+      "step": 462500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.6115587561566442e-05,
+      "loss": 1.388,
+      "step": 463000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.607899532351198e-05,
+      "loss": 1.3855,
+      "step": 463500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.6042403085457512e-05,
+      "loss": 1.3861,
+      "step": 464000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.600581084740305e-05,
+      "loss": 1.3949,
+      "step": 464500
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.5969218609348586e-05,
+      "loss": 1.39,
+      "step": 465000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.593262637129412e-05,
+      "loss": 1.3827,
+      "step": 465500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5896034133239656e-05,
+      "loss": 1.3773,
+      "step": 466000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5859441895185196e-05,
+      "loss": 1.3845,
+      "step": 466500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.582284965713073e-05,
+      "loss": 1.3946,
+      "step": 467000
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.5786257419076267e-05,
+      "loss": 1.3823,
+      "step": 467500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5749665181021803e-05,
+      "loss": 1.377,
+      "step": 468000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5713072942967337e-05,
+      "loss": 1.3842,
+      "step": 468500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5676480704912874e-05,
+      "loss": 1.3839,
+      "step": 469000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.563988846685841e-05,
+      "loss": 1.3694,
+      "step": 469500
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.5603296228803947e-05,
+      "loss": 1.3808,
+      "step": 470000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.556670399074948e-05,
+      "loss": 1.3926,
+      "step": 470500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.553011175269502e-05,
+      "loss": 1.3834,
+      "step": 471000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5493519514640555e-05,
+      "loss": 1.386,
+      "step": 471500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.545692727658609e-05,
+      "loss": 1.3794,
+      "step": 472000
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.5420335038531628e-05,
+      "loss": 1.3759,
+      "step": 472500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.538374280047716e-05,
+      "loss": 1.388,
+      "step": 473000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.53471505624227e-05,
+      "loss": 1.3761,
+      "step": 473500
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5310558324368235e-05,
+      "loss": 1.3866,
+      "step": 474000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.5273966086313772e-05,
+      "loss": 1.3835,
+      "step": 474500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5237373848259309e-05,
+      "loss": 1.378,
+      "step": 475000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5200781610204846e-05,
+      "loss": 1.3784,
+      "step": 475500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5164189372150381e-05,
+      "loss": 1.3927,
+      "step": 476000
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5127597134095916e-05,
+      "loss": 1.3864,
+      "step": 476500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.5091004896041453e-05,
+      "loss": 1.386,
+      "step": 477000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5054412657986988e-05,
+      "loss": 1.3788,
+      "step": 477500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.5017820419932523e-05,
+      "loss": 1.3707,
+      "step": 478000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.498122818187806e-05,
+      "loss": 1.3688,
+      "step": 478500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.4944635943823595e-05,
+      "loss": 1.3727,
+      "step": 479000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4908043705769134e-05,
+      "loss": 1.3761,
+      "step": 479500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.487145146771467e-05,
+      "loss": 1.3775,
+      "step": 480000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4834859229660206e-05,
+      "loss": 1.3772,
+      "step": 480500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4798266991605741e-05,
+      "loss": 1.3652,
+      "step": 481000
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.4761674753551278e-05,
+      "loss": 1.3798,
+      "step": 481500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4725082515496813e-05,
+      "loss": 1.379,
+      "step": 482000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.468849027744235e-05,
+      "loss": 1.3702,
+      "step": 482500
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4651898039387885e-05,
+      "loss": 1.365,
+      "step": 483000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.4615305801333423e-05,
+      "loss": 1.369,
+      "step": 483500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4578713563278958e-05,
+      "loss": 1.3969,
+      "step": 484000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4542121325224495e-05,
+      "loss": 1.3779,
+      "step": 484500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.450552908717003e-05,
+      "loss": 1.3796,
+      "step": 485000
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4468936849115566e-05,
+      "loss": 1.3797,
+      "step": 485500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.4432344611061102e-05,
+      "loss": 1.3757,
+      "step": 486000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4395752373006638e-05,
+      "loss": 1.3746,
+      "step": 486500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4359160134952174e-05,
+      "loss": 1.3759,
+      "step": 487000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.432256789689771e-05,
+      "loss": 1.3638,
+      "step": 487500
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.4285975658843248e-05,
+      "loss": 1.3832,
+      "step": 488000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4249383420788783e-05,
+      "loss": 1.379,
+      "step": 488500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.421279118273432e-05,
+      "loss": 1.3694,
+      "step": 489000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4176198944679855e-05,
+      "loss": 1.3698,
+      "step": 489500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4139606706625392e-05,
+      "loss": 1.3798,
+      "step": 490000
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.4103014468570927e-05,
+      "loss": 1.3824,
+      "step": 490500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4066422230516462e-05,
+      "loss": 1.3781,
+      "step": 491000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.4029829992462e-05,
+      "loss": 1.378,
+      "step": 491500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3993237754407534e-05,
+      "loss": 1.3735,
+      "step": 492000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3956645516353073e-05,
+      "loss": 1.3645,
+      "step": 492500
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.3920053278298608e-05,
+      "loss": 1.387,
+      "step": 493000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3883461040244145e-05,
+      "loss": 1.3751,
+      "step": 493500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.384686880218968e-05,
+      "loss": 1.3807,
+      "step": 494000
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3810276564135217e-05,
+      "loss": 1.3781,
+      "step": 494500
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.3773684326080752e-05,
+      "loss": 1.3763,
+      "step": 495000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3737092088026287e-05,
+      "loss": 1.3676,
+      "step": 495500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3700499849971824e-05,
+      "loss": 1.3703,
+      "step": 496000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3663907611917362e-05,
+      "loss": 1.3836,
+      "step": 496500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3627315373862898e-05,
+      "loss": 1.3748,
+      "step": 497000
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.3590723135808434e-05,
+      "loss": 1.3646,
+      "step": 497500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.355413089775397e-05,
+      "loss": 1.3874,
+      "step": 498000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3517538659699505e-05,
+      "loss": 1.3635,
+      "step": 498500
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3480946421645042e-05,
+      "loss": 1.3712,
+      "step": 499000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.3444354183590577e-05,
+      "loss": 1.373,
+      "step": 499500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3407761945536112e-05,
+      "loss": 1.378,
+      "step": 500000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3371169707481649e-05,
+      "loss": 1.372,
+      "step": 500500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3334577469427187e-05,
+      "loss": 1.3841,
+      "step": 501000
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.3297985231372722e-05,
+      "loss": 1.3825,
+      "step": 501500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.326139299331826e-05,
+      "loss": 1.3811,
+      "step": 502000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3224800755263794e-05,
+      "loss": 1.37,
+      "step": 502500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.318820851720933e-05,
+      "loss": 1.3668,
+      "step": 503000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3151616279154866e-05,
+      "loss": 1.3691,
+      "step": 503500
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.3115024041100401e-05,
+      "loss": 1.3772,
+      "step": 504000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3078431803045937e-05,
+      "loss": 1.3675,
+      "step": 504500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3041839564991477e-05,
+      "loss": 1.3739,
+      "step": 505000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.3005247326937012e-05,
+      "loss": 1.3572,
+      "step": 505500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2968655088882547e-05,
+      "loss": 1.3754,
+      "step": 506000
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2932062850828084e-05,
+      "loss": 1.3654,
+      "step": 506500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2895470612773619e-05,
+      "loss": 1.3652,
+      "step": 507000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2858878374719154e-05,
+      "loss": 1.373,
+      "step": 507500
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2822286136664691e-05,
+      "loss": 1.3841,
+      "step": 508000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2785693898610226e-05,
+      "loss": 1.3595,
+      "step": 508500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2749101660555763e-05,
+      "loss": 1.367,
+      "step": 509000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2712509422501302e-05,
+      "loss": 1.3702,
+      "step": 509500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2675917184446837e-05,
+      "loss": 1.3707,
+      "step": 510000
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2639324946392372e-05,
+      "loss": 1.3656,
+      "step": 510500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2602732708337909e-05,
+      "loss": 1.3808,
+      "step": 511000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2566140470283444e-05,
+      "loss": 1.3641,
+      "step": 511500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2529548232228979e-05,
+      "loss": 1.3663,
+      "step": 512000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2492955994174517e-05,
+      "loss": 1.3672,
+      "step": 512500
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2456363756120053e-05,
+      "loss": 1.3626,
+      "step": 513000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2419771518065588e-05,
+      "loss": 1.3692,
+      "step": 513500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2383179280011125e-05,
+      "loss": 1.3803,
+      "step": 514000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.234658704195666e-05,
+      "loss": 1.3755,
+      "step": 514500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2309994803902197e-05,
+      "loss": 1.3726,
+      "step": 515000
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2273402565847733e-05,
+      "loss": 1.3691,
+      "step": 515500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2236810327793269e-05,
+      "loss": 1.3659,
+      "step": 516000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2200218089738805e-05,
+      "loss": 1.3664,
+      "step": 516500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2163625851684342e-05,
+      "loss": 1.3668,
+      "step": 517000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2127033613629877e-05,
+      "loss": 1.3751,
+      "step": 517500
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.2090441375575414e-05,
+      "loss": 1.3743,
+      "step": 518000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.205384913752095e-05,
+      "loss": 1.3634,
+      "step": 518500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.2017256899466486e-05,
+      "loss": 1.3559,
+      "step": 519000
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1980664661412021e-05,
+      "loss": 1.3654,
+      "step": 519500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1944072423357558e-05,
+      "loss": 1.3643,
+      "step": 520000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1907480185303093e-05,
+      "loss": 1.3557,
+      "step": 520500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.187088794724863e-05,
+      "loss": 1.3618,
+      "step": 521000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1834295709194167e-05,
+      "loss": 1.367,
+      "step": 521500
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1797703471139702e-05,
+      "loss": 1.3672,
+      "step": 522000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1761111233085239e-05,
+      "loss": 1.3629,
+      "step": 522500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1724518995030774e-05,
+      "loss": 1.3636,
+      "step": 523000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1687926756976311e-05,
+      "loss": 1.3646,
+      "step": 523500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1651334518921848e-05,
+      "loss": 1.3581,
+      "step": 524000
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1614742280867383e-05,
+      "loss": 1.3597,
+      "step": 524500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1578150042812918e-05,
+      "loss": 1.3571,
+      "step": 525000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1541557804758457e-05,
+      "loss": 1.3767,
+      "step": 525500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1504965566703992e-05,
+      "loss": 1.3722,
+      "step": 526000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1468373328649527e-05,
+      "loss": 1.3621,
+      "step": 526500
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1431781090595064e-05,
+      "loss": 1.357,
+      "step": 527000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.13951888525406e-05,
+      "loss": 1.3701,
+      "step": 527500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1358596614486136e-05,
+      "loss": 1.3566,
+      "step": 528000
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1322004376431673e-05,
+      "loss": 1.3705,
+      "step": 528500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1285412138377208e-05,
+      "loss": 1.3597,
+      "step": 529000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1248819900322743e-05,
+      "loss": 1.3713,
+      "step": 529500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1212227662268281e-05,
+      "loss": 1.3763,
+      "step": 530000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1175635424213816e-05,
+      "loss": 1.3662,
+      "step": 530500
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1139043186159352e-05,
+      "loss": 1.3618,
+      "step": 531000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1102450948104888e-05,
+      "loss": 1.366,
+      "step": 531500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1065858710050425e-05,
+      "loss": 1.37,
+      "step": 532000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.102926647199596e-05,
+      "loss": 1.3671,
+      "step": 532500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0992674233941497e-05,
+      "loss": 1.3631,
+      "step": 533000
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0956081995887032e-05,
+      "loss": 1.3664,
+      "step": 533500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.091948975783257e-05,
+      "loss": 1.3647,
+      "step": 534000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0882897519778106e-05,
+      "loss": 1.3509,
+      "step": 534500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0846305281723641e-05,
+      "loss": 1.3613,
+      "step": 535000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0809713043669176e-05,
+      "loss": 1.3708,
+      "step": 535500
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0773120805614713e-05,
+      "loss": 1.3521,
+      "step": 536000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.073652856756025e-05,
+      "loss": 1.3668,
+      "step": 536500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0699936329505785e-05,
+      "loss": 1.3668,
+      "step": 537000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0663344091451322e-05,
+      "loss": 1.353,
+      "step": 537500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0626751853396857e-05,
+      "loss": 1.3575,
+      "step": 538000
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0590159615342394e-05,
+      "loss": 1.358,
+      "step": 538500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.055356737728793e-05,
+      "loss": 1.3632,
+      "step": 539000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0516975139233466e-05,
+      "loss": 1.3578,
+      "step": 539500
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0480382901179001e-05,
+      "loss": 1.3506,
+      "step": 540000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.044379066312454e-05,
+      "loss": 1.3577,
+      "step": 540500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0407198425070075e-05,
+      "loss": 1.3575,
+      "step": 541000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.037060618701561e-05,
+      "loss": 1.3581,
+      "step": 541500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0334013948961147e-05,
+      "loss": 1.357,
+      "step": 542000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0297421710906682e-05,
+      "loss": 1.3578,
+      "step": 542500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0260829472852219e-05,
+      "loss": 1.3503,
+      "step": 543000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0224237234797756e-05,
+      "loss": 1.3621,
+      "step": 543500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.018764499674329e-05,
+      "loss": 1.3604,
+      "step": 544000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0151052758688828e-05,
+      "loss": 1.359,
+      "step": 544500
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0114460520634364e-05,
+      "loss": 1.3771,
+      "step": 545000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.00778682825799e-05,
+      "loss": 1.3578,
+      "step": 545500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0041276044525436e-05,
+      "loss": 1.3601,
+      "step": 546000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0004683806470972e-05,
+      "loss": 1.3586,
+      "step": 546500
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.968091568416508e-06,
+      "loss": 1.3545,
+      "step": 547000
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 9.931499330362043e-06,
+      "loss": 1.3452,
+      "step": 547500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.89490709230758e-06,
+      "loss": 1.3501,
+      "step": 548000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.858314854253115e-06,
+      "loss": 1.3568,
+      "step": 548500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.821722616198652e-06,
+      "loss": 1.3667,
+      "step": 549000
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 9.785130378144189e-06,
+      "loss": 1.3636,
+      "step": 549500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.748538140089724e-06,
+      "loss": 1.3585,
+      "step": 550000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.711945902035261e-06,
+      "loss": 1.3552,
+      "step": 550500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.675353663980796e-06,
+      "loss": 1.3517,
+      "step": 551000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.638761425926333e-06,
+      "loss": 1.3478,
+      "step": 551500
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 9.60216918787187e-06,
+      "loss": 1.3516,
+      "step": 552000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.565576949817405e-06,
+      "loss": 1.3541,
+      "step": 552500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.52898471176294e-06,
+      "loss": 1.3551,
+      "step": 553000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.492392473708479e-06,
+      "loss": 1.359,
+      "step": 553500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.455800235654014e-06,
+      "loss": 1.3485,
+      "step": 554000
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 9.419207997599549e-06,
+      "loss": 1.3409,
+      "step": 554500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.382615759545086e-06,
+      "loss": 1.3536,
+      "step": 555000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.346023521490623e-06,
+      "loss": 1.3572,
+      "step": 555500
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.309431283436158e-06,
+      "loss": 1.3559,
+      "step": 556000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 9.272839045381695e-06,
+      "loss": 1.3642,
+      "step": 556500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.23624680732723e-06,
+      "loss": 1.3652,
+      "step": 557000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.199654569272765e-06,
+      "loss": 1.361,
+      "step": 557500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.163062331218304e-06,
+      "loss": 1.3511,
+      "step": 558000
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.126470093163839e-06,
+      "loss": 1.3614,
+      "step": 558500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 9.089877855109374e-06,
+      "loss": 1.3533,
+      "step": 559000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.05328561705491e-06,
+      "loss": 1.3564,
+      "step": 559500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 9.016693379000447e-06,
+      "loss": 1.345,
+      "step": 560000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.980101140945983e-06,
+      "loss": 1.3464,
+      "step": 560500
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 8.94350890289152e-06,
+      "loss": 1.3601,
+      "step": 561000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.906916664837055e-06,
+      "loss": 1.3547,
+      "step": 561500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.870324426782591e-06,
+      "loss": 1.3582,
+      "step": 562000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.833732188728128e-06,
+      "loss": 1.3534,
+      "step": 562500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.797139950673663e-06,
+      "loss": 1.3396,
+      "step": 563000
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 8.760547712619199e-06,
+      "loss": 1.3508,
+      "step": 563500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.723955474564735e-06,
+      "loss": 1.3511,
+      "step": 564000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.687363236510272e-06,
+      "loss": 1.3571,
+      "step": 564500
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.650770998455807e-06,
+      "loss": 1.3538,
+      "step": 565000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 8.614178760401344e-06,
+      "loss": 1.3447,
+      "step": 565500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.57758652234688e-06,
+      "loss": 1.3467,
+      "step": 566000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.540994284292416e-06,
+      "loss": 1.3417,
+      "step": 566500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.504402046237953e-06,
+      "loss": 1.346,
+      "step": 567000
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.467809808183488e-06,
+      "loss": 1.344,
+      "step": 567500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 8.431217570129023e-06,
+      "loss": 1.3515,
+      "step": 568000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.394625332074562e-06,
+      "loss": 1.3618,
+      "step": 568500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.358033094020097e-06,
+      "loss": 1.3459,
+      "step": 569000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.321440855965632e-06,
+      "loss": 1.3507,
+      "step": 569500
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 8.284848617911169e-06,
+      "loss": 1.3528,
+      "step": 570000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.248256379856706e-06,
+      "loss": 1.3498,
+      "step": 570500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.211664141802241e-06,
+      "loss": 1.3529,
+      "step": 571000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.175071903747778e-06,
+      "loss": 1.3504,
+      "step": 571500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.138479665693313e-06,
+      "loss": 1.3506,
+      "step": 572000
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 8.10188742763885e-06,
+      "loss": 1.3471,
+      "step": 572500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.065295189584387e-06,
+      "loss": 1.3551,
+      "step": 573000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 8.028702951529922e-06,
+      "loss": 1.3549,
+      "step": 573500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.992110713475459e-06,
+      "loss": 1.3472,
+      "step": 574000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.955518475420994e-06,
+      "loss": 1.3395,
+      "step": 574500
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 7.91892623736653e-06,
+      "loss": 1.3591,
+      "step": 575000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.882333999312066e-06,
+      "loss": 1.3509,
+      "step": 575500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.845741761257603e-06,
+      "loss": 1.3443,
+      "step": 576000
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.809149523203138e-06,
+      "loss": 1.3603,
+      "step": 576500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 7.772557285148674e-06,
+      "loss": 1.3436,
+      "step": 577000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.735965047094211e-06,
+      "loss": 1.3421,
+      "step": 577500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.699372809039746e-06,
+      "loss": 1.3551,
+      "step": 578000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.662780570985283e-06,
+      "loss": 1.3581,
+      "step": 578500
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.6261883329308184e-06,
+      "loss": 1.3474,
+      "step": 579000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 7.589596094876355e-06,
+      "loss": 1.3456,
+      "step": 579500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.553003856821891e-06,
+      "loss": 1.3512,
+      "step": 580000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.516411618767427e-06,
+      "loss": 1.3494,
+      "step": 580500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.479819380712963e-06,
+      "loss": 1.3323,
+      "step": 581000
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 7.4432271426585e-06,
+      "loss": 1.3456,
+      "step": 581500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.406634904604036e-06,
+      "loss": 1.3429,
+      "step": 582000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.370042666549572e-06,
+      "loss": 1.3494,
+      "step": 582500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.333450428495107e-06,
+      "loss": 1.3479,
+      "step": 583000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.296858190440645e-06,
+      "loss": 1.3486,
+      "step": 583500
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 7.260265952386181e-06,
+      "loss": 1.3491,
+      "step": 584000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.223673714331716e-06,
+      "loss": 1.3462,
+      "step": 584500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.187081476277252e-06,
+      "loss": 1.3477,
+      "step": 585000
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.150489238222789e-06,
+      "loss": 1.3465,
+      "step": 585500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 7.113897000168325e-06,
+      "loss": 1.3472,
+      "step": 586000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.077304762113861e-06,
+      "loss": 1.3523,
+      "step": 586500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.040712524059397e-06,
+      "loss": 1.3461,
+      "step": 587000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 7.004120286004932e-06,
+      "loss": 1.3461,
+      "step": 587500
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.96752804795047e-06,
+      "loss": 1.3466,
+      "step": 588000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 6.930935809896006e-06,
+      "loss": 1.3501,
+      "step": 588500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.894343571841541e-06,
+      "loss": 1.3525,
+      "step": 589000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.857751333787077e-06,
+      "loss": 1.3393,
+      "step": 589500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.8211590957326144e-06,
+      "loss": 1.3412,
+      "step": 590000
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 6.78456685767815e-06,
+      "loss": 1.3463,
+      "step": 590500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7479746196236856e-06,
+      "loss": 1.3364,
+      "step": 591000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.7113823815692216e-06,
+      "loss": 1.3401,
+      "step": 591500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.674790143514758e-06,
+      "loss": 1.3501,
+      "step": 592000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.638197905460294e-06,
+      "loss": 1.3375,
+      "step": 592500
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 6.60160566740583e-06,
+      "loss": 1.3514,
+      "step": 593000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.565013429351366e-06,
+      "loss": 1.3331,
+      "step": 593500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.5284211912969015e-06,
+      "loss": 1.3456,
+      "step": 594000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.491828953242439e-06,
+      "loss": 1.3397,
+      "step": 594500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.455236715187974e-06,
+      "loss": 1.3495,
+      "step": 595000
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 6.41864447713351e-06,
+      "loss": 1.337,
+      "step": 595500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.382052239079046e-06,
+      "loss": 1.3484,
+      "step": 596000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.345460001024583e-06,
+      "loss": 1.3352,
+      "step": 596500
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.308867762970119e-06,
+      "loss": 1.3477,
+      "step": 597000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 6.272275524915655e-06,
+      "loss": 1.3466,
+      "step": 597500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.235683286861192e-06,
+      "loss": 1.3371,
+      "step": 598000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.199091048806727e-06,
+      "loss": 1.3526,
+      "step": 598500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.162498810752264e-06,
+      "loss": 1.3368,
+      "step": 599000
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.1259065726978e-06,
+      "loss": 1.3502,
+      "step": 599500
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 6.089314334643336e-06,
+      "loss": 1.3446,
+      "step": 600000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.052722096588872e-06,
+      "loss": 1.3531,
+      "step": 600500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 6.016129858534408e-06,
+      "loss": 1.3437,
+      "step": 601000
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.979537620479944e-06,
+      "loss": 1.3506,
+      "step": 601500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.94294538242548e-06,
+      "loss": 1.3396,
+      "step": 602000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.906353144371017e-06,
+      "loss": 1.3549,
+      "step": 602500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.869760906316552e-06,
+      "loss": 1.3492,
+      "step": 603000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.833168668262089e-06,
+      "loss": 1.3498,
+      "step": 603500
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.796576430207625e-06,
+      "loss": 1.3447,
+      "step": 604000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.759984192153161e-06,
+      "loss": 1.3409,
+      "step": 604500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.723391954098697e-06,
+      "loss": 1.3428,
+      "step": 605000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6867997160442335e-06,
+      "loss": 1.341,
+      "step": 605500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.650207477989769e-06,
+      "loss": 1.3304,
+      "step": 606000
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.6136152399353055e-06,
+      "loss": 1.3499,
+      "step": 606500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5770230018808415e-06,
+      "loss": 1.3352,
+      "step": 607000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5404307638263774e-06,
+      "loss": 1.3362,
+      "step": 607500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.5038385257719134e-06,
+      "loss": 1.3503,
+      "step": 608000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.467246287717449e-06,
+      "loss": 1.3452,
+      "step": 608500
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.430654049662985e-06,
+      "loss": 1.3399,
+      "step": 609000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.394061811608521e-06,
+      "loss": 1.3465,
+      "step": 609500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.357469573554058e-06,
+      "loss": 1.3435,
+      "step": 610000
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.320877335499594e-06,
+      "loss": 1.3437,
+      "step": 610500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.28428509744513e-06,
+      "loss": 1.3381,
+      "step": 611000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.247692859390666e-06,
+      "loss": 1.3415,
+      "step": 611500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.211100621336203e-06,
+      "loss": 1.3447,
+      "step": 612000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.174508383281738e-06,
+      "loss": 1.3381,
+      "step": 612500
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.137916145227275e-06,
+      "loss": 1.3408,
+      "step": 613000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.101323907172811e-06,
+      "loss": 1.3424,
+      "step": 613500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.064731669118347e-06,
+      "loss": 1.3417,
+      "step": 614000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.028139431063883e-06,
+      "loss": 1.3431,
+      "step": 614500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.991547193009419e-06,
+      "loss": 1.3382,
+      "step": 615000
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.954954954954955e-06,
+      "loss": 1.3343,
+      "step": 615500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 4.918362716900491e-06,
+      "loss": 1.3464,
+      "step": 616000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.881770478846028e-06,
+      "loss": 1.3378,
+      "step": 616500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.845178240791563e-06,
+      "loss": 1.3455,
+      "step": 617000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.8085860027371e-06,
+      "loss": 1.3371,
+      "step": 617500
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 4.771993764682636e-06,
+      "loss": 1.3374,
+      "step": 618000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.735401526628172e-06,
+      "loss": 1.3382,
+      "step": 618500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.698809288573708e-06,
+      "loss": 1.3436,
+      "step": 619000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.6622170505192446e-06,
+      "loss": 1.3405,
+      "step": 619500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.62562481246478e-06,
+      "loss": 1.3298,
+      "step": 620000
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 4.5890325744103165e-06,
+      "loss": 1.3321,
+      "step": 620500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5524403363558525e-06,
+      "loss": 1.3368,
+      "step": 621000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.5158480983013885e-06,
+      "loss": 1.3388,
+      "step": 621500
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4792558602469245e-06,
+      "loss": 1.3348,
+      "step": 622000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 4.4426636221924605e-06,
+      "loss": 1.3473,
+      "step": 622500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.4060713841379965e-06,
+      "loss": 1.3361,
+      "step": 623000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.3694791460835325e-06,
+      "loss": 1.3396,
+      "step": 623500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.332886908029069e-06,
+      "loss": 1.3264,
+      "step": 624000
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.296294669974605e-06,
+      "loss": 1.3346,
+      "step": 624500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 4.259702431920141e-06,
+      "loss": 1.3415,
+      "step": 625000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.223110193865677e-06,
+      "loss": 1.3327,
+      "step": 625500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.186517955811214e-06,
+      "loss": 1.3395,
+      "step": 626000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.149925717756749e-06,
+      "loss": 1.3383,
+      "step": 626500
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 4.113333479702286e-06,
+      "loss": 1.3418,
+      "step": 627000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.076741241647822e-06,
+      "loss": 1.3266,
+      "step": 627500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.040149003593358e-06,
+      "loss": 1.346,
+      "step": 628000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 4.003556765538894e-06,
+      "loss": 1.3316,
+      "step": 628500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.966964527484431e-06,
+      "loss": 1.334,
+      "step": 629000
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.930372289429966e-06,
+      "loss": 1.3275,
+      "step": 629500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.893780051375502e-06,
+      "loss": 1.3259,
+      "step": 630000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.857187813321039e-06,
+      "loss": 1.334,
+      "step": 630500
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.820595575266574e-06,
+      "loss": 1.3437,
+      "step": 631000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.784003337212111e-06,
+      "loss": 1.3255,
+      "step": 631500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.747411099157647e-06,
+      "loss": 1.3243,
+      "step": 632000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.7108188611031833e-06,
+      "loss": 1.3374,
+      "step": 632500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.674226623048719e-06,
+      "loss": 1.3236,
+      "step": 633000
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6376343849942552e-06,
+      "loss": 1.3267,
+      "step": 633500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6010421469397912e-06,
+      "loss": 1.329,
+      "step": 634000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5644499088853276e-06,
+      "loss": 1.3338,
+      "step": 634500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.5278576708308636e-06,
+      "loss": 1.3339,
+      "step": 635000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4912654327764e-06,
+      "loss": 1.3229,
+      "step": 635500
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4546731947219356e-06,
+      "loss": 1.3332,
+      "step": 636000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.4180809566674716e-06,
+      "loss": 1.3309,
+      "step": 636500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.381488718613008e-06,
+      "loss": 1.3393,
+      "step": 637000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.344896480558544e-06,
+      "loss": 1.3255,
+      "step": 637500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.3083042425040804e-06,
+      "loss": 1.3316,
+      "step": 638000
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.271712004449616e-06,
+      "loss": 1.3307,
+      "step": 638500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.2351197663951524e-06,
+      "loss": 1.3238,
+      "step": 639000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1985275283406884e-06,
+      "loss": 1.3271,
+      "step": 639500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1619352902862248e-06,
+      "loss": 1.3326,
+      "step": 640000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.1253430522317608e-06,
+      "loss": 1.3344,
+      "step": 640500
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.0887508141772968e-06,
+      "loss": 1.3328,
+      "step": 641000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.0521585761228328e-06,
+      "loss": 1.3275,
+      "step": 641500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.015566338068369e-06,
+      "loss": 1.33,
+      "step": 642000
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.978974100013905e-06,
+      "loss": 1.3317,
+      "step": 642500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 2.942381861959441e-06,
+      "loss": 1.336,
+      "step": 643000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.9057896239049776e-06,
+      "loss": 1.3303,
+      "step": 643500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8691973858505135e-06,
+      "loss": 1.3353,
+      "step": 644000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.8326051477960495e-06,
+      "loss": 1.3191,
+      "step": 644500
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.796012909741586e-06,
+      "loss": 1.3381,
+      "step": 645000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 2.759420671687122e-06,
+      "loss": 1.3213,
+      "step": 645500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.722828433632658e-06,
+      "loss": 1.3411,
+      "step": 646000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6862361955781943e-06,
+      "loss": 1.3401,
+      "step": 646500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6496439575237303e-06,
+      "loss": 1.3324,
+      "step": 647000
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 2.6130517194692663e-06,
+      "loss": 1.3351,
+      "step": 647500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5764594814148023e-06,
+      "loss": 1.3283,
+      "step": 648000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5398672433603383e-06,
+      "loss": 1.3221,
+      "step": 648500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.5032750053058747e-06,
+      "loss": 1.338,
+      "step": 649000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4666827672514107e-06,
+      "loss": 1.3342,
+      "step": 649500
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 2.4300905291969467e-06,
+      "loss": 1.3353,
+      "step": 650000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.393498291142483e-06,
+      "loss": 1.3423,
+      "step": 650500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.356906053088019e-06,
+      "loss": 1.3436,
+      "step": 651000
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.320313815033555e-06,
+      "loss": 1.3203,
+      "step": 651500
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 2.2837215769790915e-06,
+      "loss": 1.3413,
+      "step": 652000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2471293389246275e-06,
+      "loss": 1.3269,
+      "step": 652500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.2105371008701635e-06,
+      "loss": 1.3249,
+      "step": 653000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.1739448628157e-06,
+      "loss": 1.3176,
+      "step": 653500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.137352624761236e-06,
+      "loss": 1.3226,
+      "step": 654000
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 2.100760386706772e-06,
+      "loss": 1.329,
+      "step": 654500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.064168148652308e-06,
+      "loss": 1.3336,
+      "step": 655000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 2.027575910597844e-06,
+      "loss": 1.3308,
+      "step": 655500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9909836725433802e-06,
+      "loss": 1.3268,
+      "step": 656000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9543914344889162e-06,
+      "loss": 1.3149,
+      "step": 656500
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.9177991964344522e-06,
+      "loss": 1.3322,
+      "step": 657000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8812069583799884e-06,
+      "loss": 1.3207,
+      "step": 657500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8446147203255246e-06,
+      "loss": 1.3294,
+      "step": 658000
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.8080224822710608e-06,
+      "loss": 1.3317,
+      "step": 658500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.7714302442165968e-06,
+      "loss": 1.3212,
+      "step": 659000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.734838006162133e-06,
+      "loss": 1.328,
+      "step": 659500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6982457681076692e-06,
+      "loss": 1.3315,
+      "step": 660000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6616535300532052e-06,
+      "loss": 1.3311,
+      "step": 660500
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6250612919987414e-06,
+      "loss": 1.3252,
+      "step": 661000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.5884690539442776e-06,
+      "loss": 1.3288,
+      "step": 661500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5518768158898136e-06,
+      "loss": 1.3344,
+      "step": 662000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.5152845778353496e-06,
+      "loss": 1.3368,
+      "step": 662500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4786923397808858e-06,
+      "loss": 1.3204,
+      "step": 663000
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.4421001017264218e-06,
+      "loss": 1.3299,
+      "step": 663500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.405507863671958e-06,
+      "loss": 1.3284,
+      "step": 664000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.368915625617494e-06,
+      "loss": 1.3319,
+      "step": 664500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.3323233875630302e-06,
+      "loss": 1.3274,
+      "step": 665000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2957311495085664e-06,
+      "loss": 1.3265,
+      "step": 665500
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.2591389114541024e-06,
+      "loss": 1.3263,
+      "step": 666000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.2225466733996386e-06,
+      "loss": 1.3355,
+      "step": 666500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1859544353451748e-06,
+      "loss": 1.3206,
+      "step": 667000
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1493621972907108e-06,
+      "loss": 1.3317,
+      "step": 667500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.1127699592362467e-06,
+      "loss": 1.3282,
+      "step": 668000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.076177721181783e-06,
+      "loss": 1.3227,
+      "step": 668500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0395854831273191e-06,
+      "loss": 1.327,
+      "step": 669000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.0029932450728551e-06,
+      "loss": 1.3196,
+      "step": 669500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.664010070183913e-07,
+      "loss": 1.3497,
+      "step": 670000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 9.298087689639275e-07,
+      "loss": 1.3402,
+      "step": 670500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.932165309094634e-07,
+      "loss": 1.3344,
+      "step": 671000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.566242928549996e-07,
+      "loss": 1.3278,
+      "step": 671500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.200320548005357e-07,
+      "loss": 1.3247,
+      "step": 672000
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.834398167460719e-07,
+      "loss": 1.3366,
+      "step": 672500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.46847578691608e-07,
+      "loss": 1.3247,
+      "step": 673000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.10255340637144e-07,
+      "loss": 1.3243,
+      "step": 673500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.736631025826802e-07,
+      "loss": 1.3274,
+      "step": 674000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.370708645282163e-07,
+      "loss": 1.3312,
+      "step": 674500
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 6.004786264737525e-07,
+      "loss": 1.3387,
+      "step": 675000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.638863884192885e-07,
+      "loss": 1.3341,
+      "step": 675500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 5.272941503648247e-07,
+      "loss": 1.3356,
+      "step": 676000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.907019123103608e-07,
+      "loss": 1.3261,
+      "step": 676500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.541096742558968e-07,
+      "loss": 1.3243,
+      "step": 677000
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.1751743620143297e-07,
+      "loss": 1.3181,
+      "step": 677500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.8092519814696907e-07,
+      "loss": 1.3265,
+      "step": 678000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.4433296009250516e-07,
+      "loss": 1.3207,
+      "step": 678500
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.077407220380413e-07,
+      "loss": 1.3251,
+      "step": 679000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 2.711484839835774e-07,
+      "loss": 1.3292,
+      "step": 679500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 2.3455624592911355e-07,
+      "loss": 1.3304,
+      "step": 680000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.9796400787464965e-07,
+      "loss": 1.3353,
+      "step": 680500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6137176982018574e-07,
+      "loss": 1.329,
+      "step": 681000
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.2477953176572186e-07,
+      "loss": 1.3176,
+      "step": 681500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.818729371125797e-08,
+      "loss": 1.3152,
+      "step": 682000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 5.159505565679408e-08,
+      "loss": 1.3272,
+      "step": 682500
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.5002817602330194e-08,
+      "loss": 1.3272,
+      "step": 683000
+    },
+    {
+      "epoch": 3.0,
+      "step": 683205,
+      "total_flos": 1.79822844129024e+18,
+      "train_loss": 1.4489240925217872,
+      "train_runtime": 502024.9182,
+      "train_samples_per_second": 13.609,
+      "train_steps_per_second": 1.361
+    }
+  ],
+  "max_steps": 683205,
+  "num_train_epochs": 3,
+  "total_flos": 1.79822844129024e+18,
+  "trial_name": null,
+  "trial_params": null
+}