diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5759 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "global_step": 478150,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9947715152148914e-05,
+      "loss": 1.9286,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.989543030429782e-05,
+      "loss": 1.4178,
+      "step": 1000
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.984314545644672e-05,
+      "loss": 1.3079,
+      "step": 1500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.979086060859563e-05,
+      "loss": 1.22,
+      "step": 2000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.973857576074454e-05,
+      "loss": 1.2108,
+      "step": 2500
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.9686290912893443e-05,
+      "loss": 1.1095,
+      "step": 3000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.9634006065042355e-05,
+      "loss": 1.1086,
+      "step": 3500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 4.958172121719126e-05,
+      "loss": 1.0555,
+      "step": 4000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 4.9529436369340164e-05,
+      "loss": 1.0431,
+      "step": 4500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 4.9477151521489075e-05,
+      "loss": 1.0312,
+      "step": 5000
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 4.942486667363798e-05,
+      "loss": 1.038,
+      "step": 5500
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 4.937258182578689e-05,
+      "loss": 0.9589,
+      "step": 6000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 4.9320296977935796e-05,
+      "loss": 0.9573,
+      "step": 6500
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 4.926801213008471e-05,
+      "loss": 0.9399,
+      "step": 7000
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 4.921572728223361e-05,
+      "loss": 0.9621,
+      "step": 7500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 4.9163442434382516e-05,
+      "loss": 0.9075,
+      "step": 8000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 4.911115758653143e-05,
+      "loss": 0.9014,
+      "step": 8500
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 4.905887273868033e-05,
+      "loss": 0.9125,
+      "step": 9000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 4.9006587890829244e-05,
+      "loss": 0.8829,
+      "step": 9500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 4.895430304297814e-05,
+      "loss": 0.9324,
+      "step": 10000
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 4.890201819512705e-05,
+      "loss": 0.8304,
+      "step": 10500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 4.8849733347275964e-05,
+      "loss": 0.8362,
+      "step": 11000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 4.879744849942487e-05,
+      "loss": 0.8292,
+      "step": 11500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 4.874516365157378e-05,
+      "loss": 0.846,
+      "step": 12000
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 4.8692878803722685e-05,
+      "loss": 0.8455,
+      "step": 12500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 4.864059395587159e-05,
+      "loss": 0.844,
+      "step": 13000
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 4.8588309108020494e-05,
+      "loss": 0.8488,
+      "step": 13500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 4.8536024260169405e-05,
+      "loss": 0.8259,
+      "step": 14000
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 4.848373941231831e-05,
+      "loss": 0.8309,
+      "step": 14500
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 4.843145456446722e-05,
+      "loss": 0.8033,
+      "step": 15000
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 4.837916971661613e-05,
+      "loss": 0.8117,
+      "step": 15500
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 4.832688486876503e-05,
+      "loss": 0.8268,
+      "step": 16000
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 4.827460002091394e-05,
+      "loss": 0.7868,
+      "step": 16500
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 4.8222315173062846e-05,
+      "loss": 0.8446,
+      "step": 17000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 4.817003032521176e-05,
+      "loss": 0.7891,
+      "step": 17500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 4.811774547736066e-05,
+      "loss": 0.7854,
+      "step": 18000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 4.806546062950957e-05,
+      "loss": 0.8087,
+      "step": 18500
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 4.801317578165848e-05,
+      "loss": 0.7793,
+      "step": 19000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 4.796089093380738e-05,
+      "loss": 0.7704,
+      "step": 19500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 4.7908606085956294e-05,
+      "loss": 0.7703,
+      "step": 20000
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 4.78563212381052e-05,
+      "loss": 0.7493,
+      "step": 20500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 4.780403639025411e-05,
+      "loss": 0.7818,
+      "step": 21000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 4.7751751542403015e-05,
+      "loss": 0.7732,
+      "step": 21500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 4.769946669455192e-05,
+      "loss": 0.7606,
+      "step": 22000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 4.764718184670083e-05,
+      "loss": 0.7694,
+      "step": 22500
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 4.7594896998849735e-05,
+      "loss": 0.7524,
+      "step": 23000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 4.7542612150998646e-05,
+      "loss": 0.7691,
+      "step": 23500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 4.749032730314755e-05,
+      "loss": 0.7804,
+      "step": 24000
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.7438042455296456e-05,
+      "loss": 0.8009,
+      "step": 24500
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.738575760744536e-05,
+      "loss": 0.7812,
+      "step": 25000
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733347275959427e-05,
+      "loss": 0.7623,
+      "step": 25500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.728118791174318e-05,
+      "loss": 0.7702,
+      "step": 26000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.722890306389209e-05,
+      "loss": 0.7669,
+      "step": 26500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.717661821604099e-05,
+      "loss": 0.7348,
+      "step": 27000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.71243333681899e-05,
+      "loss": 0.7423,
+      "step": 27500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 4.707204852033881e-05,
+      "loss": 0.7587,
+      "step": 28000
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 4.701976367248771e-05,
+      "loss": 0.736,
+      "step": 28500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 4.6967478824636624e-05,
+      "loss": 0.742,
+      "step": 29000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 4.6915193976785535e-05,
+      "loss": 0.7497,
+      "step": 29500
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 4.686290912893443e-05,
+      "loss": 0.7355,
+      "step": 30000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 4.6810624281083344e-05,
+      "loss": 0.7097,
+      "step": 30500
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 4.675833943323225e-05,
+      "loss": 0.7811,
+      "step": 31000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 4.670605458538116e-05,
+      "loss": 0.736,
+      "step": 31500
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 4.6653769737530065e-05,
+      "loss": 0.7308,
+      "step": 32000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.6601484889678976e-05,
+      "loss": 0.7214,
+      "step": 32500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654920004182788e-05,
+      "loss": 0.7604,
+      "step": 33000
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.6496915193976786e-05,
+      "loss": 0.7473,
+      "step": 33500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.64446303461257e-05,
+      "loss": 0.7504,
+      "step": 34000
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.63923454982746e-05,
+      "loss": 0.7095,
+      "step": 34500
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 4.634006065042351e-05,
+      "loss": 0.7317,
+      "step": 35000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 4.628777580257241e-05,
+      "loss": 0.7478,
+      "step": 35500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 4.623549095472132e-05,
+      "loss": 0.7212,
+      "step": 36000
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 4.618320610687023e-05,
+      "loss": 0.7275,
+      "step": 36500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 4.613092125901914e-05,
+      "loss": 0.7644,
+      "step": 37000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 4.607863641116805e-05,
+      "loss": 0.7487,
+      "step": 37500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 4.6026351563316954e-05,
+      "loss": 0.7298,
+      "step": 38000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 4.597406671546586e-05,
+      "loss": 0.7049,
+      "step": 38500
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 4.592178186761476e-05,
+      "loss": 0.7178,
+      "step": 39000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 4.5869497019763674e-05,
+      "loss": 0.7099,
+      "step": 39500
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 4.5817212171912586e-05,
+      "loss": 0.712,
+      "step": 40000
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 4.576492732406149e-05,
+      "loss": 0.7371,
+      "step": 40500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.57126424762104e-05,
+      "loss": 0.7406,
+      "step": 41000
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.56603576283593e-05,
+      "loss": 0.725,
+      "step": 41500
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.560807278050821e-05,
+      "loss": 0.715,
+      "step": 42000
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 4.5555787932657115e-05,
+      "loss": 0.7177,
+      "step": 42500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.550350308480603e-05,
+      "loss": 0.7345,
+      "step": 43000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.545121823695493e-05,
+      "loss": 0.6942,
+      "step": 43500
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.539893338910384e-05,
+      "loss": 0.7055,
+      "step": 44000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 4.534664854125275e-05,
+      "loss": 0.7284,
+      "step": 44500
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 4.529436369340165e-05,
+      "loss": 0.7422,
+      "step": 45000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 4.524207884555056e-05,
+      "loss": 0.7039,
+      "step": 45500
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.518979399769947e-05,
+      "loss": 0.7464,
+      "step": 46000
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.513750914984838e-05,
+      "loss": 0.7237,
+      "step": 46500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.5085224301997284e-05,
+      "loss": 0.7161,
+      "step": 47000
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.503293945414619e-05,
+      "loss": 0.7366,
+      "step": 47500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.49806546062951e-05,
+      "loss": 0.7339,
+      "step": 48000
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 4.4928369758444004e-05,
+      "loss": 0.6388,
+      "step": 48500
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 4.4876084910592916e-05,
+      "loss": 0.6579,
+      "step": 49000
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 4.482380006274182e-05,
+      "loss": 0.6537,
+      "step": 49500
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 4.4771515214890725e-05,
+      "loss": 0.6422,
+      "step": 50000
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 4.4719230367039636e-05,
+      "loss": 0.6894,
+      "step": 50500
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 4.466694551918854e-05,
+      "loss": 0.6774,
+      "step": 51000
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 4.461466067133745e-05,
+      "loss": 0.6605,
+      "step": 51500
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 4.456237582348636e-05,
+      "loss": 0.6687,
+      "step": 52000
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 4.451009097563527e-05,
+      "loss": 0.662,
+      "step": 52500
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 4.4457806127784166e-05,
+      "loss": 0.648,
+      "step": 53000
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 4.440552127993308e-05,
+      "loss": 0.6638,
+      "step": 53500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 4.435323643208198e-05,
+      "loss": 0.661,
+      "step": 54000
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 4.430095158423089e-05,
+      "loss": 0.6733,
+      "step": 54500
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 4.4248666736379804e-05,
+      "loss": 0.6828,
+      "step": 55000
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 4.41963818885287e-05,
+      "loss": 0.6613,
+      "step": 55500
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 4.4144097040677614e-05,
+      "loss": 0.708,
+      "step": 56000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 4.409181219282652e-05,
+      "loss": 0.6683,
+      "step": 56500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 4.403952734497543e-05,
+      "loss": 0.6493,
+      "step": 57000
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 4.3987242497124334e-05,
+      "loss": 0.6648,
+      "step": 57500
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 4.3934957649273246e-05,
+      "loss": 0.7077,
+      "step": 58000
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 4.388267280142215e-05,
+      "loss": 0.6687,
+      "step": 58500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 4.3830387953571055e-05,
+      "loss": 0.6465,
+      "step": 59000
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 4.3778103105719966e-05,
+      "loss": 0.6375,
+      "step": 59500
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 4.372581825786887e-05,
+      "loss": 0.677,
+      "step": 60000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 4.367353341001778e-05,
+      "loss": 0.6624,
+      "step": 60500
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 4.3621248562166687e-05,
+      "loss": 0.6619,
+      "step": 61000
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 4.356896371431559e-05,
+      "loss": 0.6867,
+      "step": 61500
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 4.35166788664645e-05,
+      "loss": 0.6723,
+      "step": 62000
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 4.346439401861341e-05,
+      "loss": 0.6585,
+      "step": 62500
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 4.341210917076232e-05,
+      "loss": 0.6564,
+      "step": 63000
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 4.335982432291122e-05,
+      "loss": 0.6608,
+      "step": 63500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 4.330753947506013e-05,
+      "loss": 0.657,
+      "step": 64000
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 4.325525462720903e-05,
+      "loss": 0.6957,
+      "step": 64500
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 4.3202969779357944e-05,
+      "loss": 0.6625,
+      "step": 65000
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 4.3150684931506855e-05,
+      "loss": 0.6876,
+      "step": 65500
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 4.309840008365576e-05,
+      "loss": 0.6622,
+      "step": 66000
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 4.304611523580467e-05,
+      "loss": 0.6452,
+      "step": 66500
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 4.299383038795357e-05,
+      "loss": 0.6647,
+      "step": 67000
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 4.294154554010248e-05,
+      "loss": 0.6562,
+      "step": 67500
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 4.2889260692251385e-05,
+      "loss": 0.6274,
+      "step": 68000
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 4.2836975844400296e-05,
+      "loss": 0.6503,
+      "step": 68500
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 4.27846909965492e-05,
+      "loss": 0.6427,
+      "step": 69000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 4.273240614869811e-05,
+      "loss": 0.6766,
+      "step": 69500
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 4.2680121300847016e-05,
+      "loss": 0.6578,
+      "step": 70000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 4.262783645299592e-05,
+      "loss": 0.6269,
+      "step": 70500
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 4.257555160514483e-05,
+      "loss": 0.6384,
+      "step": 71000
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.252326675729374e-05,
+      "loss": 0.6593,
+      "step": 71500
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.247098190944265e-05,
+      "loss": 0.658,
+      "step": 72000
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.241869706159155e-05,
+      "loss": 0.6884,
+      "step": 72500
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.236641221374046e-05,
+      "loss": 0.6342,
+      "step": 73000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.231412736588937e-05,
+      "loss": 0.6604,
+      "step": 73500
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.2261842518038273e-05,
+      "loss": 0.6693,
+      "step": 74000
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.2209557670187185e-05,
+      "loss": 0.6543,
+      "step": 74500
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.215727282233609e-05,
+      "loss": 0.6449,
+      "step": 75000
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.2104987974484994e-05,
+      "loss": 0.6421,
+      "step": 75500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.2052703126633905e-05,
+      "loss": 0.617,
+      "step": 76000
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.200041827878281e-05,
+      "loss": 0.6678,
+      "step": 76500
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.194813343093172e-05,
+      "loss": 0.6912,
+      "step": 77000
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.1895848583080626e-05,
+      "loss": 0.648,
+      "step": 77500
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.184356373522954e-05,
+      "loss": 0.6743,
+      "step": 78000
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.1791278887378435e-05,
+      "loss": 0.6393,
+      "step": 78500
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.1738994039527346e-05,
+      "loss": 0.6387,
+      "step": 79000
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.168670919167625e-05,
+      "loss": 0.6549,
+      "step": 79500
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.163442434382516e-05,
+      "loss": 0.6439,
+      "step": 80000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.1582139495974074e-05,
+      "loss": 0.6458,
+      "step": 80500
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.152985464812297e-05,
+      "loss": 0.663,
+      "step": 81000
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.147756980027188e-05,
+      "loss": 0.6304,
+      "step": 81500
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.142528495242079e-05,
+      "loss": 0.6471,
+      "step": 82000
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.13730001045697e-05,
+      "loss": 0.652,
+      "step": 82500
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.13207152567186e-05,
+      "loss": 0.645,
+      "step": 83000
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.1268430408867515e-05,
+      "loss": 0.6546,
+      "step": 83500
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.121614556101642e-05,
+      "loss": 0.6011,
+      "step": 84000
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.1163860713165324e-05,
+      "loss": 0.6465,
+      "step": 84500
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.1111575865314235e-05,
+      "loss": 0.6639,
+      "step": 85000
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.105929101746314e-05,
+      "loss": 0.6391,
+      "step": 85500
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.100700616961205e-05,
+      "loss": 0.6391,
+      "step": 86000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 4.0954721321760956e-05,
+      "loss": 0.6355,
+      "step": 86500
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 4.090243647390986e-05,
+      "loss": 0.6607,
+      "step": 87000
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 4.085015162605877e-05,
+      "loss": 0.6498,
+      "step": 87500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 4.0797866778207676e-05,
+      "loss": 0.6463,
+      "step": 88000
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 4.074558193035659e-05,
+      "loss": 0.6566,
+      "step": 88500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 4.069329708250549e-05,
+      "loss": 0.6586,
+      "step": 89000
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 4.06410122346544e-05,
+      "loss": 0.6191,
+      "step": 89500
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 4.05887273868033e-05,
+      "loss": 0.6107,
+      "step": 90000
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 4.053644253895221e-05,
+      "loss": 0.6505,
+      "step": 90500
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 4.0484157691101124e-05,
+      "loss": 0.6473,
+      "step": 91000
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 4.043187284325003e-05,
+      "loss": 0.6264,
+      "step": 91500
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 4.037958799539894e-05,
+      "loss": 0.6572,
+      "step": 92000
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 4.032730314754784e-05,
+      "loss": 0.6464,
+      "step": 92500
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 4.027501829969675e-05,
+      "loss": 0.6204,
+      "step": 93000
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 4.0222733451845654e-05,
+      "loss": 0.6363,
+      "step": 93500
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 4.0170448603994565e-05,
+      "loss": 0.6514,
+      "step": 94000
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.0118163756143476e-05,
+      "loss": 0.6464,
+      "step": 94500
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 4.006587890829238e-05,
+      "loss": 0.6582,
+      "step": 95000
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 4.0013594060441286e-05,
+      "loss": 0.6507,
+      "step": 95500
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.996130921259019e-05,
+      "loss": 0.5918,
+      "step": 96000
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.99090243647391e-05,
+      "loss": 0.5721,
+      "step": 96500
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.9856739516888006e-05,
+      "loss": 0.5882,
+      "step": 97000
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.980445466903692e-05,
+      "loss": 0.5759,
+      "step": 97500
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.975216982118582e-05,
+      "loss": 0.5864,
+      "step": 98000
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.969988497333473e-05,
+      "loss": 0.5741,
+      "step": 98500
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.964760012548364e-05,
+      "loss": 0.5952,
+      "step": 99000
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.959531527763254e-05,
+      "loss": 0.5823,
+      "step": 99500
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.9543030429781454e-05,
+      "loss": 0.5871,
+      "step": 100000
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.949074558193036e-05,
+      "loss": 0.5717,
+      "step": 100500
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 3.943846073407926e-05,
+      "loss": 0.6012,
+      "step": 101000
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 3.9386175886228174e-05,
+      "loss": 0.5999,
+      "step": 101500
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 3.933389103837708e-05,
+      "loss": 0.5749,
+      "step": 102000
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 3.928160619052599e-05,
+      "loss": 0.6024,
+      "step": 102500
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 3.9229321342674895e-05,
+      "loss": 0.5815,
+      "step": 103000
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 3.9177036494823806e-05,
+      "loss": 0.6153,
+      "step": 103500
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 3.9124751646972704e-05,
+      "loss": 0.5985,
+      "step": 104000
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 3.9072466799121616e-05,
+      "loss": 0.5789,
+      "step": 104500
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 3.902018195127053e-05,
+      "loss": 0.5979,
+      "step": 105000
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 3.896789710341943e-05,
+      "loss": 0.6111,
+      "step": 105500
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 3.891561225556834e-05,
+      "loss": 0.5985,
+      "step": 106000
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 3.886332740771724e-05,
+      "loss": 0.603,
+      "step": 106500
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 3.881104255986615e-05,
+      "loss": 0.6152,
+      "step": 107000
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 3.8758757712015057e-05,
+      "loss": 0.5891,
+      "step": 107500
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 3.870647286416397e-05,
+      "loss": 0.6007,
+      "step": 108000
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 3.865418801631287e-05,
+      "loss": 0.5796,
+      "step": 108500
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 3.8601903168461784e-05,
+      "loss": 0.6049,
+      "step": 109000
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 3.854961832061069e-05,
+      "loss": 0.5707,
+      "step": 109500
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 3.849733347275959e-05,
+      "loss": 0.6083,
+      "step": 110000
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 3.8445048624908504e-05,
+      "loss": 0.5984,
+      "step": 110500
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 3.839276377705741e-05,
+      "loss": 0.6061,
+      "step": 111000
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 3.834047892920632e-05,
+      "loss": 0.585,
+      "step": 111500
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 3.8288194081355225e-05,
+      "loss": 0.5908,
+      "step": 112000
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 3.823590923350413e-05,
+      "loss": 0.5628,
+      "step": 112500
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 3.818362438565304e-05,
+      "loss": 0.5913,
+      "step": 113000
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 3.8131339537801945e-05,
+      "loss": 0.6095,
+      "step": 113500
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 3.807905468995086e-05,
+      "loss": 0.5652,
+      "step": 114000
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 3.802676984209976e-05,
+      "loss": 0.5884,
+      "step": 114500
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 3.7974484994248666e-05,
+      "loss": 0.5881,
+      "step": 115000
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 3.792220014639758e-05,
+      "loss": 0.572,
+      "step": 115500
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 3.786991529854648e-05,
+      "loss": 0.613,
+      "step": 116000
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 3.781763045069539e-05,
+      "loss": 0.5828,
+      "step": 116500
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 3.77653456028443e-05,
+      "loss": 0.593,
+      "step": 117000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 3.771306075499321e-05,
+      "loss": 0.5822,
+      "step": 117500
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 3.766077590714211e-05,
+      "loss": 0.584,
+      "step": 118000
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 3.760849105929102e-05,
+      "loss": 0.6153,
+      "step": 118500
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 3.755620621143992e-05,
+      "loss": 0.576,
+      "step": 119000
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 3.7503921363588834e-05,
+      "loss": 0.6093,
+      "step": 119500
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 3.7451636515737746e-05,
+      "loss": 0.5884,
+      "step": 120000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 3.739935166788665e-05,
+      "loss": 0.6014,
+      "step": 120500
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 3.7347066820035555e-05,
+      "loss": 0.6056,
+      "step": 121000
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 3.729478197218446e-05,
+      "loss": 0.6095,
+      "step": 121500
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 3.724249712433337e-05,
+      "loss": 0.5864,
+      "step": 122000
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 3.7190212276482275e-05,
+      "loss": 0.5991,
+      "step": 122500
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 3.713792742863119e-05,
+      "loss": 0.6125,
+      "step": 123000
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 3.708564258078009e-05,
+      "loss": 0.5813,
+      "step": 123500
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 3.7033357732928996e-05,
+      "loss": 0.5866,
+      "step": 124000
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 3.698107288507791e-05,
+      "loss": 0.5924,
+      "step": 124500
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 3.692878803722681e-05,
+      "loss": 0.5912,
+      "step": 125000
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 3.687650318937572e-05,
+      "loss": 0.5514,
+      "step": 125500
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 3.682421834152463e-05,
+      "loss": 0.5862,
+      "step": 126000
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 3.677193349367353e-05,
+      "loss": 0.5608,
+      "step": 126500
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 3.6719648645822444e-05,
+      "loss": 0.5791,
+      "step": 127000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 3.666736379797135e-05,
+      "loss": 0.6113,
+      "step": 127500
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 3.661507895012026e-05,
+      "loss": 0.5795,
+      "step": 128000
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 3.6562794102269164e-05,
+      "loss": 0.6109,
+      "step": 128500
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 3.6510509254418076e-05,
+      "loss": 0.6091,
+      "step": 129000
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 3.645822440656697e-05,
+      "loss": 0.5855,
+      "step": 129500
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 3.6405939558715885e-05,
+      "loss": 0.5788,
+      "step": 130000
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 3.6353654710864796e-05,
+      "loss": 0.6089,
+      "step": 130500
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 3.63013698630137e-05,
+      "loss": 0.6061,
+      "step": 131000
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 3.624908501516261e-05,
+      "loss": 0.6076,
+      "step": 131500
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 3.6196800167311517e-05,
+      "loss": 0.592,
+      "step": 132000
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 3.614451531946042e-05,
+      "loss": 0.5624,
+      "step": 132500
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 3.6092230471609326e-05,
+      "loss": 0.592,
+      "step": 133000
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 3.603994562375824e-05,
+      "loss": 0.6006,
+      "step": 133500
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 3.598766077590714e-05,
+      "loss": 0.5696,
+      "step": 134000
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 3.593537592805605e-05,
+      "loss": 0.5638,
+      "step": 134500
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 3.588309108020496e-05,
+      "loss": 0.5895,
+      "step": 135000
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 3.583080623235386e-05,
+      "loss": 0.5862,
+      "step": 135500
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 3.5778521384502774e-05,
+      "loss": 0.5675,
+      "step": 136000
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 3.572623653665168e-05,
+      "loss": 0.5469,
+      "step": 136500
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 3.567395168880059e-05,
+      "loss": 0.5871,
+      "step": 137000
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.5621666840949494e-05,
+      "loss": 0.5943,
+      "step": 137500
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.55693819930984e-05,
+      "loss": 0.5586,
+      "step": 138000
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.551709714524731e-05,
+      "loss": 0.5978,
+      "step": 138500
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.5464812297396215e-05,
+      "loss": 0.579,
+      "step": 139000
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 3.5412527449545126e-05,
+      "loss": 0.6138,
+      "step": 139500
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 3.536024260169403e-05,
+      "loss": 0.5756,
+      "step": 140000
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 3.530795775384294e-05,
+      "loss": 0.5899,
+      "step": 140500
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 3.5255672905991846e-05,
+      "loss": 0.5765,
+      "step": 141000
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 3.520338805814075e-05,
+      "loss": 0.5767,
+      "step": 141500
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 3.515110321028966e-05,
+      "loss": 0.5916,
+      "step": 142000
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 3.509881836243857e-05,
+      "loss": 0.6079,
+      "step": 142500
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 3.504653351458748e-05,
+      "loss": 0.608,
+      "step": 143000
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 3.4994248666736376e-05,
+      "loss": 0.5729,
+      "step": 143500
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 3.494196381888529e-05,
+      "loss": 0.5501,
+      "step": 144000
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 3.488967897103419e-05,
+      "loss": 0.5655,
+      "step": 144500
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 3.4837394123183103e-05,
+      "loss": 0.5459,
+      "step": 145000
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 3.4785109275332015e-05,
+      "loss": 0.5261,
+      "step": 145500
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 3.473282442748092e-05,
+      "loss": 0.5256,
+      "step": 146000
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 3.4680539579629824e-05,
+      "loss": 0.553,
+      "step": 146500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 3.462825473177873e-05,
+      "loss": 0.5718,
+      "step": 147000
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 3.457596988392764e-05,
+      "loss": 0.5577,
+      "step": 147500
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 3.4523685036076545e-05,
+      "loss": 0.5396,
+      "step": 148000
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 3.4471400188225456e-05,
+      "loss": 0.5378,
+      "step": 148500
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 3.441911534037436e-05,
+      "loss": 0.5366,
+      "step": 149000
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 3.4366830492523265e-05,
+      "loss": 0.5619,
+      "step": 149500
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 3.4314545644672176e-05,
+      "loss": 0.5326,
+      "step": 150000
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 3.426226079682108e-05,
+      "loss": 0.5735,
+      "step": 150500
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 3.420997594896999e-05,
+      "loss": 0.5345,
+      "step": 151000
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 3.41576911011189e-05,
+      "loss": 0.5283,
+      "step": 151500
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 3.41054062532678e-05,
+      "loss": 0.5424,
+      "step": 152000
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 3.405312140541671e-05,
+      "loss": 0.5694,
+      "step": 152500
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 3.400083655756562e-05,
+      "loss": 0.5309,
+      "step": 153000
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 3.394855170971453e-05,
+      "loss": 0.5426,
+      "step": 153500
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 3.389626686186343e-05,
+      "loss": 0.5597,
+      "step": 154000
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 3.3843982014012345e-05,
+      "loss": 0.5433,
+      "step": 154500
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 3.379169716616124e-05,
+      "loss": 0.542,
+      "step": 155000
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 3.3739412318310154e-05,
+      "loss": 0.5371,
+      "step": 155500
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 3.3687127470459065e-05,
+      "loss": 0.5698,
+      "step": 156000
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 3.363484262260797e-05,
+      "loss": 0.5431,
+      "step": 156500
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 3.358255777475688e-05,
+      "loss": 0.5284,
+      "step": 157000
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 3.3530272926905786e-05,
+      "loss": 0.536,
+      "step": 157500
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 3.347798807905469e-05,
+      "loss": 0.5591,
+      "step": 158000
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 3.3425703231203595e-05,
+      "loss": 0.5197,
+      "step": 158500
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 3.3373418383352506e-05,
+      "loss": 0.5357,
+      "step": 159000
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 3.332113353550142e-05,
+      "loss": 0.5549,
+      "step": 159500
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 3.326884868765032e-05,
+      "loss": 0.556,
+      "step": 160000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 3.321656383979923e-05,
+      "loss": 0.5425,
+      "step": 160500
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 3.316427899194813e-05,
+      "loss": 0.5247,
+      "step": 161000
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 3.311199414409704e-05,
+      "loss": 0.5339,
+      "step": 161500
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 3.305970929624595e-05,
+      "loss": 0.5495,
+      "step": 162000
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 3.300742444839486e-05,
+      "loss": 0.5362,
+      "step": 162500
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 3.295513960054376e-05,
+      "loss": 0.556,
+      "step": 163000
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 3.290285475269267e-05,
+      "loss": 0.5521,
+      "step": 163500
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.285056990484158e-05,
+      "loss": 0.5553,
+      "step": 164000
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.2798285056990484e-05,
+      "loss": 0.5434,
+      "step": 164500
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.2746000209139395e-05,
+      "loss": 0.5498,
+      "step": 165000
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.26937153612883e-05,
+      "loss": 0.5435,
+      "step": 165500
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.264143051343721e-05,
+      "loss": 0.5384,
+      "step": 166000
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.2589145665586116e-05,
+      "loss": 0.5705,
+      "step": 166500
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.253686081773502e-05,
+      "loss": 0.5452,
+      "step": 167000
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.248457596988393e-05,
+      "loss": 0.5341,
+      "step": 167500
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.2432291122032836e-05,
+      "loss": 0.5492,
+      "step": 168000
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.238000627418175e-05,
+      "loss": 0.5583,
+      "step": 168500
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.2327721426330645e-05,
+      "loss": 0.5837,
+      "step": 169000
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.227543657847956e-05,
+      "loss": 0.5318,
+      "step": 169500
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.222315173062847e-05,
+      "loss": 0.5835,
+      "step": 170000
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.217086688277737e-05,
+      "loss": 0.5547,
+      "step": 170500
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.2118582034926284e-05,
+      "loss": 0.5549,
+      "step": 171000
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.206629718707519e-05,
+      "loss": 0.5426,
+      "step": 171500
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.201401233922409e-05,
+      "loss": 0.5464,
+      "step": 172000
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.1961727491373e-05,
+      "loss": 0.5396,
+      "step": 172500
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.190944264352191e-05,
+      "loss": 0.563,
+      "step": 173000
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.1857157795670814e-05,
+      "loss": 0.5453,
+      "step": 173500
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.1804872947819725e-05,
+      "loss": 0.5593,
+      "step": 174000
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.1752588099968636e-05,
+      "loss": 0.5374,
+      "step": 174500
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.1700303252117534e-05,
+      "loss": 0.549,
+      "step": 175000
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.1648018404266446e-05,
+      "loss": 0.5501,
+      "step": 175500
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.159573355641535e-05,
+      "loss": 0.5716,
+      "step": 176000
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.154344870856426e-05,
+      "loss": 0.5449,
+      "step": 176500
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.1491163860713166e-05,
+      "loss": 0.5678,
+      "step": 177000
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.143887901286207e-05,
+      "loss": 0.5384,
+      "step": 177500
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.138659416501098e-05,
+      "loss": 0.5327,
+      "step": 178000
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1334309317159887e-05,
+      "loss": 0.546,
+      "step": 178500
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.12820244693088e-05,
+      "loss": 0.5257,
+      "step": 179000
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.12297396214577e-05,
+      "loss": 0.5232,
+      "step": 179500
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.1177454773606614e-05,
+      "loss": 0.5351,
+      "step": 180000
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.112516992575552e-05,
+      "loss": 0.5379,
+      "step": 180500
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.107288507790442e-05,
+      "loss": 0.5343,
+      "step": 181000
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.1020600230053334e-05,
+      "loss": 0.5496,
+      "step": 181500
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.096831538220224e-05,
+      "loss": 0.5387,
+      "step": 182000
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.091603053435115e-05,
+      "loss": 0.5361,
+      "step": 182500
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.0863745686500055e-05,
+      "loss": 0.5321,
+      "step": 183000
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.081146083864896e-05,
+      "loss": 0.5526,
+      "step": 183500
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.0759175990797864e-05,
+      "loss": 0.5866,
+      "step": 184000
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.0706891142946775e-05,
+      "loss": 0.5566,
+      "step": 184500
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.065460629509569e-05,
+      "loss": 0.5216,
+      "step": 185000
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.060232144724459e-05,
+      "loss": 0.5304,
+      "step": 185500
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.0550036599393496e-05,
+      "loss": 0.5371,
+      "step": 186000
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.0497751751542404e-05,
+      "loss": 0.5406,
+      "step": 186500
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.0445466903691312e-05,
+      "loss": 0.5361,
+      "step": 187000
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.039318205584022e-05,
+      "loss": 0.5458,
+      "step": 187500
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.0340897207989128e-05,
+      "loss": 0.5337,
+      "step": 188000
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0288612360138036e-05,
+      "loss": 0.544,
+      "step": 188500
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0236327512286937e-05,
+      "loss": 0.5252,
+      "step": 189000
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0184042664435845e-05,
+      "loss": 0.5183,
+      "step": 189500
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0131757816584756e-05,
+      "loss": 0.5715,
+      "step": 190000
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0079472968733664e-05,
+      "loss": 0.5432,
+      "step": 190500
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 3.0027188120882572e-05,
+      "loss": 0.5498,
+      "step": 191000
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 2.997490327303148e-05,
+      "loss": 0.5321,
+      "step": 191500
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 2.992261842518038e-05,
+      "loss": 0.5073,
+      "step": 192000
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 2.987033357732929e-05,
+      "loss": 0.5032,
+      "step": 192500
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 2.9818048729478197e-05,
+      "loss": 0.4929,
+      "step": 193000
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 2.9765763881627105e-05,
+      "loss": 0.5216,
+      "step": 193500
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 2.9713479033776013e-05,
+      "loss": 0.496,
+      "step": 194000
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 2.9661194185924918e-05,
+      "loss": 0.4708,
+      "step": 194500
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 2.9608909338073826e-05,
+      "loss": 0.4902,
+      "step": 195000
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 2.9556624490222734e-05,
+      "loss": 0.4774,
+      "step": 195500
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 2.9504339642371642e-05,
+      "loss": 0.5148,
+      "step": 196000
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 2.945205479452055e-05,
+      "loss": 0.5185,
+      "step": 196500
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 2.9399769946669458e-05,
+      "loss": 0.4926,
+      "step": 197000
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 2.9347485098818362e-05,
+      "loss": 0.5151,
+      "step": 197500
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 2.929520025096727e-05,
+      "loss": 0.5046,
+      "step": 198000
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 2.9242915403116178e-05,
+      "loss": 0.514,
+      "step": 198500
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 2.9190630555265086e-05,
+      "loss": 0.51,
+      "step": 199000
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 2.9138345707413994e-05,
+      "loss": 0.5434,
+      "step": 199500
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 2.9086060859562902e-05,
+      "loss": 0.5082,
+      "step": 200000
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 2.9033776011711807e-05,
+      "loss": 0.4826,
+      "step": 200500
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 2.8981491163860715e-05,
+      "loss": 0.4887,
+      "step": 201000
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 2.8929206316009623e-05,
+      "loss": 0.5012,
+      "step": 201500
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 2.887692146815853e-05,
+      "loss": 0.4757,
+      "step": 202000
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 2.882463662030744e-05,
+      "loss": 0.5505,
+      "step": 202500
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.8772351772456347e-05,
+      "loss": 0.5012,
+      "step": 203000
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.8720066924605248e-05,
+      "loss": 0.4849,
+      "step": 203500
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.8667782076754156e-05,
+      "loss": 0.4962,
+      "step": 204000
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8615497228903064e-05,
+      "loss": 0.5078,
+      "step": 204500
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8563212381051975e-05,
+      "loss": 0.5322,
+      "step": 205000
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8510927533200883e-05,
+      "loss": 0.5212,
+      "step": 205500
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.8458642685349784e-05,
+      "loss": 0.5152,
+      "step": 206000
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.8406357837498692e-05,
+      "loss": 0.5157,
+      "step": 206500
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.83540729896476e-05,
+      "loss": 0.4864,
+      "step": 207000
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.8301788141796508e-05,
+      "loss": 0.4851,
+      "step": 207500
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.8249503293945416e-05,
+      "loss": 0.5083,
+      "step": 208000
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.8197218446094324e-05,
+      "loss": 0.5059,
+      "step": 208500
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.814493359824323e-05,
+      "loss": 0.5203,
+      "step": 209000
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.8092648750392137e-05,
+      "loss": 0.5268,
+      "step": 209500
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.8040363902541045e-05,
+      "loss": 0.5143,
+      "step": 210000
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.7988079054689953e-05,
+      "loss": 0.5081,
+      "step": 210500
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.793579420683886e-05,
+      "loss": 0.4966,
+      "step": 211000
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.788350935898777e-05,
+      "loss": 0.4854,
+      "step": 211500
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.7831224511136673e-05,
+      "loss": 0.5144,
+      "step": 212000
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.777893966328558e-05,
+      "loss": 0.4946,
+      "step": 212500
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.772665481543449e-05,
+      "loss": 0.4746,
+      "step": 213000
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.7674369967583397e-05,
+      "loss": 0.5224,
+      "step": 213500
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.7622085119732305e-05,
+      "loss": 0.5116,
+      "step": 214000
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.7569800271881206e-05,
+      "loss": 0.476,
+      "step": 214500
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.7517515424030114e-05,
+      "loss": 0.5014,
+      "step": 215000
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 2.7465230576179026e-05,
+      "loss": 0.5107,
+      "step": 215500
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 2.7412945728327933e-05,
+      "loss": 0.5157,
+      "step": 216000
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 2.736066088047684e-05,
+      "loss": 0.5458,
+      "step": 216500
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 2.730837603262575e-05,
+      "loss": 0.5031,
+      "step": 217000
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 2.725609118477465e-05,
+      "loss": 0.5098,
+      "step": 217500
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 2.720380633692356e-05,
+      "loss": 0.4895,
+      "step": 218000
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 2.7151521489072467e-05,
+      "loss": 0.5169,
+      "step": 218500
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 2.7099236641221375e-05,
+      "loss": 0.5024,
+      "step": 219000
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 2.7046951793370286e-05,
+      "loss": 0.4962,
+      "step": 219500
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 2.6994666945519194e-05,
+      "loss": 0.5244,
+      "step": 220000
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 2.6942382097668095e-05,
+      "loss": 0.5145,
+      "step": 220500
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.6890097249817003e-05,
+      "loss": 0.4924,
+      "step": 221000
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.683781240196591e-05,
+      "loss": 0.5127,
+      "step": 221500
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.678552755411482e-05,
+      "loss": 0.5052,
+      "step": 222000
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.6733242706263727e-05,
+      "loss": 0.4872,
+      "step": 222500
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.668095785841263e-05,
+      "loss": 0.5164,
+      "step": 223000
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.662867301056154e-05,
+      "loss": 0.5303,
+      "step": 223500
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.6576388162710447e-05,
+      "loss": 0.5112,
+      "step": 224000
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 2.6524103314859355e-05,
+      "loss": 0.5119,
+      "step": 224500
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 2.6471818467008263e-05,
+      "loss": 0.5028,
+      "step": 225000
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 2.641953361915717e-05,
+      "loss": 0.5166,
+      "step": 225500
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 2.6367248771306076e-05,
+      "loss": 0.5139,
+      "step": 226000
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 2.6314963923454984e-05,
+      "loss": 0.5063,
+      "step": 226500
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 2.6262679075603892e-05,
+      "loss": 0.5236,
+      "step": 227000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 2.62103942277528e-05,
+      "loss": 0.5249,
+      "step": 227500
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 2.6158109379901708e-05,
+      "loss": 0.524,
+      "step": 228000
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 2.6105824532050616e-05,
+      "loss": 0.4759,
+      "step": 228500
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 2.6053539684199517e-05,
+      "loss": 0.5111,
+      "step": 229000
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 2.6001254836348425e-05,
+      "loss": 0.5016,
+      "step": 229500
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 2.5948969988497336e-05,
+      "loss": 0.5448,
+      "step": 230000
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 2.5896685140646244e-05,
+      "loss": 0.5204,
+      "step": 230500
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 2.5844400292795152e-05,
+      "loss": 0.5324,
+      "step": 231000
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 2.5792115444944053e-05,
+      "loss": 0.5069,
+      "step": 231500
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 2.573983059709296e-05,
+      "loss": 0.5221,
+      "step": 232000
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 2.568754574924187e-05,
+      "loss": 0.506,
+      "step": 232500
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 2.5635260901390777e-05,
+      "loss": 0.5418,
+      "step": 233000
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.5582976053539685e-05,
+      "loss": 0.4877,
+      "step": 233500
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5530691205688593e-05,
+      "loss": 0.4889,
+      "step": 234000
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.5478406357837498e-05,
+      "loss": 0.5019,
+      "step": 234500
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 2.5426121509986406e-05,
+      "loss": 0.5124,
+      "step": 235000
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.5373836662135314e-05,
+      "loss": 0.5066,
+      "step": 235500
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.5321551814284222e-05,
+      "loss": 0.5014,
+      "step": 236000
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.526926696643313e-05,
+      "loss": 0.5057,
+      "step": 236500
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.5216982118582038e-05,
+      "loss": 0.5125,
+      "step": 237000
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.5164697270730942e-05,
+      "loss": 0.5012,
+      "step": 237500
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 2.511241242287985e-05,
+      "loss": 0.4988,
+      "step": 238000
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.5060127575028758e-05,
+      "loss": 0.4761,
+      "step": 238500
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 2.5007842727177666e-05,
+      "loss": 0.518,
+      "step": 239000
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 2.495555787932657e-05,
+      "loss": 0.49,
+      "step": 239500
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 2.490327303147548e-05,
+      "loss": 0.4793,
+      "step": 240000
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 2.4850988183624387e-05,
+      "loss": 0.4624,
+      "step": 240500
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 2.4798703335773295e-05,
+      "loss": 0.4706,
+      "step": 241000
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 2.4746418487922203e-05,
+      "loss": 0.4716,
+      "step": 241500
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 2.469413364007111e-05,
+      "loss": 0.4775,
+      "step": 242000
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 2.4641848792220015e-05,
+      "loss": 0.4865,
+      "step": 242500
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 2.4589563944368923e-05,
+      "loss": 0.4824,
+      "step": 243000
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 2.4537279096517828e-05,
+      "loss": 0.4781,
+      "step": 243500
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 2.4484994248666736e-05,
+      "loss": 0.445,
+      "step": 244000
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 2.4432709400815647e-05,
+      "loss": 0.4782,
+      "step": 244500
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 2.438042455296455e-05,
+      "loss": 0.4768,
+      "step": 245000
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 2.432813970511346e-05,
+      "loss": 0.4527,
+      "step": 245500
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 2.4275854857262368e-05,
+      "loss": 0.4633,
+      "step": 246000
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 2.4223570009411272e-05,
+      "loss": 0.4677,
+      "step": 246500
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 2.417128516156018e-05,
+      "loss": 0.4615,
+      "step": 247000
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 2.4119000313709088e-05,
+      "loss": 0.4708,
+      "step": 247500
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 2.4066715465857996e-05,
+      "loss": 0.4835,
+      "step": 248000
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 2.4014430618006904e-05,
+      "loss": 0.4554,
+      "step": 248500
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 2.3962145770155812e-05,
+      "loss": 0.4758,
+      "step": 249000
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 2.3909860922304717e-05,
+      "loss": 0.5039,
+      "step": 249500
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 2.3857576074453625e-05,
+      "loss": 0.4834,
+      "step": 250000
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 2.3805291226602533e-05,
+      "loss": 0.4782,
+      "step": 250500
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 2.3753006378751437e-05,
+      "loss": 0.487,
+      "step": 251000
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 2.3700721530900345e-05,
+      "loss": 0.4716,
+      "step": 251500
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 2.3648436683049253e-05,
+      "loss": 0.4902,
+      "step": 252000
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 2.359615183519816e-05,
+      "loss": 0.4802,
+      "step": 252500
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 2.354386698734707e-05,
+      "loss": 0.4816,
+      "step": 253000
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 2.3491582139495974e-05,
+      "loss": 0.4725,
+      "step": 253500
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 2.343929729164488e-05,
+      "loss": 0.4646,
+      "step": 254000
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 2.338701244379379e-05,
+      "loss": 0.4651,
+      "step": 254500
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 2.3334727595942697e-05,
+      "loss": 0.4932,
+      "step": 255000
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 2.3282442748091605e-05,
+      "loss": 0.466,
+      "step": 255500
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 2.3230157900240513e-05,
+      "loss": 0.4887,
+      "step": 256000
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 2.3177873052389418e-05,
+      "loss": 0.4856,
+      "step": 256500
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 2.3125588204538326e-05,
+      "loss": 0.5018,
+      "step": 257000
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 2.3073303356687234e-05,
+      "loss": 0.4869,
+      "step": 257500
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 2.302101850883614e-05,
+      "loss": 0.4699,
+      "step": 258000
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 2.2968733660985047e-05,
+      "loss": 0.484,
+      "step": 258500
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 2.2916448813133954e-05,
+      "loss": 0.4564,
+      "step": 259000
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 2.2864163965282862e-05,
+      "loss": 0.5038,
+      "step": 259500
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 2.281187911743177e-05,
+      "loss": 0.4859,
+      "step": 260000
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 2.275959426958068e-05,
+      "loss": 0.4783,
+      "step": 260500
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 2.2707309421729583e-05,
+      "loss": 0.4773,
+      "step": 261000
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 2.265502457387849e-05,
+      "loss": 0.4729,
+      "step": 261500
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 2.2602739726027396e-05,
+      "loss": 0.4682,
+      "step": 262000
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 2.2550454878176307e-05,
+      "loss": 0.4682,
+      "step": 262500
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 2.2498170030325215e-05,
+      "loss": 0.4819,
+      "step": 263000
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 2.244588518247412e-05,
+      "loss": 0.4896,
+      "step": 263500
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 2.2393600334623027e-05,
+      "loss": 0.4475,
+      "step": 264000
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 2.2341315486771935e-05,
+      "loss": 0.464,
+      "step": 264500
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 2.228903063892084e-05,
+      "loss": 0.4981,
+      "step": 265000
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 2.2236745791069748e-05,
+      "loss": 0.4741,
+      "step": 265500
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 2.2184460943218656e-05,
+      "loss": 0.4592,
+      "step": 266000
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 2.2132176095367564e-05,
+      "loss": 0.4636,
+      "step": 266500
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 2.2079891247516472e-05,
+      "loss": 0.4954,
+      "step": 267000
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 2.202760639966538e-05,
+      "loss": 0.4981,
+      "step": 267500
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 2.1975321551814284e-05,
+      "loss": 0.4602,
+      "step": 268000
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 2.1923036703963192e-05,
+      "loss": 0.4694,
+      "step": 268500
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 2.18707518561121e-05,
+      "loss": 0.4881,
+      "step": 269000
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 2.1818467008261005e-05,
+      "loss": 0.4973,
+      "step": 269500
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 2.1766182160409916e-05,
+      "loss": 0.4687,
+      "step": 270000
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 2.171389731255882e-05,
+      "loss": 0.4862,
+      "step": 270500
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 2.166161246470773e-05,
+      "loss": 0.4734,
+      "step": 271000
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 2.1609327616856637e-05,
+      "loss": 0.5074,
+      "step": 271500
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 2.155704276900554e-05,
+      "loss": 0.4922,
+      "step": 272000
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 2.150475792115445e-05,
+      "loss": 0.498,
+      "step": 272500
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 2.1452473073303357e-05,
+      "loss": 0.4725,
+      "step": 273000
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 2.1400188225452265e-05,
+      "loss": 0.429,
+      "step": 273500
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 2.1347903377601173e-05,
+      "loss": 0.4755,
+      "step": 274000
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 2.129561852975008e-05,
+      "loss": 0.4757,
+      "step": 274500
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 2.1243333681898986e-05,
+      "loss": 0.4728,
+      "step": 275000
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 2.1191048834047894e-05,
+      "loss": 0.491,
+      "step": 275500
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 2.1138763986196802e-05,
+      "loss": 0.4939,
+      "step": 276000
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 2.1086479138345706e-05,
+      "loss": 0.4773,
+      "step": 276500
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 2.1034194290494618e-05,
+      "loss": 0.4812,
+      "step": 277000
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 2.0981909442643526e-05,
+      "loss": 0.4796,
+      "step": 277500
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 2.092962459479243e-05,
+      "loss": 0.48,
+      "step": 278000
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 2.0877339746941338e-05,
+      "loss": 0.4863,
+      "step": 278500
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 2.0825054899090243e-05,
+      "loss": 0.4961,
+      "step": 279000
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 2.077277005123915e-05,
+      "loss": 0.4645,
+      "step": 279500
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 2.072048520338806e-05,
+      "loss": 0.4926,
+      "step": 280000
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 2.0668200355536967e-05,
+      "loss": 0.4935,
+      "step": 280500
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 2.0615915507685875e-05,
+      "loss": 0.5012,
+      "step": 281000
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 2.0563630659834783e-05,
+      "loss": 0.4922,
+      "step": 281500
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 2.0511345811983687e-05,
+      "loss": 0.4614,
+      "step": 282000
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 2.0459060964132595e-05,
+      "loss": 0.4737,
+      "step": 282500
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 2.0406776116281503e-05,
+      "loss": 0.482,
+      "step": 283000
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 2.0354491268430408e-05,
+      "loss": 0.4773,
+      "step": 283500
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 2.0302206420579316e-05,
+      "loss": 0.5035,
+      "step": 284000
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 2.0249921572728227e-05,
+      "loss": 0.4404,
+      "step": 284500
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 2.019763672487713e-05,
+      "loss": 0.4864,
+      "step": 285000
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 2.014535187702604e-05,
+      "loss": 0.4959,
+      "step": 285500
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 2.0093067029174948e-05,
+      "loss": 0.4792,
+      "step": 286000
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 2.0040782181323852e-05,
+      "loss": 0.4821,
+      "step": 286500
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 1.998849733347276e-05,
+      "loss": 0.4649,
+      "step": 287000
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 1.9936212485621668e-05,
+      "loss": 0.4552,
+      "step": 287500
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 1.9883927637770576e-05,
+      "loss": 0.4687,
+      "step": 288000
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 1.9831642789919484e-05,
+      "loss": 0.4455,
+      "step": 288500
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 1.977935794206839e-05,
+      "loss": 0.4571,
+      "step": 289000
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 1.9727073094217297e-05,
+      "loss": 0.4498,
+      "step": 289500
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 1.9674788246366205e-05,
+      "loss": 0.4616,
+      "step": 290000
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 1.962250339851511e-05,
+      "loss": 0.4694,
+      "step": 290500
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 1.9570218550664017e-05,
+      "loss": 0.4594,
+      "step": 291000
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 1.9517933702812925e-05,
+      "loss": 0.4337,
+      "step": 291500
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 1.9465648854961833e-05,
+      "loss": 0.4548,
+      "step": 292000
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 1.941336400711074e-05,
+      "loss": 0.4498,
+      "step": 292500
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 1.936107915925965e-05,
+      "loss": 0.4649,
+      "step": 293000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 1.9308794311408554e-05,
+      "loss": 0.4529,
+      "step": 293500
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 1.925650946355746e-05,
+      "loss": 0.4513,
+      "step": 294000
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 1.920422461570637e-05,
+      "loss": 0.437,
+      "step": 294500
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 1.9151939767855277e-05,
+      "loss": 0.4541,
+      "step": 295000
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 1.9099654920004185e-05,
+      "loss": 0.464,
+      "step": 295500
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 1.9047370072153093e-05,
+      "loss": 0.4681,
+      "step": 296000
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 1.8995085224301998e-05,
+      "loss": 0.4354,
+      "step": 296500
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 1.8942800376450906e-05,
+      "loss": 0.4457,
+      "step": 297000
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 1.889051552859981e-05,
+      "loss": 0.4584,
+      "step": 297500
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 1.883823068074872e-05,
+      "loss": 0.4815,
+      "step": 298000
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 1.8785945832897626e-05,
+      "loss": 0.4383,
+      "step": 298500
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 1.8733660985046534e-05,
+      "loss": 0.446,
+      "step": 299000
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 1.8681376137195442e-05,
+      "loss": 0.4733,
+      "step": 299500
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 1.862909128934435e-05,
+      "loss": 0.449,
+      "step": 300000
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 1.8576806441493255e-05,
+      "loss": 0.4419,
+      "step": 300500
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 1.8524521593642163e-05,
+      "loss": 0.4374,
+      "step": 301000
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 1.847223674579107e-05,
+      "loss": 0.4565,
+      "step": 301500
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 1.8419951897939975e-05,
+      "loss": 0.4434,
+      "step": 302000
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 1.8367667050088887e-05,
+      "loss": 0.4394,
+      "step": 302500
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 1.8315382202237795e-05,
+      "loss": 0.4649,
+      "step": 303000
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 1.82630973543867e-05,
+      "loss": 0.4655,
+      "step": 303500
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 1.8210812506535607e-05,
+      "loss": 0.4712,
+      "step": 304000
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 1.8158527658684515e-05,
+      "loss": 0.4672,
+      "step": 304500
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 1.810624281083342e-05,
+      "loss": 0.4626,
+      "step": 305000
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 1.8053957962982328e-05,
+      "loss": 0.4649,
+      "step": 305500
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 1.8001673115131236e-05,
+      "loss": 0.4477,
+      "step": 306000
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 1.7949388267280144e-05,
+      "loss": 0.45,
+      "step": 306500
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 1.7897103419429052e-05,
+      "loss": 0.4577,
+      "step": 307000
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 1.7844818571577956e-05,
+      "loss": 0.4795,
+      "step": 307500
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 1.7792533723726864e-05,
+      "loss": 0.4645,
+      "step": 308000
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 1.7740248875875772e-05,
+      "loss": 0.4423,
+      "step": 308500
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 1.7687964028024677e-05,
+      "loss": 0.4497,
+      "step": 309000
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 1.7635679180173588e-05,
+      "loss": 0.4769,
+      "step": 309500
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 1.7583394332322496e-05,
+      "loss": 0.4639,
+      "step": 310000
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 1.75311094844714e-05,
+      "loss": 0.4703,
+      "step": 310500
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 1.747882463662031e-05,
+      "loss": 0.4541,
+      "step": 311000
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 1.7426539788769217e-05,
+      "loss": 0.454,
+      "step": 311500
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 1.737425494091812e-05,
+      "loss": 0.4483,
+      "step": 312000
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 1.732197009306703e-05,
+      "loss": 0.4234,
+      "step": 312500
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 1.7269685245215937e-05,
+      "loss": 0.4451,
+      "step": 313000
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 1.7217400397364845e-05,
+      "loss": 0.4651,
+      "step": 313500
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 1.7165115549513753e-05,
+      "loss": 0.4331,
+      "step": 314000
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 1.7112830701662658e-05,
+      "loss": 0.4765,
+      "step": 314500
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 1.7060545853811566e-05,
+      "loss": 0.4552,
+      "step": 315000
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 1.7008261005960474e-05,
+      "loss": 0.4791,
+      "step": 315500
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 1.6955976158109378e-05,
+      "loss": 0.4552,
+      "step": 316000
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 1.6903691310258286e-05,
+      "loss": 0.455,
+      "step": 316500
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 1.6851406462407198e-05,
+      "loss": 0.4444,
+      "step": 317000
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 1.6799121614556102e-05,
+      "loss": 0.4469,
+      "step": 317500
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 1.674683676670501e-05,
+      "loss": 0.4334,
+      "step": 318000
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 1.6694551918853918e-05,
+      "loss": 0.4568,
+      "step": 318500
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 1.6642267071002823e-05,
+      "loss": 0.4446,
+      "step": 319000
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 1.658998222315173e-05,
+      "loss": 0.4535,
+      "step": 319500
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 1.653769737530064e-05,
+      "loss": 0.4797,
+      "step": 320000
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 1.6485412527449547e-05,
+      "loss": 0.4491,
+      "step": 320500
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 1.6433127679598455e-05,
+      "loss": 0.4702,
+      "step": 321000
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 1.6380842831747363e-05,
+      "loss": 0.4596,
+      "step": 321500
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 1.6328557983896267e-05,
+      "loss": 0.4359,
+      "step": 322000
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 1.6276273136045175e-05,
+      "loss": 0.4612,
+      "step": 322500
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 1.622398828819408e-05,
+      "loss": 0.4545,
+      "step": 323000
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 1.6171703440342988e-05,
+      "loss": 0.4274,
+      "step": 323500
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 1.6119418592491896e-05,
+      "loss": 0.4545,
+      "step": 324000
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 1.6067133744640804e-05,
+      "loss": 0.4704,
+      "step": 324500
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 1.601484889678971e-05,
+      "loss": 0.4611,
+      "step": 325000
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 1.596256404893862e-05,
+      "loss": 0.4661,
+      "step": 325500
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 1.5910279201087524e-05,
+      "loss": 0.4599,
+      "step": 326000
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 1.5857994353236432e-05,
+      "loss": 0.4543,
+      "step": 326500
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 1.580570950538534e-05,
+      "loss": 0.4599,
+      "step": 327000
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 1.5753424657534248e-05,
+      "loss": 0.4739,
+      "step": 327500
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 1.5701139809683156e-05,
+      "loss": 0.4799,
+      "step": 328000
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 1.5648854961832064e-05,
+      "loss": 0.4452,
+      "step": 328500
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 1.559657011398097e-05,
+      "loss": 0.4733,
+      "step": 329000
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 1.5544285266129877e-05,
+      "loss": 0.4738,
+      "step": 329500
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 1.5492000418278784e-05,
+      "loss": 0.4626,
+      "step": 330000
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 1.543971557042769e-05,
+      "loss": 0.4565,
+      "step": 330500
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 1.5387430722576597e-05,
+      "loss": 0.443,
+      "step": 331000
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 1.5335145874725505e-05,
+      "loss": 0.4594,
+      "step": 331500
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 1.5282861026874413e-05,
+      "loss": 0.4415,
+      "step": 332000
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 1.5230576179023321e-05,
+      "loss": 0.4454,
+      "step": 332500
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 1.5178291331172226e-05,
+      "loss": 0.4572,
+      "step": 333000
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 1.5126006483321133e-05,
+      "loss": 0.4609,
+      "step": 333500
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 1.5073721635470041e-05,
+      "loss": 0.4617,
+      "step": 334000
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 1.5021436787618948e-05,
+      "loss": 0.4703,
+      "step": 334500
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 1.4969151939767856e-05,
+      "loss": 0.4387,
+      "step": 335000
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 1.4916867091916764e-05,
+      "loss": 0.4427,
+      "step": 335500
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 1.486458224406567e-05,
+      "loss": 0.4316,
+      "step": 336000
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 1.4812297396214578e-05,
+      "loss": 0.4496,
+      "step": 336500
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 1.4760012548363486e-05,
+      "loss": 0.4646,
+      "step": 337000
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 1.4707727700512392e-05,
+      "loss": 0.4382,
+      "step": 337500
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 1.46554428526613e-05,
+      "loss": 0.4402,
+      "step": 338000
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 1.4603158004810208e-05,
+      "loss": 0.4444,
+      "step": 338500
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 1.4550873156959113e-05,
+      "loss": 0.4309,
+      "step": 339000
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 1.4498588309108022e-05,
+      "loss": 0.4195,
+      "step": 339500
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 1.444630346125693e-05,
+      "loss": 0.4596,
+      "step": 340000
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 1.4394018613405835e-05,
+      "loss": 0.4394,
+      "step": 340500
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 1.4341733765554743e-05,
+      "loss": 0.4309,
+      "step": 341000
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 1.4289448917703649e-05,
+      "loss": 0.4517,
+      "step": 341500
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 1.4237164069852557e-05,
+      "loss": 0.4504,
+      "step": 342000
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 1.4184879222001465e-05,
+      "loss": 0.4289,
+      "step": 342500
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 1.4132594374150371e-05,
+      "loss": 0.4268,
+      "step": 343000
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 1.408030952629928e-05,
+      "loss": 0.4317,
+      "step": 343500
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 1.4028024678448187e-05,
+      "loss": 0.4535,
+      "step": 344000
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 1.3975739830597092e-05,
+      "loss": 0.4405,
+      "step": 344500
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 1.3923454982746002e-05,
+      "loss": 0.4287,
+      "step": 345000
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 1.387117013489491e-05,
+      "loss": 0.4418,
+      "step": 345500
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 1.3818885287043814e-05,
+      "loss": 0.4316,
+      "step": 346000
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 1.3766600439192722e-05,
+      "loss": 0.4319,
+      "step": 346500
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 1.3714315591341632e-05,
+      "loss": 0.4426,
+      "step": 347000
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 1.3662030743490536e-05,
+      "loss": 0.4215,
+      "step": 347500
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 1.3609745895639444e-05,
+      "loss": 0.4526,
+      "step": 348000
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 1.3557461047788352e-05,
+      "loss": 0.4335,
+      "step": 348500
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 1.3505176199937259e-05,
+      "loss": 0.4294,
+      "step": 349000
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 1.3452891352086166e-05,
+      "loss": 0.4141,
+      "step": 349500
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 1.3400606504235073e-05,
+      "loss": 0.4348,
+      "step": 350000
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 1.334832165638398e-05,
+      "loss": 0.4515,
+      "step": 350500
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 1.3296036808532889e-05,
+      "loss": 0.4326,
+      "step": 351000
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 1.3243751960681793e-05,
+      "loss": 0.4244,
+      "step": 351500
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 1.3191467112830703e-05,
+      "loss": 0.4621,
+      "step": 352000
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 1.3139182264979611e-05,
+      "loss": 0.4211,
+      "step": 352500
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 1.3086897417128516e-05,
+      "loss": 0.4295,
+      "step": 353000
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 1.3034612569277423e-05,
+      "loss": 0.4276,
+      "step": 353500
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 1.2982327721426331e-05,
+      "loss": 0.4339,
+      "step": 354000
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 1.2930042873575238e-05,
+      "loss": 0.4433,
+      "step": 354500
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 1.2877758025724146e-05,
+      "loss": 0.4503,
+      "step": 355000
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 1.2825473177873054e-05,
+      "loss": 0.4386,
+      "step": 355500
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 1.277318833002196e-05,
+      "loss": 0.4357,
+      "step": 356000
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 1.2720903482170868e-05,
+      "loss": 0.4371,
+      "step": 356500
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 1.2668618634319776e-05,
+      "loss": 0.4332,
+      "step": 357000
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 1.2616333786468682e-05,
+      "loss": 0.425,
+      "step": 357500
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 1.256404893861759e-05,
+      "loss": 0.4317,
+      "step": 358000
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 1.2511764090766495e-05,
+      "loss": 0.428,
+      "step": 358500
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 1.2459479242915403e-05,
+      "loss": 0.4694,
+      "step": 359000
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 1.2407194395064312e-05,
+      "loss": 0.412,
+      "step": 359500
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 1.2354909547213219e-05,
+      "loss": 0.4387,
+      "step": 360000
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 1.2302624699362125e-05,
+      "loss": 0.4412,
+      "step": 360500
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 1.2250339851511033e-05,
+      "loss": 0.4401,
+      "step": 361000
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 1.219805500365994e-05,
+      "loss": 0.4325,
+      "step": 361500
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 1.2145770155808847e-05,
+      "loss": 0.4294,
+      "step": 362000
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 1.2093485307957753e-05,
+      "loss": 0.4468,
+      "step": 362500
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 1.2041200460106663e-05,
+      "loss": 0.4422,
+      "step": 363000
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 1.198891561225557e-05,
+      "loss": 0.4261,
+      "step": 363500
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 1.1936630764404476e-05,
+      "loss": 0.4385,
+      "step": 364000
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 1.1884345916553384e-05,
+      "loss": 0.4453,
+      "step": 364500
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 1.1832061068702292e-05,
+      "loss": 0.4743,
+      "step": 365000
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 1.1779776220851198e-05,
+      "loss": 0.4223,
+      "step": 365500
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 1.1727491373000104e-05,
+      "loss": 0.4491,
+      "step": 366000
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 1.1675206525149012e-05,
+      "loss": 0.4185,
+      "step": 366500
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 1.162292167729792e-05,
+      "loss": 0.4402,
+      "step": 367000
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 1.1570636829446826e-05,
+      "loss": 0.4612,
+      "step": 367500
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 1.1518351981595734e-05,
+      "loss": 0.4621,
+      "step": 368000
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 1.1466067133744642e-05,
+      "loss": 0.4304,
+      "step": 368500
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 1.1413782285893549e-05,
+      "loss": 0.4245,
+      "step": 369000
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 1.1361497438042455e-05,
+      "loss": 0.4259,
+      "step": 369500
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 1.1309212590191363e-05,
+      "loss": 0.428,
+      "step": 370000
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 1.125692774234027e-05,
+      "loss": 0.4337,
+      "step": 370500
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 1.1204642894489177e-05,
+      "loss": 0.4452,
+      "step": 371000
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 1.1152358046638085e-05,
+      "loss": 0.4485,
+      "step": 371500
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 1.1100073198786993e-05,
+      "loss": 0.4324,
+      "step": 372000
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 1.10477883509359e-05,
+      "loss": 0.43,
+      "step": 372500
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 1.0995503503084807e-05,
+      "loss": 0.4408,
+      "step": 373000
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 1.0943218655233713e-05,
+      "loss": 0.4352,
+      "step": 373500
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 1.0890933807382621e-05,
+      "loss": 0.4282,
+      "step": 374000
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 1.0838648959531528e-05,
+      "loss": 0.4512,
+      "step": 374500
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 1.0786364111680436e-05,
+      "loss": 0.4499,
+      "step": 375000
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 1.0734079263829342e-05,
+      "loss": 0.4457,
+      "step": 375500
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 1.068179441597825e-05,
+      "loss": 0.4276,
+      "step": 376000
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 1.0629509568127158e-05,
+      "loss": 0.4353,
+      "step": 376500
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 1.0577224720276064e-05,
+      "loss": 0.4521,
+      "step": 377000
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 1.0524939872424972e-05,
+      "loss": 0.4409,
+      "step": 377500
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 1.0472655024573878e-05,
+      "loss": 0.4217,
+      "step": 378000
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 1.0420370176722786e-05,
+      "loss": 0.453,
+      "step": 378500
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 1.0368085328871693e-05,
+      "loss": 0.4401,
+      "step": 379000
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 1.03158004810206e-05,
+      "loss": 0.4541,
+      "step": 379500
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 1.0263515633169509e-05,
+      "loss": 0.432,
+      "step": 380000
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 1.0211230785318415e-05,
+      "loss": 0.4453,
+      "step": 380500
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 1.0158945937467323e-05,
+      "loss": 0.4389,
+      "step": 381000
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 1.010666108961623e-05,
+      "loss": 0.4453,
+      "step": 381500
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 1.0054376241765137e-05,
+      "loss": 0.4338,
+      "step": 382000
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 1.0002091393914043e-05,
+      "loss": 0.4269,
+      "step": 382500
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 9.949806546062951e-06,
+      "loss": 0.4261,
+      "step": 383000
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 9.89752169821186e-06,
+      "loss": 0.4249,
+      "step": 383500
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 9.845236850360766e-06,
+      "loss": 0.4067,
+      "step": 384000
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 9.792952002509674e-06,
+      "loss": 0.4065,
+      "step": 384500
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 9.740667154658582e-06,
+      "loss": 0.4301,
+      "step": 385000
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 9.688382306807488e-06,
+      "loss": 0.4121,
+      "step": 385500
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 9.636097458956394e-06,
+      "loss": 0.4413,
+      "step": 386000
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 9.583812611105302e-06,
+      "loss": 0.4113,
+      "step": 386500
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 9.53152776325421e-06,
+      "loss": 0.4107,
+      "step": 387000
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 9.479242915403116e-06,
+      "loss": 0.4356,
+      "step": 387500
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 9.426958067552023e-06,
+      "loss": 0.4213,
+      "step": 388000
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 9.374673219700932e-06,
+      "loss": 0.4323,
+      "step": 388500
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 9.322388371849838e-06,
+      "loss": 0.4215,
+      "step": 389000
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 9.270103523998745e-06,
+      "loss": 0.4348,
+      "step": 389500
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 9.217818676147653e-06,
+      "loss": 0.4179,
+      "step": 390000
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 9.16553382829656e-06,
+      "loss": 0.4253,
+      "step": 390500
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 9.113248980445467e-06,
+      "loss": 0.4407,
+      "step": 391000
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 9.060964132594373e-06,
+      "loss": 0.4426,
+      "step": 391500
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 9.008679284743283e-06,
+      "loss": 0.4223,
+      "step": 392000
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 8.95639443689219e-06,
+      "loss": 0.399,
+      "step": 392500
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 8.904109589041095e-06,
+      "loss": 0.4347,
+      "step": 393000
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 8.851824741190003e-06,
+      "loss": 0.418,
+      "step": 393500
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 8.799539893338911e-06,
+      "loss": 0.4381,
+      "step": 394000
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 8.747255045487818e-06,
+      "loss": 0.4165,
+      "step": 394500
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 8.694970197636726e-06,
+      "loss": 0.4168,
+      "step": 395000
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 8.642685349785634e-06,
+      "loss": 0.4169,
+      "step": 395500
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 8.59040050193454e-06,
+      "loss": 0.3928,
+      "step": 396000
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 8.538115654083446e-06,
+      "loss": 0.4138,
+      "step": 396500
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 8.485830806232354e-06,
+      "loss": 0.4091,
+      "step": 397000
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 8.433545958381262e-06,
+      "loss": 0.4141,
+      "step": 397500
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 8.381261110530168e-06,
+      "loss": 0.4111,
+      "step": 398000
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 8.328976262679076e-06,
+      "loss": 0.4271,
+      "step": 398500
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 8.276691414827983e-06,
+      "loss": 0.4164,
+      "step": 399000
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 8.22440656697689e-06,
+      "loss": 0.4158,
+      "step": 399500
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 8.172121719125797e-06,
+      "loss": 0.4256,
+      "step": 400000
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 8.119836871274705e-06,
+      "loss": 0.4206,
+      "step": 400500
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 8.067552023423613e-06,
+      "loss": 0.4123,
+      "step": 401000
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 8.015267175572519e-06,
+      "loss": 0.4244,
+      "step": 401500
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 7.962982327721427e-06,
+      "loss": 0.42,
+      "step": 402000
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 7.910697479870333e-06,
+      "loss": 0.4067,
+      "step": 402500
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 7.858412632019241e-06,
+      "loss": 0.4307,
+      "step": 403000
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 7.80612778416815e-06,
+      "loss": 0.4354,
+      "step": 403500
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 7.753842936317056e-06,
+      "loss": 0.4007,
+      "step": 404000
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 7.701558088465964e-06,
+      "loss": 0.4185,
+      "step": 404500
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 7.64927324061487e-06,
+      "loss": 0.4448,
+      "step": 405000
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 7.596988392763778e-06,
+      "loss": 0.4314,
+      "step": 405500
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 7.544703544912685e-06,
+      "loss": 0.4176,
+      "step": 406000
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 7.492418697061591e-06,
+      "loss": 0.4242,
+      "step": 406500
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 7.4401338492105e-06,
+      "loss": 0.4259,
+      "step": 407000
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 7.387849001359406e-06,
+      "loss": 0.3979,
+      "step": 407500
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 7.335564153508313e-06,
+      "loss": 0.4005,
+      "step": 408000
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 7.283279305657221e-06,
+      "loss": 0.4435,
+      "step": 408500
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 7.2309944578061285e-06,
+      "loss": 0.4414,
+      "step": 409000
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 7.178709609955036e-06,
+      "loss": 0.4335,
+      "step": 409500
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 7.126424762103942e-06,
+      "loss": 0.4032,
+      "step": 410000
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 7.074139914252851e-06,
+      "loss": 0.4205,
+      "step": 410500
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 7.021855066401757e-06,
+      "loss": 0.4307,
+      "step": 411000
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 6.969570218550664e-06,
+      "loss": 0.4257,
+      "step": 411500
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 6.917285370699572e-06,
+      "loss": 0.3947,
+      "step": 412000
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 6.865000522848479e-06,
+      "loss": 0.4222,
+      "step": 412500
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 6.8127156749973854e-06,
+      "loss": 0.4179,
+      "step": 413000
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 6.7604308271462926e-06,
+      "loss": 0.4095,
+      "step": 413500
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 6.7081459792952005e-06,
+      "loss": 0.4268,
+      "step": 414000
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 6.655861131444108e-06,
+      "loss": 0.4201,
+      "step": 414500
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 6.603576283593015e-06,
+      "loss": 0.4435,
+      "step": 415000
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 6.551291435741923e-06,
+      "loss": 0.4201,
+      "step": 415500
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 6.49900658789083e-06,
+      "loss": 0.4152,
+      "step": 416000
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 6.446721740039736e-06,
+      "loss": 0.4389,
+      "step": 416500
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 6.394436892188645e-06,
+      "loss": 0.4174,
+      "step": 417000
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 6.342152044337551e-06,
+      "loss": 0.4222,
+      "step": 417500
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 6.289867196486458e-06,
+      "loss": 0.4301,
+      "step": 418000
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 6.2375823486353655e-06,
+      "loss": 0.4279,
+      "step": 418500
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 6.185297500784273e-06,
+      "loss": 0.4346,
+      "step": 419000
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 6.1330126529331806e-06,
+      "loss": 0.4296,
+      "step": 419500
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 6.080727805082088e-06,
+      "loss": 0.4194,
+      "step": 420000
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 6.028442957230995e-06,
+      "loss": 0.412,
+      "step": 420500
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 5.976158109379902e-06,
+      "loss": 0.4128,
+      "step": 421000
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 5.923873261528809e-06,
+      "loss": 0.4182,
+      "step": 421500
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 5.871588413677716e-06,
+      "loss": 0.4338,
+      "step": 422000
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 5.819303565826624e-06,
+      "loss": 0.4509,
+      "step": 422500
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 5.7670187179755304e-06,
+      "loss": 0.4219,
+      "step": 423000
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 5.714733870124438e-06,
+      "loss": 0.4517,
+      "step": 423500
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 5.6624490222733455e-06,
+      "loss": 0.4208,
+      "step": 424000
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 5.610164174422253e-06,
+      "loss": 0.4269,
+      "step": 424500
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 5.557879326571161e-06,
+      "loss": 0.4002,
+      "step": 425000
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 5.505594478720067e-06,
+      "loss": 0.4041,
+      "step": 425500
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 5.453309630868975e-06,
+      "loss": 0.4152,
+      "step": 426000
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 5.401024783017881e-06,
+      "loss": 0.4241,
+      "step": 426500
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.348739935166789e-06,
+      "loss": 0.4164,
+      "step": 427000
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.296455087315696e-06,
+      "loss": 0.4242,
+      "step": 427500
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.244170239464603e-06,
+      "loss": 0.4456,
+      "step": 428000
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.1918853916135105e-06,
+      "loss": 0.4229,
+      "step": 428500
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.139600543762418e-06,
+      "loss": 0.4097,
+      "step": 429000
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.0873156959113256e-06,
+      "loss": 0.4169,
+      "step": 429500
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.035030848060232e-06,
+      "loss": 0.4466,
+      "step": 430000
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 4.98274600020914e-06,
+      "loss": 0.4345,
+      "step": 430500
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 4.930461152358047e-06,
+      "loss": 0.4037,
+      "step": 431000
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.878176304506954e-06,
+      "loss": 0.4221,
+      "step": 431500
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.825891456655861e-06,
+      "loss": 0.3757,
+      "step": 432000
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.773606608804768e-06,
+      "loss": 0.4079,
+      "step": 432500
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.721321760953676e-06,
+      "loss": 0.4045,
+      "step": 433000
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.669036913102583e-06,
+      "loss": 0.4178,
+      "step": 433500
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.6167520652514905e-06,
+      "loss": 0.404,
+      "step": 434000
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.564467217400398e-06,
+      "loss": 0.4363,
+      "step": 434500
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.512182369549305e-06,
+      "loss": 0.4086,
+      "step": 435000
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.459897521698212e-06,
+      "loss": 0.4141,
+      "step": 435500
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.40761267384712e-06,
+      "loss": 0.4174,
+      "step": 436000
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 4.355327825996026e-06,
+      "loss": 0.4218,
+      "step": 436500
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 4.303042978144934e-06,
+      "loss": 0.404,
+      "step": 437000
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 4.250758130293841e-06,
+      "loss": 0.415,
+      "step": 437500
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 4.198473282442748e-06,
+      "loss": 0.4141,
+      "step": 438000
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 4.1461884345916555e-06,
+      "loss": 0.3927,
+      "step": 438500
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 4.0939035867405626e-06,
+      "loss": 0.4302,
+      "step": 439000
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 4.0416187388894705e-06,
+      "loss": 0.4193,
+      "step": 439500
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.989333891038377e-06,
+      "loss": 0.4089,
+      "step": 440000
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.937049043187285e-06,
+      "loss": 0.4003,
+      "step": 440500
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.884764195336191e-06,
+      "loss": 0.4011,
+      "step": 441000
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.832479347485099e-06,
+      "loss": 0.406,
+      "step": 441500
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.7801944996340066e-06,
+      "loss": 0.416,
+      "step": 442000
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 3.7279096517829133e-06,
+      "loss": 0.4303,
+      "step": 442500
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 3.675624803931821e-06,
+      "loss": 0.3928,
+      "step": 443000
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 3.6233399560807275e-06,
+      "loss": 0.4058,
+      "step": 443500
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 3.571055108229635e-06,
+      "loss": 0.4145,
+      "step": 444000
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 3.5187702603785426e-06,
+      "loss": 0.405,
+      "step": 444500
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 3.4664854125274497e-06,
+      "loss": 0.4074,
+      "step": 445000
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 3.4142005646763573e-06,
+      "loss": 0.4068,
+      "step": 445500
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 3.361915716825264e-06,
+      "loss": 0.4059,
+      "step": 446000
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 3.3096308689741715e-06,
+      "loss": 0.4371,
+      "step": 446500
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 3.257346021123079e-06,
+      "loss": 0.4203,
+      "step": 447000
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 3.2050611732719858e-06,
+      "loss": 0.407,
+      "step": 447500
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 3.1527763254208933e-06,
+      "loss": 0.4107,
+      "step": 448000
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 3.1004914775698004e-06,
+      "loss": 0.4242,
+      "step": 448500
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 3.0482066297187076e-06,
+      "loss": 0.4185,
+      "step": 449000
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 2.9959217818676147e-06,
+      "loss": 0.4028,
+      "step": 449500
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 2.9436369340165222e-06,
+      "loss": 0.3876,
+      "step": 450000
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 2.8913520861654294e-06,
+      "loss": 0.3924,
+      "step": 450500
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 2.839067238314337e-06,
+      "loss": 0.4232,
+      "step": 451000
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 2.786782390463244e-06,
+      "loss": 0.4339,
+      "step": 451500
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 2.734497542612151e-06,
+      "loss": 0.4155,
+      "step": 452000
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 2.6822126947610583e-06,
+      "loss": 0.4212,
+      "step": 452500
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 2.6299278469099654e-06,
+      "loss": 0.4205,
+      "step": 453000
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 2.577642999058873e-06,
+      "loss": 0.4055,
+      "step": 453500
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 2.52535815120778e-06,
+      "loss": 0.4073,
+      "step": 454000
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 2.4730733033566876e-06,
+      "loss": 0.4086,
+      "step": 454500
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 2.4207884555055947e-06,
+      "loss": 0.4472,
+      "step": 455000
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 2.368503607654502e-06,
+      "loss": 0.4278,
+      "step": 455500
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 2.316218759803409e-06,
+      "loss": 0.4158,
+      "step": 456000
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 2.2639339119523165e-06,
+      "loss": 0.4119,
+      "step": 456500
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 2.2116490641012236e-06,
+      "loss": 0.3938,
+      "step": 457000
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 2.1593642162501308e-06,
+      "loss": 0.4163,
+      "step": 457500
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 2.107079368399038e-06,
+      "loss": 0.3945,
+      "step": 458000
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 2.054794520547945e-06,
+      "loss": 0.4338,
+      "step": 458500
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 2.0025096726968526e-06,
+      "loss": 0.4274,
+      "step": 459000
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 1.95022482484576e-06,
+      "loss": 0.4125,
+      "step": 459500
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 1.897939976994667e-06,
+      "loss": 0.4229,
+      "step": 460000
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 1.8456551291435743e-06,
+      "loss": 0.4063,
+      "step": 460500
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 1.7933702812924815e-06,
+      "loss": 0.4322,
+      "step": 461000
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 1.7410854334413886e-06,
+      "loss": 0.41,
+      "step": 461500
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 1.6888005855902961e-06,
+      "loss": 0.3928,
+      "step": 462000
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 1.6365157377392033e-06,
+      "loss": 0.4201,
+      "step": 462500
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 1.5842308898881106e-06,
+      "loss": 0.4173,
+      "step": 463000
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 1.5319460420370177e-06,
+      "loss": 0.417,
+      "step": 463500
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 1.479661194185925e-06,
+      "loss": 0.3899,
+      "step": 464000
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 1.4273763463348322e-06,
+      "loss": 0.419,
+      "step": 464500
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 1.3750914984837395e-06,
+      "loss": 0.4127,
+      "step": 465000
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 1.3228066506326468e-06,
+      "loss": 0.4289,
+      "step": 465500
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 1.270521802781554e-06,
+      "loss": 0.4348,
+      "step": 466000
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 1.218236954930461e-06,
+      "loss": 0.3801,
+      "step": 466500
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 1.1659521070793684e-06,
+      "loss": 0.4107,
+      "step": 467000
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 1.1136672592282758e-06,
+      "loss": 0.4241,
+      "step": 467500
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 1.061382411377183e-06,
+      "loss": 0.4128,
+      "step": 468000
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 1.0090975635260902e-06,
+      "loss": 0.4336,
+      "step": 468500
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 9.568127156749973e-07,
+      "loss": 0.4109,
+      "step": 469000
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 9.045278678239048e-07,
+      "loss": 0.3986,
+      "step": 469500
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 8.522430199728119e-07,
+      "loss": 0.4232,
+      "step": 470000
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 7.999581721217191e-07,
+      "loss": 0.4168,
+      "step": 470500
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 7.476733242706265e-07,
+      "loss": 0.4261,
+      "step": 471000
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 6.953884764195336e-07,
+      "loss": 0.399,
+      "step": 471500
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 6.431036285684409e-07,
+      "loss": 0.416,
+      "step": 472000
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 5.908187807173481e-07,
+      "loss": 0.4144,
+      "step": 472500
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 5.385339328662554e-07,
+      "loss": 0.3898,
+      "step": 473000
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 4.862490850151626e-07,
+      "loss": 0.392,
+      "step": 473500
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.339642371640699e-07,
+      "loss": 0.425,
+      "step": 474000
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.816793893129771e-07,
+      "loss": 0.4121,
+      "step": 474500
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 3.293945414618844e-07,
+      "loss": 0.4141,
+      "step": 475000
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 2.771096936107916e-07,
+      "loss": 0.4202,
+      "step": 475500
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 2.2482484575969887e-07,
+      "loss": 0.4116,
+      "step": 476000
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 1.725399979086061e-07,
+      "loss": 0.4211,
+      "step": 476500
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.2025515005751333e-07,
+      "loss": 0.3955,
+      "step": 477000
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 6.797030220642058e-08,
+      "loss": 0.4151,
+      "step": 477500
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 1.5685454355327827e-08,
+      "loss": 0.4201,
+      "step": 478000
+    },
+    {
+      "epoch": 10.0,
+      "step": 478150,
+      "total_flos": 2.0234881444387553e+18,
+      "train_runtime": 143076.1453,
+      "train_samples_per_second": 3.342
+    }
+  ],
+  "max_steps": 478150,
+  "num_train_epochs": 10,
+  "total_flos": 2.0234881444387553e+18,
+  "trial_name": null,
+  "trial_params": null
+}