diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -1,8 +1,8 @@
 {
   "best_metric": null,
   "best_model_checkpoint": null,
-  "epoch": 10.0,
-  "global_step": 2018430,
+  "epoch": 17.0,
+  "global_step": 3431331,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -24332,11 +24332,17044 @@
       "eval_samples_per_second": 1745.728,
       "eval_steps_per_second": 72.743,
       "step": 2018430
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.500164236560099e-05,
+      "loss": 2.4274,
+      "step": 2018500
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 4.500040377917491e-05,
+      "loss": 2.4295,
+      "step": 2019000
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.4999165192748824e-05,
+      "loss": 2.3972,
+      "step": 2019500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499792660632274e-05,
+      "loss": 2.4053,
+      "step": 2020000
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499668801989666e-05,
+      "loss": 2.4178,
+      "step": 2020500
+    },
+    {
+      "epoch": 10.01,
+      "learning_rate": 4.499544943347057e-05,
+      "loss": 2.4151,
+      "step": 2021000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.4994210847044485e-05,
+      "loss": 2.4157,
+      "step": 2021500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.49929722606184e-05,
+      "loss": 2.4145,
+      "step": 2022000
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.499173367419232e-05,
+      "loss": 2.4341,
+      "step": 2022500
+    },
+    {
+      "epoch": 10.02,
+      "learning_rate": 4.499050004211194e-05,
+      "loss": 2.4384,
+      "step": 2023000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.4989261455685856e-05,
+      "loss": 2.4209,
+      "step": 2023500
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.498802286925977e-05,
+      "loss": 2.4392,
+      "step": 2024000
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.498678428283369e-05,
+      "loss": 2.4234,
+      "step": 2024500
+    },
+    {
+      "epoch": 10.03,
+      "learning_rate": 4.498554569640761e-05,
+      "loss": 2.412,
+      "step": 2025000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4984307109981524e-05,
+      "loss": 2.4119,
+      "step": 2025500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.498306852355544e-05,
+      "loss": 2.4244,
+      "step": 2026000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.498182993712936e-05,
+      "loss": 2.4498,
+      "step": 2026500
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.4980591350703275e-05,
+      "loss": 2.4358,
+      "step": 2027000
+    },
+    {
+      "epoch": 10.04,
+      "learning_rate": 4.497935276427719e-05,
+      "loss": 2.4245,
+      "step": 2027500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.497811913219681e-05,
+      "loss": 2.4284,
+      "step": 2028000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.497688054577073e-05,
+      "loss": 2.4131,
+      "step": 2028500
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.497564443651749e-05,
+      "loss": 2.4187,
+      "step": 2029000
+    },
+    {
+      "epoch": 10.05,
+      "learning_rate": 4.497440585009141e-05,
+      "loss": 2.4189,
+      "step": 2029500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.4973167263665325e-05,
+      "loss": 2.4364,
+      "step": 2030000
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.497192867723924e-05,
+      "loss": 2.4455,
+      "step": 2030500
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.497069009081316e-05,
+      "loss": 2.4064,
+      "step": 2031000
+    },
+    {
+      "epoch": 10.06,
+      "learning_rate": 4.4969451504387076e-05,
+      "loss": 2.4481,
+      "step": 2031500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.4968212917960986e-05,
+      "loss": 2.4151,
+      "step": 2032000
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.49669743315349e-05,
+      "loss": 2.4254,
+      "step": 2032500
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.496573574510882e-05,
+      "loss": 2.428,
+      "step": 2033000
+    },
+    {
+      "epoch": 10.07,
+      "learning_rate": 4.496449715868274e-05,
+      "loss": 2.4263,
+      "step": 2033500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4963258572256654e-05,
+      "loss": 2.4114,
+      "step": 2034000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.496201998583057e-05,
+      "loss": 2.4332,
+      "step": 2034500
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.496078139940449e-05,
+      "loss": 2.4325,
+      "step": 2035000
+    },
+    {
+      "epoch": 10.08,
+      "learning_rate": 4.4959545290151257e-05,
+      "loss": 2.4009,
+      "step": 2035500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4958306703725173e-05,
+      "loss": 2.435,
+      "step": 2036000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.495706811729909e-05,
+      "loss": 2.4103,
+      "step": 2036500
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.495582953087301e-05,
+      "loss": 2.4359,
+      "step": 2037000
+    },
+    {
+      "epoch": 10.09,
+      "learning_rate": 4.4954590944446924e-05,
+      "loss": 2.4256,
+      "step": 2037500
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.495335235802084e-05,
+      "loss": 2.4242,
+      "step": 2038000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.495211377159476e-05,
+      "loss": 2.4267,
+      "step": 2038500
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4950875185168675e-05,
+      "loss": 2.4368,
+      "step": 2039000
+    },
+    {
+      "epoch": 10.1,
+      "learning_rate": 4.4949641553088296e-05,
+      "loss": 2.4211,
+      "step": 2039500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494840296666221e-05,
+      "loss": 2.4296,
+      "step": 2040000
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494716438023613e-05,
+      "loss": 2.4401,
+      "step": 2040500
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.4945925793810047e-05,
+      "loss": 2.4441,
+      "step": 2041000
+    },
+    {
+      "epoch": 10.11,
+      "learning_rate": 4.494468720738396e-05,
+      "loss": 2.4434,
+      "step": 2041500
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4943448620957874e-05,
+      "loss": 2.4475,
+      "step": 2042000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.494221003453179e-05,
+      "loss": 2.4398,
+      "step": 2042500
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.494097144810571e-05,
+      "loss": 2.4278,
+      "step": 2043000
+    },
+    {
+      "epoch": 10.12,
+      "learning_rate": 4.4939732861679624e-05,
+      "loss": 2.4079,
+      "step": 2043500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.493849675242639e-05,
+      "loss": 2.4734,
+      "step": 2044000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.49372581660003e-05,
+      "loss": 2.4516,
+      "step": 2044500
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.493601957957422e-05,
+      "loss": 2.4456,
+      "step": 2045000
+    },
+    {
+      "epoch": 10.13,
+      "learning_rate": 4.493478099314814e-05,
+      "loss": 2.4319,
+      "step": 2045500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.4933542406722054e-05,
+      "loss": 2.4348,
+      "step": 2046000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.493230382029597e-05,
+      "loss": 2.4237,
+      "step": 2046500
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.493106771104275e-05,
+      "loss": 2.4202,
+      "step": 2047000
+    },
+    {
+      "epoch": 10.14,
+      "learning_rate": 4.492982912461666e-05,
+      "loss": 2.421,
+      "step": 2047500
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.492859301536343e-05,
+      "loss": 2.4291,
+      "step": 2048000
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.4927356906110194e-05,
+      "loss": 2.422,
+      "step": 2048500
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.492611831968411e-05,
+      "loss": 2.4306,
+      "step": 2049000
+    },
+    {
+      "epoch": 10.15,
+      "learning_rate": 4.492487973325803e-05,
+      "loss": 2.4154,
+      "step": 2049500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4923641146831945e-05,
+      "loss": 2.4207,
+      "step": 2050000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.492240256040586e-05,
+      "loss": 2.4277,
+      "step": 2050500
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.492116397397978e-05,
+      "loss": 2.4082,
+      "step": 2051000
+    },
+    {
+      "epoch": 10.16,
+      "learning_rate": 4.4919925387553696e-05,
+      "loss": 2.4284,
+      "step": 2051500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.491868680112761e-05,
+      "loss": 2.4324,
+      "step": 2052000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.491744821470153e-05,
+      "loss": 2.418,
+      "step": 2052500
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.491620962827545e-05,
+      "loss": 2.4193,
+      "step": 2053000
+    },
+    {
+      "epoch": 10.17,
+      "learning_rate": 4.4914971041849364e-05,
+      "loss": 2.4341,
+      "step": 2053500
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4913732455423274e-05,
+      "loss": 2.4189,
+      "step": 2054000
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.491249634617005e-05,
+      "loss": 2.4323,
+      "step": 2054500
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.4911257759743966e-05,
+      "loss": 2.4587,
+      "step": 2055000
+    },
+    {
+      "epoch": 10.18,
+      "learning_rate": 4.491002165049073e-05,
+      "loss": 2.4278,
+      "step": 2055500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.4908783064064645e-05,
+      "loss": 2.4202,
+      "step": 2056000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490754447763856e-05,
+      "loss": 2.4481,
+      "step": 2056500
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490630589121248e-05,
+      "loss": 2.4114,
+      "step": 2057000
+    },
+    {
+      "epoch": 10.19,
+      "learning_rate": 4.490506978195925e-05,
+      "loss": 2.4111,
+      "step": 2057500
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4903831195533165e-05,
+      "loss": 2.4391,
+      "step": 2058000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.490259260910708e-05,
+      "loss": 2.4327,
+      "step": 2058500
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4901354022681e-05,
+      "loss": 2.4279,
+      "step": 2059000
+    },
+    {
+      "epoch": 10.2,
+      "learning_rate": 4.4900115436254916e-05,
+      "loss": 2.4355,
+      "step": 2059500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4898881804174536e-05,
+      "loss": 2.4252,
+      "step": 2060000
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.489764321774845e-05,
+      "loss": 2.4196,
+      "step": 2060500
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.4896404631322363e-05,
+      "loss": 2.4315,
+      "step": 2061000
+    },
+    {
+      "epoch": 10.21,
+      "learning_rate": 4.489516604489628e-05,
+      "loss": 2.4199,
+      "step": 2061500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.48939274584702e-05,
+      "loss": 2.4235,
+      "step": 2062000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.4892688872044114e-05,
+      "loss": 2.3984,
+      "step": 2062500
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.489145028561803e-05,
+      "loss": 2.4339,
+      "step": 2063000
+    },
+    {
+      "epoch": 10.22,
+      "learning_rate": 4.489021169919195e-05,
+      "loss": 2.3881,
+      "step": 2063500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4888973112765865e-05,
+      "loss": 2.4048,
+      "step": 2064000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.488773452633978e-05,
+      "loss": 2.426,
+      "step": 2064500
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.48864959399137e-05,
+      "loss": 2.4169,
+      "step": 2065000
+    },
+    {
+      "epoch": 10.23,
+      "learning_rate": 4.4885257353487616e-05,
+      "loss": 2.4291,
+      "step": 2065500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488402124423438e-05,
+      "loss": 2.422,
+      "step": 2066000
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.4882782657808295e-05,
+      "loss": 2.4173,
+      "step": 2066500
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488154407138221e-05,
+      "loss": 2.415,
+      "step": 2067000
+    },
+    {
+      "epoch": 10.24,
+      "learning_rate": 4.488030548495613e-05,
+      "loss": 2.4058,
+      "step": 2067500
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.4879066898530046e-05,
+      "loss": 2.4282,
+      "step": 2068000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.4877830789276814e-05,
+      "loss": 2.4527,
+      "step": 2068500
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.487659220285073e-05,
+      "loss": 2.4305,
+      "step": 2069000
+    },
+    {
+      "epoch": 10.25,
+      "learning_rate": 4.487535361642465e-05,
+      "loss": 2.4422,
+      "step": 2069500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4874115029998565e-05,
+      "loss": 2.4138,
+      "step": 2070000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.4872878920745334e-05,
+      "loss": 2.4326,
+      "step": 2070500
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.487164033431925e-05,
+      "loss": 2.4561,
+      "step": 2071000
+    },
+    {
+      "epoch": 10.26,
+      "learning_rate": 4.487040174789317e-05,
+      "loss": 2.4219,
+      "step": 2071500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4869163161467085e-05,
+      "loss": 2.4363,
+      "step": 2072000
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.4867924575040995e-05,
+      "loss": 2.4182,
+      "step": 2072500
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.486668598861491e-05,
+      "loss": 2.4498,
+      "step": 2073000
+    },
+    {
+      "epoch": 10.27,
+      "learning_rate": 4.486544740218883e-05,
+      "loss": 2.4262,
+      "step": 2073500
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.4864208815762746e-05,
+      "loss": 2.4582,
+      "step": 2074000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.4862972706509515e-05,
+      "loss": 2.4353,
+      "step": 2074500
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.486173412008343e-05,
+      "loss": 2.4298,
+      "step": 2075000
+    },
+    {
+      "epoch": 10.28,
+      "learning_rate": 4.486049553365735e-05,
+      "loss": 2.4233,
+      "step": 2075500
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4859259424404124e-05,
+      "loss": 2.4328,
+      "step": 2076000
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.4858023315150886e-05,
+      "loss": 2.4591,
+      "step": 2076500
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.48567847287248e-05,
+      "loss": 2.4568,
+      "step": 2077000
+    },
+    {
+      "epoch": 10.29,
+      "learning_rate": 4.485554614229872e-05,
+      "loss": 2.4302,
+      "step": 2077500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.485430755587264e-05,
+      "loss": 2.4473,
+      "step": 2078000
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4853071446619406e-05,
+      "loss": 2.4297,
+      "step": 2078500
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.4851835337366174e-05,
+      "loss": 2.4359,
+      "step": 2079000
+    },
+    {
+      "epoch": 10.3,
+      "learning_rate": 4.485059675094009e-05,
+      "loss": 2.4481,
+      "step": 2079500
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4849358164514e-05,
+      "loss": 2.4354,
+      "step": 2080000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.484811957808792e-05,
+      "loss": 2.435,
+      "step": 2080500
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.484688346883469e-05,
+      "loss": 2.4235,
+      "step": 2081000
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.4845644882408604e-05,
+      "loss": 2.419,
+      "step": 2081500
+    },
+    {
+      "epoch": 10.31,
+      "learning_rate": 4.484440629598252e-05,
+      "loss": 2.4439,
+      "step": 2082000
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.484316770955644e-05,
+      "loss": 2.423,
+      "step": 2082500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.4841929123130355e-05,
+      "loss": 2.4478,
+      "step": 2083000
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.484069053670427e-05,
+      "loss": 2.4356,
+      "step": 2083500
+    },
+    {
+      "epoch": 10.32,
+      "learning_rate": 4.483945195027819e-05,
+      "loss": 2.4367,
+      "step": 2084000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4838213363852106e-05,
+      "loss": 2.4409,
+      "step": 2084500
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.483697477742602e-05,
+      "loss": 2.4238,
+      "step": 2085000
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.483573619099994e-05,
+      "loss": 2.4366,
+      "step": 2085500
+    },
+    {
+      "epoch": 10.33,
+      "learning_rate": 4.4834497604573856e-05,
+      "loss": 2.4146,
+      "step": 2086000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4833259018147773e-05,
+      "loss": 2.4158,
+      "step": 2086500
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.483202043172169e-05,
+      "loss": 2.4511,
+      "step": 2087000
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.483078184529561e-05,
+      "loss": 2.42,
+      "step": 2087500
+    },
+    {
+      "epoch": 10.34,
+      "learning_rate": 4.4829543258869524e-05,
+      "loss": 2.4163,
+      "step": 2088000
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.482830467244344e-05,
+      "loss": 2.4279,
+      "step": 2088500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.482706608601735e-05,
+      "loss": 2.451,
+      "step": 2089000
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.482582749959127e-05,
+      "loss": 2.4472,
+      "step": 2089500
+    },
+    {
+      "epoch": 10.35,
+      "learning_rate": 4.4824588913165185e-05,
+      "loss": 2.4138,
+      "step": 2090000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.48233503267391e-05,
+      "loss": 2.4263,
+      "step": 2090500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.482211174031302e-05,
+      "loss": 2.4066,
+      "step": 2091000
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.482087563105979e-05,
+      "loss": 2.4424,
+      "step": 2091500
+    },
+    {
+      "epoch": 10.36,
+      "learning_rate": 4.48196370446337e-05,
+      "loss": 2.4485,
+      "step": 2092000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4818398458207615e-05,
+      "loss": 2.4437,
+      "step": 2092500
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.481716234895439e-05,
+      "loss": 2.4505,
+      "step": 2093000
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.481592376252831e-05,
+      "loss": 2.4309,
+      "step": 2093500
+    },
+    {
+      "epoch": 10.37,
+      "learning_rate": 4.4814685176102224e-05,
+      "loss": 2.4408,
+      "step": 2094000
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481344658967614e-05,
+      "loss": 2.4208,
+      "step": 2094500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481220800325005e-05,
+      "loss": 2.3977,
+      "step": 2095000
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.481097189399682e-05,
+      "loss": 2.4398,
+      "step": 2095500
+    },
+    {
+      "epoch": 10.38,
+      "learning_rate": 4.480973330757074e-05,
+      "loss": 2.4252,
+      "step": 2096000
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4808494721144654e-05,
+      "loss": 2.4061,
+      "step": 2096500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.480725613471857e-05,
+      "loss": 2.4416,
+      "step": 2097000
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.480601754829249e-05,
+      "loss": 2.4346,
+      "step": 2097500
+    },
+    {
+      "epoch": 10.39,
+      "learning_rate": 4.4804778961866405e-05,
+      "loss": 2.4364,
+      "step": 2098000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4803540375440315e-05,
+      "loss": 2.4197,
+      "step": 2098500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.480230178901423e-05,
+      "loss": 2.423,
+      "step": 2099000
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.480106320258815e-05,
+      "loss": 2.4381,
+      "step": 2099500
+    },
+    {
+      "epoch": 10.4,
+      "learning_rate": 4.4799824616162066e-05,
+      "loss": 2.4665,
+      "step": 2100000
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.479858602973598e-05,
+      "loss": 2.4273,
+      "step": 2100500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.479734992048276e-05,
+      "loss": 2.4372,
+      "step": 2101000
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.479611133405667e-05,
+      "loss": 2.429,
+      "step": 2101500
+    },
+    {
+      "epoch": 10.41,
+      "learning_rate": 4.4794872747630585e-05,
+      "loss": 2.4399,
+      "step": 2102000
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.47936341612045e-05,
+      "loss": 2.4392,
+      "step": 2102500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.479239805195127e-05,
+      "loss": 2.4241,
+      "step": 2103000
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.479115946552519e-05,
+      "loss": 2.4679,
+      "step": 2103500
+    },
+    {
+      "epoch": 10.42,
+      "learning_rate": 4.4789920879099105e-05,
+      "loss": 2.4386,
+      "step": 2104000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.4788682292673015e-05,
+      "loss": 2.4394,
+      "step": 2104500
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.478744370624693e-05,
+      "loss": 2.4168,
+      "step": 2105000
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.478620511982085e-05,
+      "loss": 2.4562,
+      "step": 2105500
+    },
+    {
+      "epoch": 10.43,
+      "learning_rate": 4.4784969010567625e-05,
+      "loss": 2.4321,
+      "step": 2106000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.478373042414154e-05,
+      "loss": 2.4339,
+      "step": 2106500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.478249183771546e-05,
+      "loss": 2.4364,
+      "step": 2107000
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.478125325128937e-05,
+      "loss": 2.3967,
+      "step": 2107500
+    },
+    {
+      "epoch": 10.44,
+      "learning_rate": 4.4780014664863286e-05,
+      "loss": 2.4516,
+      "step": 2108000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.47787760784372e-05,
+      "loss": 2.4217,
+      "step": 2108500
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.477753749201112e-05,
+      "loss": 2.418,
+      "step": 2109000
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.4776298905585036e-05,
+      "loss": 2.4375,
+      "step": 2109500
+    },
+    {
+      "epoch": 10.45,
+      "learning_rate": 4.477506031915895e-05,
+      "loss": 2.4179,
+      "step": 2110000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.477382173273287e-05,
+      "loss": 2.4415,
+      "step": 2110500
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.477258562347963e-05,
+      "loss": 2.4335,
+      "step": 2111000
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.477134703705355e-05,
+      "loss": 2.4394,
+      "step": 2111500
+    },
+    {
+      "epoch": 10.46,
+      "learning_rate": 4.4770108450627466e-05,
+      "loss": 2.441,
+      "step": 2112000
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.476886986420138e-05,
+      "loss": 2.4271,
+      "step": 2112500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.47676312777753e-05,
+      "loss": 2.4607,
+      "step": 2113000
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.476639269134922e-05,
+      "loss": 2.4462,
+      "step": 2113500
+    },
+    {
+      "epoch": 10.47,
+      "learning_rate": 4.4765154104923134e-05,
+      "loss": 2.4493,
+      "step": 2114000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.47639179956699e-05,
+      "loss": 2.4344,
+      "step": 2114500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.476267940924382e-05,
+      "loss": 2.4534,
+      "step": 2115000
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.4761440822817736e-05,
+      "loss": 2.4273,
+      "step": 2115500
+    },
+    {
+      "epoch": 10.48,
+      "learning_rate": 4.476020223639165e-05,
+      "loss": 2.4293,
+      "step": 2116000
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.475896612713842e-05,
+      "loss": 2.4547,
+      "step": 2116500
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.475773001788519e-05,
+      "loss": 2.4524,
+      "step": 2117000
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.475649143145911e-05,
+      "loss": 2.4532,
+      "step": 2117500
+    },
+    {
+      "epoch": 10.49,
+      "learning_rate": 4.4755252845033025e-05,
+      "loss": 2.4419,
+      "step": 2118000
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.475401425860694e-05,
+      "loss": 2.4467,
+      "step": 2118500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.475277567218086e-05,
+      "loss": 2.4355,
+      "step": 2119000
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.475153956292763e-05,
+      "loss": 2.4465,
+      "step": 2119500
+    },
+    {
+      "epoch": 10.5,
+      "learning_rate": 4.4750300976501544e-05,
+      "loss": 2.4484,
+      "step": 2120000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.474906239007546e-05,
+      "loss": 2.4718,
+      "step": 2120500
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.474782380364937e-05,
+      "loss": 2.4429,
+      "step": 2121000
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.474658521722329e-05,
+      "loss": 2.461,
+      "step": 2121500
+    },
+    {
+      "epoch": 10.51,
+      "learning_rate": 4.4745346630797205e-05,
+      "loss": 2.4328,
+      "step": 2122000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474410804437112e-05,
+      "loss": 2.457,
+      "step": 2122500
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474286945794504e-05,
+      "loss": 2.4409,
+      "step": 2123000
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.474163087151895e-05,
+      "loss": 2.4528,
+      "step": 2123500
+    },
+    {
+      "epoch": 10.52,
+      "learning_rate": 4.4740392285092866e-05,
+      "loss": 2.4398,
+      "step": 2124000
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.473915617583964e-05,
+      "loss": 2.419,
+      "step": 2124500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.473791758941356e-05,
+      "loss": 2.4351,
+      "step": 2125000
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.4736679002987476e-05,
+      "loss": 2.4138,
+      "step": 2125500
+    },
+    {
+      "epoch": 10.53,
+      "learning_rate": 4.473544041656139e-05,
+      "loss": 2.4318,
+      "step": 2126000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.47342018301353e-05,
+      "loss": 2.4301,
+      "step": 2126500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473296324370922e-05,
+      "loss": 2.4201,
+      "step": 2127000
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.473172465728314e-05,
+      "loss": 2.439,
+      "step": 2127500
+    },
+    {
+      "epoch": 10.54,
+      "learning_rate": 4.4730488548029905e-05,
+      "loss": 2.4401,
+      "step": 2128000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.472924996160382e-05,
+      "loss": 2.4265,
+      "step": 2128500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.472801137517774e-05,
+      "loss": 2.4333,
+      "step": 2129000
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.4726772788751656e-05,
+      "loss": 2.4448,
+      "step": 2129500
+    },
+    {
+      "epoch": 10.55,
+      "learning_rate": 4.4725534202325566e-05,
+      "loss": 2.4431,
+      "step": 2130000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.472429561589948e-05,
+      "loss": 2.4097,
+      "step": 2130500
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.47230570294734e-05,
+      "loss": 2.4432,
+      "step": 2131000
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.472181844304732e-05,
+      "loss": 2.4459,
+      "step": 2131500
+    },
+    {
+      "epoch": 10.56,
+      "learning_rate": 4.472058233379409e-05,
+      "loss": 2.4461,
+      "step": 2132000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.4719343747368e-05,
+      "loss": 2.4396,
+      "step": 2132500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.471810763811478e-05,
+      "loss": 2.4402,
+      "step": 2133000
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.4716869051688695e-05,
+      "loss": 2.4444,
+      "step": 2133500
+    },
+    {
+      "epoch": 10.57,
+      "learning_rate": 4.471563046526261e-05,
+      "loss": 2.4261,
+      "step": 2134000
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.471439187883652e-05,
+      "loss": 2.4395,
+      "step": 2134500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.471315329241044e-05,
+      "loss": 2.4222,
+      "step": 2135000
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.4711914705984356e-05,
+      "loss": 2.4348,
+      "step": 2135500
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.4710678596731125e-05,
+      "loss": 2.4261,
+      "step": 2136000
+    },
+    {
+      "epoch": 10.58,
+      "learning_rate": 4.470944001030504e-05,
+      "loss": 2.4462,
+      "step": 2136500
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.470820142387896e-05,
+      "loss": 2.4554,
+      "step": 2137000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.4706962837452876e-05,
+      "loss": 2.4547,
+      "step": 2137500
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.470572425102679e-05,
+      "loss": 2.4258,
+      "step": 2138000
+    },
+    {
+      "epoch": 10.59,
+      "learning_rate": 4.470448566460071e-05,
+      "loss": 2.4368,
+      "step": 2138500
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.470324707817462e-05,
+      "loss": 2.4417,
+      "step": 2139000
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.470200849174854e-05,
+      "loss": 2.4279,
+      "step": 2139500
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.470077238249531e-05,
+      "loss": 2.4251,
+      "step": 2140000
+    },
+    {
+      "epoch": 10.6,
+      "learning_rate": 4.469953379606923e-05,
+      "loss": 2.4378,
+      "step": 2140500
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.469829520964314e-05,
+      "loss": 2.4518,
+      "step": 2141000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.4697056623217057e-05,
+      "loss": 2.4638,
+      "step": 2141500
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.4695818036790973e-05,
+      "loss": 2.4405,
+      "step": 2142000
+    },
+    {
+      "epoch": 10.61,
+      "learning_rate": 4.469457945036489e-05,
+      "loss": 2.444,
+      "step": 2142500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.469334334111166e-05,
+      "loss": 2.4427,
+      "step": 2143000
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.4692104754685576e-05,
+      "loss": 2.4335,
+      "step": 2143500
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.469086616825949e-05,
+      "loss": 2.4229,
+      "step": 2144000
+    },
+    {
+      "epoch": 10.62,
+      "learning_rate": 4.468962758183341e-05,
+      "loss": 2.4291,
+      "step": 2144500
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.468838899540732e-05,
+      "loss": 2.4256,
+      "step": 2145000
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.4687152886154096e-05,
+      "loss": 2.4089,
+      "step": 2145500
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.468591429972801e-05,
+      "loss": 2.4351,
+      "step": 2146000
+    },
+    {
+      "epoch": 10.63,
+      "learning_rate": 4.468467571330193e-05,
+      "loss": 2.4628,
+      "step": 2146500
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.468343960404869e-05,
+      "loss": 2.4254,
+      "step": 2147000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.468220349479546e-05,
+      "loss": 2.4318,
+      "step": 2147500
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.468096490836938e-05,
+      "loss": 2.4339,
+      "step": 2148000
+    },
+    {
+      "epoch": 10.64,
+      "learning_rate": 4.4679726321943294e-05,
+      "loss": 2.4517,
+      "step": 2148500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.467848773551721e-05,
+      "loss": 2.4626,
+      "step": 2149000
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.467725162626398e-05,
+      "loss": 2.4349,
+      "step": 2149500
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.467601551701075e-05,
+      "loss": 2.4428,
+      "step": 2150000
+    },
+    {
+      "epoch": 10.65,
+      "learning_rate": 4.4674776930584666e-05,
+      "loss": 2.4328,
+      "step": 2150500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467353834415858e-05,
+      "loss": 2.4354,
+      "step": 2151000
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.46722997577325e-05,
+      "loss": 2.4413,
+      "step": 2151500
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.467106117130641e-05,
+      "loss": 2.4063,
+      "step": 2152000
+    },
+    {
+      "epoch": 10.66,
+      "learning_rate": 4.466982258488033e-05,
+      "loss": 2.4428,
+      "step": 2152500
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.4668583998454244e-05,
+      "loss": 2.4312,
+      "step": 2153000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.466734541202816e-05,
+      "loss": 2.4444,
+      "step": 2153500
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.466610682560208e-05,
+      "loss": 2.4416,
+      "step": 2154000
+    },
+    {
+      "epoch": 10.67,
+      "learning_rate": 4.4664868239175994e-05,
+      "loss": 2.4138,
+      "step": 2154500
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.466362965274991e-05,
+      "loss": 2.4313,
+      "step": 2155000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.466239106632383e-05,
+      "loss": 2.419,
+      "step": 2155500
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.4661152479897745e-05,
+      "loss": 2.4566,
+      "step": 2156000
+    },
+    {
+      "epoch": 10.68,
+      "learning_rate": 4.4659916370644514e-05,
+      "loss": 2.4375,
+      "step": 2156500
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.465867778421843e-05,
+      "loss": 2.4304,
+      "step": 2157000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.465743919779235e-05,
+      "loss": 2.4378,
+      "step": 2157500
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.465620061136626e-05,
+      "loss": 2.4383,
+      "step": 2158000
+    },
+    {
+      "epoch": 10.69,
+      "learning_rate": 4.4654962024940175e-05,
+      "loss": 2.4265,
+      "step": 2158500
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.4653725915686944e-05,
+      "loss": 2.4481,
+      "step": 2159000
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.465248732926086e-05,
+      "loss": 2.4098,
+      "step": 2159500
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.4651251220007636e-05,
+      "loss": 2.4331,
+      "step": 2160000
+    },
+    {
+      "epoch": 10.7,
+      "learning_rate": 4.465001263358155e-05,
+      "loss": 2.419,
+      "step": 2160500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.464877404715547e-05,
+      "loss": 2.43,
+      "step": 2161000
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.464753546072938e-05,
+      "loss": 2.4379,
+      "step": 2161500
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.46462968743033e-05,
+      "loss": 2.4677,
+      "step": 2162000
+    },
+    {
+      "epoch": 10.71,
+      "learning_rate": 4.4645058287877214e-05,
+      "loss": 2.431,
+      "step": 2162500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.464381970145113e-05,
+      "loss": 2.4455,
+      "step": 2163000
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.464258111502505e-05,
+      "loss": 2.4502,
+      "step": 2163500
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.4641342528598965e-05,
+      "loss": 2.4302,
+      "step": 2164000
+    },
+    {
+      "epoch": 10.72,
+      "learning_rate": 4.464010641934573e-05,
+      "loss": 2.4198,
+      "step": 2164500
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.4638867832919644e-05,
+      "loss": 2.4332,
+      "step": 2165000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.463762924649356e-05,
+      "loss": 2.4356,
+      "step": 2165500
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.463639066006748e-05,
+      "loss": 2.422,
+      "step": 2166000
+    },
+    {
+      "epoch": 10.73,
+      "learning_rate": 4.4635152073641395e-05,
+      "loss": 2.4193,
+      "step": 2166500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.463391348721531e-05,
+      "loss": 2.4501,
+      "step": 2167000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.463267490078923e-05,
+      "loss": 2.4621,
+      "step": 2167500
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.4631436314363145e-05,
+      "loss": 2.4338,
+      "step": 2168000
+    },
+    {
+      "epoch": 10.74,
+      "learning_rate": 4.4630200205109914e-05,
+      "loss": 2.4167,
+      "step": 2168500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.462896161868383e-05,
+      "loss": 2.4113,
+      "step": 2169000
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.462772303225775e-05,
+      "loss": 2.4667,
+      "step": 2169500
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.462648692300452e-05,
+      "loss": 2.4525,
+      "step": 2170000
+    },
+    {
+      "epoch": 10.75,
+      "learning_rate": 4.4625248336578434e-05,
+      "loss": 2.4503,
+      "step": 2170500
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.4624009750152344e-05,
+      "loss": 2.4539,
+      "step": 2171000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462277116372626e-05,
+      "loss": 2.4368,
+      "step": 2171500
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.462153257730018e-05,
+      "loss": 2.4427,
+      "step": 2172000
+    },
+    {
+      "epoch": 10.76,
+      "learning_rate": 4.4620293990874095e-05,
+      "loss": 2.4173,
+      "step": 2172500
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.461905788162087e-05,
+      "loss": 2.4316,
+      "step": 2173000
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.461781929519479e-05,
+      "loss": 2.4271,
+      "step": 2173500
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.461658318594155e-05,
+      "loss": 2.4018,
+      "step": 2174000
+    },
+    {
+      "epoch": 10.77,
+      "learning_rate": 4.4615344599515466e-05,
+      "loss": 2.427,
+      "step": 2174500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.461410601308938e-05,
+      "loss": 2.4359,
+      "step": 2175000
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.46128674266633e-05,
+      "loss": 2.4292,
+      "step": 2175500
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.461163131741007e-05,
+      "loss": 2.4656,
+      "step": 2176000
+    },
+    {
+      "epoch": 10.78,
+      "learning_rate": 4.4610392730983986e-05,
+      "loss": 2.4482,
+      "step": 2176500
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.46091541445579e-05,
+      "loss": 2.4331,
+      "step": 2177000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.460791555813182e-05,
+      "loss": 2.4638,
+      "step": 2177500
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.4606676971705737e-05,
+      "loss": 2.4594,
+      "step": 2178000
+    },
+    {
+      "epoch": 10.79,
+      "learning_rate": 4.4605438385279654e-05,
+      "loss": 2.4464,
+      "step": 2178500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.460419979885357e-05,
+      "loss": 2.4558,
+      "step": 2179000
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.460296121242749e-05,
+      "loss": 2.4394,
+      "step": 2179500
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.46017226260014e-05,
+      "loss": 2.4307,
+      "step": 2180000
+    },
+    {
+      "epoch": 10.8,
+      "learning_rate": 4.4600484039575314e-05,
+      "loss": 2.4527,
+      "step": 2180500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.459924545314923e-05,
+      "loss": 2.4379,
+      "step": 2181000
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.459800686672315e-05,
+      "loss": 2.4372,
+      "step": 2181500
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.4596768280297065e-05,
+      "loss": 2.4373,
+      "step": 2182000
+    },
+    {
+      "epoch": 10.81,
+      "learning_rate": 4.4595532171043834e-05,
+      "loss": 2.4486,
+      "step": 2182500
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.459429358461775e-05,
+      "loss": 2.4385,
+      "step": 2183000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.459305747536452e-05,
+      "loss": 2.4328,
+      "step": 2183500
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.459181888893844e-05,
+      "loss": 2.4374,
+      "step": 2184000
+    },
+    {
+      "epoch": 10.82,
+      "learning_rate": 4.4590580302512354e-05,
+      "loss": 2.4451,
+      "step": 2184500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.458934171608627e-05,
+      "loss": 2.4589,
+      "step": 2185000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.458810312966019e-05,
+      "loss": 2.451,
+      "step": 2185500
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.458686702040695e-05,
+      "loss": 2.418,
+      "step": 2186000
+    },
+    {
+      "epoch": 10.83,
+      "learning_rate": 4.4585628433980866e-05,
+      "loss": 2.4478,
+      "step": 2186500
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.4584389847554783e-05,
+      "loss": 2.4271,
+      "step": 2187000
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.45831512611287e-05,
+      "loss": 2.4425,
+      "step": 2187500
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.458191267470262e-05,
+      "loss": 2.452,
+      "step": 2188000
+    },
+    {
+      "epoch": 10.84,
+      "learning_rate": 4.4580674088276534e-05,
+      "loss": 2.4415,
+      "step": 2188500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.457943550185045e-05,
+      "loss": 2.4458,
+      "step": 2189000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.457819691542436e-05,
+      "loss": 2.4406,
+      "step": 2189500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.457696080617114e-05,
+      "loss": 2.4222,
+      "step": 2190000
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.4575722219745054e-05,
+      "loss": 2.458,
+      "step": 2190500
+    },
+    {
+      "epoch": 10.85,
+      "learning_rate": 4.457448363331897e-05,
+      "loss": 2.4399,
+      "step": 2191000
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.457324504689289e-05,
+      "loss": 2.4346,
+      "step": 2191500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.4572008937639656e-05,
+      "loss": 2.4664,
+      "step": 2192000
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.4570770351213567e-05,
+      "loss": 2.4396,
+      "step": 2192500
+    },
+    {
+      "epoch": 10.86,
+      "learning_rate": 4.4569531764787484e-05,
+      "loss": 2.4306,
+      "step": 2193000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.45682931783614e-05,
+      "loss": 2.4317,
+      "step": 2193500
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.456705459193532e-05,
+      "loss": 2.4156,
+      "step": 2194000
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.4565816005509234e-05,
+      "loss": 2.444,
+      "step": 2194500
+    },
+    {
+      "epoch": 10.87,
+      "learning_rate": 4.456457741908315e-05,
+      "loss": 2.4486,
+      "step": 2195000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.456333883265707e-05,
+      "loss": 2.4602,
+      "step": 2195500
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.456210024623098e-05,
+      "loss": 2.4304,
+      "step": 2196000
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.4560861659804895e-05,
+      "loss": 2.4429,
+      "step": 2196500
+    },
+    {
+      "epoch": 10.88,
+      "learning_rate": 4.455962307337881e-05,
+      "loss": 2.4199,
+      "step": 2197000
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.455838448695273e-05,
+      "loss": 2.4348,
+      "step": 2197500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.4557145900526646e-05,
+      "loss": 2.4346,
+      "step": 2198000
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.455590979127342e-05,
+      "loss": 2.4507,
+      "step": 2198500
+    },
+    {
+      "epoch": 10.89,
+      "learning_rate": 4.455467120484733e-05,
+      "loss": 2.443,
+      "step": 2199000
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.455343261842125e-05,
+      "loss": 2.4219,
+      "step": 2199500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.455219650916802e-05,
+      "loss": 2.4231,
+      "step": 2200000
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.4550957922741934e-05,
+      "loss": 2.4443,
+      "step": 2200500
+    },
+    {
+      "epoch": 10.9,
+      "learning_rate": 4.454971933631585e-05,
+      "loss": 2.4357,
+      "step": 2201000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.454848570423547e-05,
+      "loss": 2.4494,
+      "step": 2201500
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.454724711780939e-05,
+      "loss": 2.4197,
+      "step": 2202000
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.4546008531383306e-05,
+      "loss": 2.4512,
+      "step": 2202500
+    },
+    {
+      "epoch": 10.91,
+      "learning_rate": 4.454476994495722e-05,
+      "loss": 2.445,
+      "step": 2203000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.454353135853114e-05,
+      "loss": 2.449,
+      "step": 2203500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.45422952492779e-05,
+      "loss": 2.4383,
+      "step": 2204000
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.454105666285182e-05,
+      "loss": 2.4595,
+      "step": 2204500
+    },
+    {
+      "epoch": 10.92,
+      "learning_rate": 4.4539818076425736e-05,
+      "loss": 2.4192,
+      "step": 2205000
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.453857948999965e-05,
+      "loss": 2.4381,
+      "step": 2205500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.453734090357357e-05,
+      "loss": 2.429,
+      "step": 2206000
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.453610479432034e-05,
+      "loss": 2.4534,
+      "step": 2206500
+    },
+    {
+      "epoch": 10.93,
+      "learning_rate": 4.4534866207894255e-05,
+      "loss": 2.4423,
+      "step": 2207000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.453362762146817e-05,
+      "loss": 2.4272,
+      "step": 2207500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.453239151221494e-05,
+      "loss": 2.4552,
+      "step": 2208000
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.453115292578886e-05,
+      "loss": 2.4172,
+      "step": 2208500
+    },
+    {
+      "epoch": 10.94,
+      "learning_rate": 4.4529914339362775e-05,
+      "loss": 2.4602,
+      "step": 2209000
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.4528675752936685e-05,
+      "loss": 2.4529,
+      "step": 2209500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.45274371665106e-05,
+      "loss": 2.4541,
+      "step": 2210000
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.452619858008452e-05,
+      "loss": 2.4447,
+      "step": 2210500
+    },
+    {
+      "epoch": 10.95,
+      "learning_rate": 4.4524959993658436e-05,
+      "loss": 2.4507,
+      "step": 2211000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.452372140723235e-05,
+      "loss": 2.4288,
+      "step": 2211500
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.452248282080627e-05,
+      "loss": 2.4157,
+      "step": 2212000
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4521244234380187e-05,
+      "loss": 2.4446,
+      "step": 2212500
+    },
+    {
+      "epoch": 10.96,
+      "learning_rate": 4.4520005647954103e-05,
+      "loss": 2.4222,
+      "step": 2213000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.451876953870087e-05,
+      "loss": 2.4498,
+      "step": 2213500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.451753095227479e-05,
+      "loss": 2.4253,
+      "step": 2214000
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.4516292365848706e-05,
+      "loss": 2.4469,
+      "step": 2214500
+    },
+    {
+      "epoch": 10.97,
+      "learning_rate": 4.451505377942262e-05,
+      "loss": 2.4336,
+      "step": 2215000
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451381519299654e-05,
+      "loss": 2.4368,
+      "step": 2215500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451257908374331e-05,
+      "loss": 2.4251,
+      "step": 2216000
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.451134297449008e-05,
+      "loss": 2.4311,
+      "step": 2216500
+    },
+    {
+      "epoch": 10.98,
+      "learning_rate": 4.4510104388063995e-05,
+      "loss": 2.4489,
+      "step": 2217000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.450886580163791e-05,
+      "loss": 2.4204,
+      "step": 2217500
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.450762721521183e-05,
+      "loss": 2.4315,
+      "step": 2218000
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.450638862878574e-05,
+      "loss": 2.4432,
+      "step": 2218500
+    },
+    {
+      "epoch": 10.99,
+      "learning_rate": 4.4505150042359655e-05,
+      "loss": 2.4581,
+      "step": 2219000
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.450391145593357e-05,
+      "loss": 2.4339,
+      "step": 2219500
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.450267286950749e-05,
+      "loss": 2.4147,
+      "step": 2220000
+    },
+    {
+      "epoch": 11.0,
+      "eval_accuracy": 0.6435940740781358,
+      "eval_accuracy_mlm": 0.596974884280312,
+      "eval_accuracy_nsp": 0.863742797861617,
+      "eval_loss": 2.406216859817505,
+      "eval_runtime": 146.12,
+      "eval_samples_per_second": 1744.86,
+      "eval_steps_per_second": 72.707,
+      "step": 2220273
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.4501434283081406e-05,
+      "loss": 2.4115,
+      "step": 2220500
+    },
+    {
+      "epoch": 11.0,
+      "learning_rate": 4.450019569665532e-05,
+      "loss": 2.3987,
+      "step": 2221000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.449895958740209e-05,
+      "loss": 2.392,
+      "step": 2221500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.449772100097601e-05,
+      "loss": 2.3835,
+      "step": 2222000
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.4496482414549926e-05,
+      "loss": 2.4125,
+      "step": 2222500
+    },
+    {
+      "epoch": 11.01,
+      "learning_rate": 4.4495243828123836e-05,
+      "loss": 2.4218,
+      "step": 2223000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.449400524169775e-05,
+      "loss": 2.4222,
+      "step": 2223500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.449276665527167e-05,
+      "loss": 2.3963,
+      "step": 2224000
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.449152806884559e-05,
+      "loss": 2.4003,
+      "step": 2224500
+    },
+    {
+      "epoch": 11.02,
+      "learning_rate": 4.4490289482419504e-05,
+      "loss": 2.4131,
+      "step": 2225000
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.448905089599342e-05,
+      "loss": 2.426,
+      "step": 2225500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.448781230956734e-05,
+      "loss": 2.4027,
+      "step": 2226000
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.4486573723141255e-05,
+      "loss": 2.4331,
+      "step": 2226500
+    },
+    {
+      "epoch": 11.03,
+      "learning_rate": 4.448533761388802e-05,
+      "loss": 2.4009,
+      "step": 2227000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448409902746194e-05,
+      "loss": 2.4009,
+      "step": 2227500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448286044103586e-05,
+      "loss": 2.4034,
+      "step": 2228000
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.4481621854609774e-05,
+      "loss": 2.4055,
+      "step": 2228500
+    },
+    {
+      "epoch": 11.04,
+      "learning_rate": 4.448038326818369e-05,
+      "loss": 2.3944,
+      "step": 2229000
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.447914468175761e-05,
+      "loss": 2.4153,
+      "step": 2229500
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.4477906095331525e-05,
+      "loss": 2.4075,
+      "step": 2230000
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.447666750890544e-05,
+      "loss": 2.4052,
+      "step": 2230500
+    },
+    {
+      "epoch": 11.05,
+      "learning_rate": 4.447542892247936e-05,
+      "loss": 2.4021,
+      "step": 2231000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.447419281322612e-05,
+      "loss": 2.3982,
+      "step": 2231500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.447295422680004e-05,
+      "loss": 2.4161,
+      "step": 2232000
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.4471715640373955e-05,
+      "loss": 2.4319,
+      "step": 2232500
+    },
+    {
+      "epoch": 11.06,
+      "learning_rate": 4.447047705394787e-05,
+      "loss": 2.4021,
+      "step": 2233000
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.446923846752179e-05,
+      "loss": 2.4029,
+      "step": 2233500
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.4467999881095705e-05,
+      "loss": 2.4136,
+      "step": 2234000
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.446676129466962e-05,
+      "loss": 2.4027,
+      "step": 2234500
+    },
+    {
+      "epoch": 11.07,
+      "learning_rate": 4.446552518541639e-05,
+      "loss": 2.4079,
+      "step": 2235000
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.446428659899031e-05,
+      "loss": 2.4259,
+      "step": 2235500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.4463048012564225e-05,
+      "loss": 2.4226,
+      "step": 2236000
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.446180942613814e-05,
+      "loss": 2.4198,
+      "step": 2236500
+    },
+    {
+      "epoch": 11.08,
+      "learning_rate": 4.446057083971206e-05,
+      "loss": 2.3826,
+      "step": 2237000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.4459332253285976e-05,
+      "loss": 2.4163,
+      "step": 2237500
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.445809366685989e-05,
+      "loss": 2.403,
+      "step": 2238000
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.445685508043381e-05,
+      "loss": 2.4073,
+      "step": 2238500
+    },
+    {
+      "epoch": 11.09,
+      "learning_rate": 4.445561897118057e-05,
+      "loss": 2.4129,
+      "step": 2239000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445438038475449e-05,
+      "loss": 2.4009,
+      "step": 2239500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.4453141798328406e-05,
+      "loss": 2.4227,
+      "step": 2240000
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445190321190232e-05,
+      "loss": 2.4006,
+      "step": 2240500
+    },
+    {
+      "epoch": 11.1,
+      "learning_rate": 4.445066710264909e-05,
+      "loss": 2.3994,
+      "step": 2241000
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.444942851622301e-05,
+      "loss": 2.4206,
+      "step": 2241500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4448189929796925e-05,
+      "loss": 2.4191,
+      "step": 2242000
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.4446953820543694e-05,
+      "loss": 2.3991,
+      "step": 2242500
+    },
+    {
+      "epoch": 11.11,
+      "learning_rate": 4.444571523411761e-05,
+      "loss": 2.4203,
+      "step": 2243000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.444447664769152e-05,
+      "loss": 2.4075,
+      "step": 2243500
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.444323806126544e-05,
+      "loss": 2.4173,
+      "step": 2244000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.4441999474839355e-05,
+      "loss": 2.4242,
+      "step": 2244500
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.444076088841327e-05,
+      "loss": 2.4012,
+      "step": 2245000
+    },
+    {
+      "epoch": 11.12,
+      "learning_rate": 4.443952230198719e-05,
+      "loss": 2.4417,
+      "step": 2245500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.4438283715561106e-05,
+      "loss": 2.3909,
+      "step": 2246000
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.443704512913502e-05,
+      "loss": 2.4142,
+      "step": 2246500
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.443580901988179e-05,
+      "loss": 2.4145,
+      "step": 2247000
+    },
+    {
+      "epoch": 11.13,
+      "learning_rate": 4.443457043345571e-05,
+      "loss": 2.4039,
+      "step": 2247500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.4433331847029625e-05,
+      "loss": 2.4166,
+      "step": 2248000
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.443209326060354e-05,
+      "loss": 2.4373,
+      "step": 2248500
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.443085467417746e-05,
+      "loss": 2.3937,
+      "step": 2249000
+    },
+    {
+      "epoch": 11.14,
+      "learning_rate": 4.4429616087751376e-05,
+      "loss": 2.4121,
+      "step": 2249500
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.442837997849814e-05,
+      "loss": 2.4216,
+      "step": 2250000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.4427141392072055e-05,
+      "loss": 2.4062,
+      "step": 2250500
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.442590280564597e-05,
+      "loss": 2.3873,
+      "step": 2251000
+    },
+    {
+      "epoch": 11.15,
+      "learning_rate": 4.442466421921989e-05,
+      "loss": 2.42,
+      "step": 2251500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.4423425632793806e-05,
+      "loss": 2.3876,
+      "step": 2252000
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.442218704636772e-05,
+      "loss": 2.4366,
+      "step": 2252500
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.442095093711449e-05,
+      "loss": 2.4273,
+      "step": 2253000
+    },
+    {
+      "epoch": 11.16,
+      "learning_rate": 4.441971235068841e-05,
+      "loss": 2.4109,
+      "step": 2253500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.4418473764262325e-05,
+      "loss": 2.4171,
+      "step": 2254000
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.441723517783624e-05,
+      "loss": 2.4281,
+      "step": 2254500
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.441599659141016e-05,
+      "loss": 2.4044,
+      "step": 2255000
+    },
+    {
+      "epoch": 11.17,
+      "learning_rate": 4.441476048215693e-05,
+      "loss": 2.3969,
+      "step": 2255500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.4413521895730845e-05,
+      "loss": 2.4335,
+      "step": 2256000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.441228330930476e-05,
+      "loss": 2.4196,
+      "step": 2256500
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.441104472287867e-05,
+      "loss": 2.4018,
+      "step": 2257000
+    },
+    {
+      "epoch": 11.18,
+      "learning_rate": 4.440980613645259e-05,
+      "loss": 2.4158,
+      "step": 2257500
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.4408567550026506e-05,
+      "loss": 2.4362,
+      "step": 2258000
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.4407331440773275e-05,
+      "loss": 2.4383,
+      "step": 2258500
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.440609285434719e-05,
+      "loss": 2.4255,
+      "step": 2259000
+    },
+    {
+      "epoch": 11.19,
+      "learning_rate": 4.440485426792111e-05,
+      "loss": 2.4434,
+      "step": 2259500
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.4403615681495026e-05,
+      "loss": 2.4004,
+      "step": 2260000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.440237709506894e-05,
+      "loss": 2.4139,
+      "step": 2260500
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.440113850864286e-05,
+      "loss": 2.4226,
+      "step": 2261000
+    },
+    {
+      "epoch": 11.2,
+      "learning_rate": 4.439990239938963e-05,
+      "loss": 2.4301,
+      "step": 2261500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.4398663812963545e-05,
+      "loss": 2.4222,
+      "step": 2262000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.439742522653746e-05,
+      "loss": 2.4225,
+      "step": 2262500
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.439618664011138e-05,
+      "loss": 2.4212,
+      "step": 2263000
+    },
+    {
+      "epoch": 11.21,
+      "learning_rate": 4.439494805368529e-05,
+      "loss": 2.4104,
+      "step": 2263500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.439371194443206e-05,
+      "loss": 2.4221,
+      "step": 2264000
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.4392473358005975e-05,
+      "loss": 2.4218,
+      "step": 2264500
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.439123477157989e-05,
+      "loss": 2.4348,
+      "step": 2265000
+    },
+    {
+      "epoch": 11.22,
+      "learning_rate": 4.438999618515381e-05,
+      "loss": 2.4062,
+      "step": 2265500
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.4388757598727726e-05,
+      "loss": 2.4061,
+      "step": 2266000
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.438751901230164e-05,
+      "loss": 2.4101,
+      "step": 2266500
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.438628290304841e-05,
+      "loss": 2.4078,
+      "step": 2267000
+    },
+    {
+      "epoch": 11.23,
+      "learning_rate": 4.438504431662233e-05,
+      "loss": 2.4107,
+      "step": 2267500
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4383805730196245e-05,
+      "loss": 2.412,
+      "step": 2268000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.438256714377016e-05,
+      "loss": 2.4134,
+      "step": 2268500
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.4381331034516924e-05,
+      "loss": 2.4223,
+      "step": 2269000
+    },
+    {
+      "epoch": 11.24,
+      "learning_rate": 4.438009244809084e-05,
+      "loss": 2.4328,
+      "step": 2269500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.437885386166476e-05,
+      "loss": 2.4224,
+      "step": 2270000
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.4377615275238675e-05,
+      "loss": 2.4191,
+      "step": 2270500
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.437637916598545e-05,
+      "loss": 2.4188,
+      "step": 2271000
+    },
+    {
+      "epoch": 11.25,
+      "learning_rate": 4.437514057955936e-05,
+      "loss": 2.4173,
+      "step": 2271500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.437390199313328e-05,
+      "loss": 2.3998,
+      "step": 2272000
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.4372663406707195e-05,
+      "loss": 2.4385,
+      "step": 2272500
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.437142482028111e-05,
+      "loss": 2.3898,
+      "step": 2273000
+    },
+    {
+      "epoch": 11.26,
+      "learning_rate": 4.437018623385503e-05,
+      "loss": 2.4457,
+      "step": 2273500
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.4368947647428945e-05,
+      "loss": 2.4537,
+      "step": 2274000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.436770906100286e-05,
+      "loss": 2.4273,
+      "step": 2274500
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.436647047457678e-05,
+      "loss": 2.4318,
+      "step": 2275000
+    },
+    {
+      "epoch": 11.27,
+      "learning_rate": 4.4365231888150696e-05,
+      "loss": 2.4194,
+      "step": 2275500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.436399330172461e-05,
+      "loss": 2.4104,
+      "step": 2276000
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.4362757192471375e-05,
+      "loss": 2.4375,
+      "step": 2276500
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.436152108321815e-05,
+      "loss": 2.4367,
+      "step": 2277000
+    },
+    {
+      "epoch": 11.28,
+      "learning_rate": 4.436028249679207e-05,
+      "loss": 2.4193,
+      "step": 2277500
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.435904638753883e-05,
+      "loss": 2.4272,
+      "step": 2278000
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.4357807801112747e-05,
+      "loss": 2.441,
+      "step": 2278500
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.4356569214686664e-05,
+      "loss": 2.4035,
+      "step": 2279000
+    },
+    {
+      "epoch": 11.29,
+      "learning_rate": 4.435533062826058e-05,
+      "loss": 2.4012,
+      "step": 2279500
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.43540920418345e-05,
+      "loss": 2.4267,
+      "step": 2280000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.4352853455408414e-05,
+      "loss": 2.4075,
+      "step": 2280500
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.4351614868982324e-05,
+      "loss": 2.4112,
+      "step": 2281000
+    },
+    {
+      "epoch": 11.3,
+      "learning_rate": 4.435037628255624e-05,
+      "loss": 2.4298,
+      "step": 2281500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434913769613016e-05,
+      "loss": 2.4208,
+      "step": 2282000
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.4347899109704075e-05,
+      "loss": 2.4449,
+      "step": 2282500
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434666052327799e-05,
+      "loss": 2.4218,
+      "step": 2283000
+    },
+    {
+      "epoch": 11.31,
+      "learning_rate": 4.434542193685191e-05,
+      "loss": 2.4302,
+      "step": 2283500
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.4344183350425826e-05,
+      "loss": 2.4257,
+      "step": 2284000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.4342947241172595e-05,
+      "loss": 2.4383,
+      "step": 2284500
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.434170865474651e-05,
+      "loss": 2.3936,
+      "step": 2285000
+    },
+    {
+      "epoch": 11.32,
+      "learning_rate": 4.434047254549328e-05,
+      "loss": 2.4184,
+      "step": 2285500
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.43392339590672e-05,
+      "loss": 2.413,
+      "step": 2286000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.4337995372641114e-05,
+      "loss": 2.4455,
+      "step": 2286500
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.4336756786215025e-05,
+      "loss": 2.4197,
+      "step": 2287000
+    },
+    {
+      "epoch": 11.33,
+      "learning_rate": 4.433551819978894e-05,
+      "loss": 2.4045,
+      "step": 2287500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.433427961336286e-05,
+      "loss": 2.4081,
+      "step": 2288000
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.4333041026936775e-05,
+      "loss": 2.4123,
+      "step": 2288500
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.433180244051069e-05,
+      "loss": 2.4212,
+      "step": 2289000
+    },
+    {
+      "epoch": 11.34,
+      "learning_rate": 4.433056633125747e-05,
+      "loss": 2.4364,
+      "step": 2289500
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.4329327744831385e-05,
+      "loss": 2.4244,
+      "step": 2290000
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.4328091635578154e-05,
+      "loss": 2.439,
+      "step": 2290500
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.432685304915207e-05,
+      "loss": 2.4081,
+      "step": 2291000
+    },
+    {
+      "epoch": 11.35,
+      "learning_rate": 4.432561446272598e-05,
+      "loss": 2.4084,
+      "step": 2291500
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.43243758762999e-05,
+      "loss": 2.4155,
+      "step": 2292000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.4323137289873815e-05,
+      "loss": 2.4206,
+      "step": 2292500
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.432189870344773e-05,
+      "loss": 2.4201,
+      "step": 2293000
+    },
+    {
+      "epoch": 11.36,
+      "learning_rate": 4.43206625941945e-05,
+      "loss": 2.4092,
+      "step": 2293500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.431942400776842e-05,
+      "loss": 2.4223,
+      "step": 2294000
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.4318185421342334e-05,
+      "loss": 2.4183,
+      "step": 2294500
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.431694683491625e-05,
+      "loss": 2.424,
+      "step": 2295000
+    },
+    {
+      "epoch": 11.37,
+      "learning_rate": 4.431570824849017e-05,
+      "loss": 2.4262,
+      "step": 2295500
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.4314469662064085e-05,
+      "loss": 2.4002,
+      "step": 2296000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.4313231075637995e-05,
+      "loss": 2.4274,
+      "step": 2296500
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.431199496638477e-05,
+      "loss": 2.4295,
+      "step": 2297000
+    },
+    {
+      "epoch": 11.38,
+      "learning_rate": 4.431075637995869e-05,
+      "loss": 2.4161,
+      "step": 2297500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4309517793532605e-05,
+      "loss": 2.4199,
+      "step": 2298000
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4308279207106515e-05,
+      "loss": 2.4241,
+      "step": 2298500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.430704062068043e-05,
+      "loss": 2.4267,
+      "step": 2299000
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.430580203425435e-05,
+      "loss": 2.4256,
+      "step": 2299500
+    },
+    {
+      "epoch": 11.39,
+      "learning_rate": 4.4304563447828265e-05,
+      "loss": 2.4038,
+      "step": 2300000
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.4303324861402176e-05,
+      "loss": 2.419,
+      "step": 2300500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.430208627497609e-05,
+      "loss": 2.4068,
+      "step": 2301000
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.430085016572287e-05,
+      "loss": 2.4023,
+      "step": 2301500
+    },
+    {
+      "epoch": 11.4,
+      "learning_rate": 4.4299611579296785e-05,
+      "loss": 2.4277,
+      "step": 2302000
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.42983729928707e-05,
+      "loss": 2.4221,
+      "step": 2302500
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.429713440644461e-05,
+      "loss": 2.429,
+      "step": 2303000
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.429589582001853e-05,
+      "loss": 2.429,
+      "step": 2303500
+    },
+    {
+      "epoch": 11.41,
+      "learning_rate": 4.4294657233592446e-05,
+      "loss": 2.4134,
+      "step": 2304000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.429341864716636e-05,
+      "loss": 2.4422,
+      "step": 2304500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.429218253791313e-05,
+      "loss": 2.438,
+      "step": 2305000
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.429094395148705e-05,
+      "loss": 2.4249,
+      "step": 2305500
+    },
+    {
+      "epoch": 11.42,
+      "learning_rate": 4.428970784223382e-05,
+      "loss": 2.4145,
+      "step": 2306000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.4288469255807734e-05,
+      "loss": 2.4202,
+      "step": 2306500
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.428723066938165e-05,
+      "loss": 2.4222,
+      "step": 2307000
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.428599208295557e-05,
+      "loss": 2.4162,
+      "step": 2307500
+    },
+    {
+      "epoch": 11.43,
+      "learning_rate": 4.428475597370234e-05,
+      "loss": 2.4277,
+      "step": 2308000
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.4283517387276254e-05,
+      "loss": 2.4503,
+      "step": 2308500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.4282281278023016e-05,
+      "loss": 2.4174,
+      "step": 2309000
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.428104269159693e-05,
+      "loss": 2.4125,
+      "step": 2309500
+    },
+    {
+      "epoch": 11.44,
+      "learning_rate": 4.427980410517085e-05,
+      "loss": 2.4267,
+      "step": 2310000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.427856551874477e-05,
+      "loss": 2.4294,
+      "step": 2310500
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.4277326932318684e-05,
+      "loss": 2.4163,
+      "step": 2311000
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.42760883458926e-05,
+      "loss": 2.406,
+      "step": 2311500
+    },
+    {
+      "epoch": 11.45,
+      "learning_rate": 4.427484975946652e-05,
+      "loss": 2.4035,
+      "step": 2312000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.4273611173040435e-05,
+      "loss": 2.4194,
+      "step": 2312500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.42723750637872e-05,
+      "loss": 2.4203,
+      "step": 2313000
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.427113647736112e-05,
+      "loss": 2.4323,
+      "step": 2313500
+    },
+    {
+      "epoch": 11.46,
+      "learning_rate": 4.426989789093504e-05,
+      "loss": 2.4078,
+      "step": 2314000
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.4268659304508954e-05,
+      "loss": 2.4321,
+      "step": 2314500
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.426742071808287e-05,
+      "loss": 2.4087,
+      "step": 2315000
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.426618213165679e-05,
+      "loss": 2.4277,
+      "step": 2315500
+    },
+    {
+      "epoch": 11.47,
+      "learning_rate": 4.4264943545230705e-05,
+      "loss": 2.41,
+      "step": 2316000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.426370495880462e-05,
+      "loss": 2.4286,
+      "step": 2316500
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.426246637237854e-05,
+      "loss": 2.4222,
+      "step": 2317000
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.4261227785952456e-05,
+      "loss": 2.4014,
+      "step": 2317500
+    },
+    {
+      "epoch": 11.48,
+      "learning_rate": 4.425998919952637e-05,
+      "loss": 2.4204,
+      "step": 2318000
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.425875061310028e-05,
+      "loss": 2.3936,
+      "step": 2318500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.42575120266742e-05,
+      "loss": 2.4184,
+      "step": 2319000
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.425627591742097e-05,
+      "loss": 2.4393,
+      "step": 2319500
+    },
+    {
+      "epoch": 11.49,
+      "learning_rate": 4.4255037330994885e-05,
+      "loss": 2.4238,
+      "step": 2320000
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.4253801221741654e-05,
+      "loss": 2.4157,
+      "step": 2320500
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.425256263531557e-05,
+      "loss": 2.4123,
+      "step": 2321000
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.425132404888949e-05,
+      "loss": 2.4241,
+      "step": 2321500
+    },
+    {
+      "epoch": 11.5,
+      "learning_rate": 4.425008793963625e-05,
+      "loss": 2.4004,
+      "step": 2322000
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.424884935321017e-05,
+      "loss": 2.4401,
+      "step": 2322500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.4247610766784084e-05,
+      "loss": 2.4317,
+      "step": 2323000
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.4246372180358e-05,
+      "loss": 2.4115,
+      "step": 2323500
+    },
+    {
+      "epoch": 11.51,
+      "learning_rate": 4.424513359393192e-05,
+      "loss": 2.4012,
+      "step": 2324000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.424389748467869e-05,
+      "loss": 2.4115,
+      "step": 2324500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.4242658898252604e-05,
+      "loss": 2.392,
+      "step": 2325000
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.424142031182652e-05,
+      "loss": 2.4307,
+      "step": 2325500
+    },
+    {
+      "epoch": 11.52,
+      "learning_rate": 4.424018172540044e-05,
+      "loss": 2.4094,
+      "step": 2326000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4238943138974354e-05,
+      "loss": 2.4216,
+      "step": 2326500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.423770455254827e-05,
+      "loss": 2.4106,
+      "step": 2327000
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.423646596612219e-05,
+      "loss": 2.4049,
+      "step": 2327500
+    },
+    {
+      "epoch": 11.53,
+      "learning_rate": 4.4235227379696105e-05,
+      "loss": 2.4113,
+      "step": 2328000
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423398879327002e-05,
+      "loss": 2.4132,
+      "step": 2328500
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423275020684394e-05,
+      "loss": 2.4204,
+      "step": 2329000
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.42315140975907e-05,
+      "loss": 2.4299,
+      "step": 2329500
+    },
+    {
+      "epoch": 11.54,
+      "learning_rate": 4.423027551116462e-05,
+      "loss": 2.419,
+      "step": 2330000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4229036924738535e-05,
+      "loss": 2.4446,
+      "step": 2330500
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.422779833831245e-05,
+      "loss": 2.4427,
+      "step": 2331000
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.422655975188637e-05,
+      "loss": 2.4376,
+      "step": 2331500
+    },
+    {
+      "epoch": 11.55,
+      "learning_rate": 4.4225321165460286e-05,
+      "loss": 2.4157,
+      "step": 2332000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.42240825790342e-05,
+      "loss": 2.4434,
+      "step": 2332500
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.422284399260812e-05,
+      "loss": 2.4163,
+      "step": 2333000
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4221605406182036e-05,
+      "loss": 2.4218,
+      "step": 2333500
+    },
+    {
+      "epoch": 11.56,
+      "learning_rate": 4.4220369296928805e-05,
+      "loss": 2.4099,
+      "step": 2334000
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.421913071050272e-05,
+      "loss": 2.4475,
+      "step": 2334500
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.421789460124949e-05,
+      "loss": 2.4169,
+      "step": 2335000
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.42166560148234e-05,
+      "loss": 2.4357,
+      "step": 2335500
+    },
+    {
+      "epoch": 11.57,
+      "learning_rate": 4.421541742839732e-05,
+      "loss": 2.4186,
+      "step": 2336000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.4214178841971235e-05,
+      "loss": 2.438,
+      "step": 2336500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.421294025554515e-05,
+      "loss": 2.4189,
+      "step": 2337000
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.421170414629192e-05,
+      "loss": 2.4401,
+      "step": 2337500
+    },
+    {
+      "epoch": 11.58,
+      "learning_rate": 4.421046555986584e-05,
+      "loss": 2.4355,
+      "step": 2338000
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.4209226973439755e-05,
+      "loss": 2.4278,
+      "step": 2338500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.420798838701367e-05,
+      "loss": 2.4345,
+      "step": 2339000
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.420674980058759e-05,
+      "loss": 2.4363,
+      "step": 2339500
+    },
+    {
+      "epoch": 11.59,
+      "learning_rate": 4.4205511214161505e-05,
+      "loss": 2.4035,
+      "step": 2340000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.4204275104908274e-05,
+      "loss": 2.4417,
+      "step": 2340500
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.420303651848219e-05,
+      "loss": 2.4219,
+      "step": 2341000
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.420179793205611e-05,
+      "loss": 2.4465,
+      "step": 2341500
+    },
+    {
+      "epoch": 11.6,
+      "learning_rate": 4.420056182280287e-05,
+      "loss": 2.4282,
+      "step": 2342000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.419932323637679e-05,
+      "loss": 2.4328,
+      "step": 2342500
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.4198084649950704e-05,
+      "loss": 2.4098,
+      "step": 2343000
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.419684606352462e-05,
+      "loss": 2.4214,
+      "step": 2343500
+    },
+    {
+      "epoch": 11.61,
+      "learning_rate": 4.419560747709854e-05,
+      "loss": 2.442,
+      "step": 2344000
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.4194368890672455e-05,
+      "loss": 2.4259,
+      "step": 2344500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.419313030424637e-05,
+      "loss": 2.4322,
+      "step": 2345000
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.419189171782029e-05,
+      "loss": 2.4194,
+      "step": 2345500
+    },
+    {
+      "epoch": 11.62,
+      "learning_rate": 4.4190653131394206e-05,
+      "loss": 2.4452,
+      "step": 2346000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418941454496812e-05,
+      "loss": 2.4208,
+      "step": 2346500
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418817595854204e-05,
+      "loss": 2.4285,
+      "step": 2347000
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418693984928881e-05,
+      "loss": 2.4065,
+      "step": 2347500
+    },
+    {
+      "epoch": 11.63,
+      "learning_rate": 4.418570374003557e-05,
+      "loss": 2.42,
+      "step": 2348000
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418446515360949e-05,
+      "loss": 2.4344,
+      "step": 2348500
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.4183226567183404e-05,
+      "loss": 2.4254,
+      "step": 2349000
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418198798075732e-05,
+      "loss": 2.4336,
+      "step": 2349500
+    },
+    {
+      "epoch": 11.64,
+      "learning_rate": 4.418074939433124e-05,
+      "loss": 2.4124,
+      "step": 2350000
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.4179510807905155e-05,
+      "loss": 2.4268,
+      "step": 2350500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.417827222147907e-05,
+      "loss": 2.4369,
+      "step": 2351000
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.417703363505299e-05,
+      "loss": 2.4431,
+      "step": 2351500
+    },
+    {
+      "epoch": 11.65,
+      "learning_rate": 4.4175795048626906e-05,
+      "loss": 2.4248,
+      "step": 2352000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.417455646220082e-05,
+      "loss": 2.3927,
+      "step": 2352500
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.417331787577474e-05,
+      "loss": 2.4374,
+      "step": 2353000
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.4172079289348656e-05,
+      "loss": 2.4195,
+      "step": 2353500
+    },
+    {
+      "epoch": 11.66,
+      "learning_rate": 4.417084070292257e-05,
+      "loss": 2.4295,
+      "step": 2354000
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.416960459366934e-05,
+      "loss": 2.4114,
+      "step": 2354500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.416836600724326e-05,
+      "loss": 2.4245,
+      "step": 2355000
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.416712742081717e-05,
+      "loss": 2.4333,
+      "step": 2355500
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.416589131156394e-05,
+      "loss": 2.4191,
+      "step": 2356000
+    },
+    {
+      "epoch": 11.67,
+      "learning_rate": 4.4164652725137855e-05,
+      "loss": 2.3961,
+      "step": 2356500
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4163416615884624e-05,
+      "loss": 2.4087,
+      "step": 2357000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.416217802945854e-05,
+      "loss": 2.4227,
+      "step": 2357500
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.416093944303246e-05,
+      "loss": 2.4177,
+      "step": 2358000
+    },
+    {
+      "epoch": 11.68,
+      "learning_rate": 4.4159700856606375e-05,
+      "loss": 2.4067,
+      "step": 2358500
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.415846227018029e-05,
+      "loss": 2.4215,
+      "step": 2359000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.4157226160927054e-05,
+      "loss": 2.4338,
+      "step": 2359500
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.415598757450097e-05,
+      "loss": 2.4406,
+      "step": 2360000
+    },
+    {
+      "epoch": 11.69,
+      "learning_rate": 4.415474898807489e-05,
+      "loss": 2.4358,
+      "step": 2360500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.4153510401648804e-05,
+      "loss": 2.4452,
+      "step": 2361000
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.415227181522272e-05,
+      "loss": 2.4173,
+      "step": 2361500
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.415103322879664e-05,
+      "loss": 2.4111,
+      "step": 2362000
+    },
+    {
+      "epoch": 11.7,
+      "learning_rate": 4.4149794642370555e-05,
+      "loss": 2.4227,
+      "step": 2362500
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.414855605594447e-05,
+      "loss": 2.4011,
+      "step": 2363000
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.414731994669124e-05,
+      "loss": 2.4455,
+      "step": 2363500
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.414608136026516e-05,
+      "loss": 2.4459,
+      "step": 2364000
+    },
+    {
+      "epoch": 11.71,
+      "learning_rate": 4.4144842773839075e-05,
+      "loss": 2.4196,
+      "step": 2364500
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.414360418741299e-05,
+      "loss": 2.4346,
+      "step": 2365000
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.414236560098691e-05,
+      "loss": 2.4307,
+      "step": 2365500
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.414113196890653e-05,
+      "loss": 2.4386,
+      "step": 2366000
+    },
+    {
+      "epoch": 11.72,
+      "learning_rate": 4.4139893382480446e-05,
+      "loss": 2.4333,
+      "step": 2366500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.413865479605436e-05,
+      "loss": 2.4199,
+      "step": 2367000
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.413741620962828e-05,
+      "loss": 2.4048,
+      "step": 2367500
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.41361776232022e-05,
+      "loss": 2.4399,
+      "step": 2368000
+    },
+    {
+      "epoch": 11.73,
+      "learning_rate": 4.4134939036776114e-05,
+      "loss": 2.4399,
+      "step": 2368500
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.4133700450350024e-05,
+      "loss": 2.4265,
+      "step": 2369000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.413246186392394e-05,
+      "loss": 2.4374,
+      "step": 2369500
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.413122327749786e-05,
+      "loss": 2.4099,
+      "step": 2370000
+    },
+    {
+      "epoch": 11.74,
+      "learning_rate": 4.4129984691071775e-05,
+      "loss": 2.3965,
+      "step": 2370500
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.4128748581818544e-05,
+      "loss": 2.4322,
+      "step": 2371000
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.412750999539246e-05,
+      "loss": 2.4613,
+      "step": 2371500
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.412627140896638e-05,
+      "loss": 2.4158,
+      "step": 2372000
+    },
+    {
+      "epoch": 11.75,
+      "learning_rate": 4.412503282254029e-05,
+      "loss": 2.4332,
+      "step": 2372500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.4123794236114205e-05,
+      "loss": 2.4335,
+      "step": 2373000
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.412255812686098e-05,
+      "loss": 2.4264,
+      "step": 2373500
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.41213195404349e-05,
+      "loss": 2.3871,
+      "step": 2374000
+    },
+    {
+      "epoch": 11.76,
+      "learning_rate": 4.4120080954008814e-05,
+      "loss": 2.4349,
+      "step": 2374500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.411884236758273e-05,
+      "loss": 2.4322,
+      "step": 2375000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.411760378115664e-05,
+      "loss": 2.4205,
+      "step": 2375500
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.411637014907626e-05,
+      "loss": 2.427,
+      "step": 2376000
+    },
+    {
+      "epoch": 11.77,
+      "learning_rate": 4.411513403982303e-05,
+      "loss": 2.4348,
+      "step": 2376500
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.411389545339695e-05,
+      "loss": 2.4465,
+      "step": 2377000
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.4112656866970864e-05,
+      "loss": 2.4235,
+      "step": 2377500
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.411141828054478e-05,
+      "loss": 2.4328,
+      "step": 2378000
+    },
+    {
+      "epoch": 11.78,
+      "learning_rate": 4.41101796941187e-05,
+      "loss": 2.3982,
+      "step": 2378500
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.4108941107692615e-05,
+      "loss": 2.4139,
+      "step": 2379000
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.410770252126653e-05,
+      "loss": 2.4416,
+      "step": 2379500
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.410646393484045e-05,
+      "loss": 2.4149,
+      "step": 2380000
+    },
+    {
+      "epoch": 11.79,
+      "learning_rate": 4.4105225348414366e-05,
+      "loss": 2.4093,
+      "step": 2380500
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.410398676198828e-05,
+      "loss": 2.4322,
+      "step": 2381000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.41027481755622e-05,
+      "loss": 2.4391,
+      "step": 2381500
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.410150958913612e-05,
+      "loss": 2.4264,
+      "step": 2382000
+    },
+    {
+      "epoch": 11.8,
+      "learning_rate": 4.4100271002710034e-05,
+      "loss": 2.4302,
+      "step": 2382500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.409903241628395e-05,
+      "loss": 2.4337,
+      "step": 2383000
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.409779630703071e-05,
+      "loss": 2.404,
+      "step": 2383500
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.409655772060463e-05,
+      "loss": 2.4127,
+      "step": 2384000
+    },
+    {
+      "epoch": 11.81,
+      "learning_rate": 4.4095319134178547e-05,
+      "loss": 2.4261,
+      "step": 2384500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.4094080547752463e-05,
+      "loss": 2.4263,
+      "step": 2385000
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.409284196132638e-05,
+      "loss": 2.4344,
+      "step": 2385500
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.4091608329246e-05,
+      "loss": 2.4243,
+      "step": 2386000
+    },
+    {
+      "epoch": 11.82,
+      "learning_rate": 4.409036974281992e-05,
+      "loss": 2.4161,
+      "step": 2386500
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.408913115639383e-05,
+      "loss": 2.4354,
+      "step": 2387000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.4087895047140604e-05,
+      "loss": 2.398,
+      "step": 2387500
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.408665646071452e-05,
+      "loss": 2.4091,
+      "step": 2388000
+    },
+    {
+      "epoch": 11.83,
+      "learning_rate": 4.408541787428843e-05,
+      "loss": 2.4046,
+      "step": 2388500
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.408417928786235e-05,
+      "loss": 2.4116,
+      "step": 2389000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.4082940701436265e-05,
+      "loss": 2.4345,
+      "step": 2389500
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.408170211501018e-05,
+      "loss": 2.4339,
+      "step": 2390000
+    },
+    {
+      "epoch": 11.84,
+      "learning_rate": 4.40804635285841e-05,
+      "loss": 2.4014,
+      "step": 2390500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.4079224942158015e-05,
+      "loss": 2.4364,
+      "step": 2391000
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.407798635573193e-05,
+      "loss": 2.4408,
+      "step": 2391500
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.407674776930585e-05,
+      "loss": 2.4218,
+      "step": 2392000
+    },
+    {
+      "epoch": 11.85,
+      "learning_rate": 4.4075509182879766e-05,
+      "loss": 2.4203,
+      "step": 2392500
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.407427059645368e-05,
+      "loss": 2.4449,
+      "step": 2393000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.40730320100276e-05,
+      "loss": 2.4359,
+      "step": 2393500
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.407179342360152e-05,
+      "loss": 2.4611,
+      "step": 2394000
+    },
+    {
+      "epoch": 11.86,
+      "learning_rate": 4.4070554837175434e-05,
+      "loss": 2.4342,
+      "step": 2394500
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.406931625074935e-05,
+      "loss": 2.4223,
+      "step": 2395000
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.406807766432327e-05,
+      "loss": 2.4426,
+      "step": 2395500
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.4066839077897185e-05,
+      "loss": 2.4305,
+      "step": 2396000
+    },
+    {
+      "epoch": 11.87,
+      "learning_rate": 4.406560296864395e-05,
+      "loss": 2.4321,
+      "step": 2396500
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.4064366859390716e-05,
+      "loss": 2.4089,
+      "step": 2397000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.4063133227310336e-05,
+      "loss": 2.414,
+      "step": 2397500
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.406189464088425e-05,
+      "loss": 2.4312,
+      "step": 2398000
+    },
+    {
+      "epoch": 11.88,
+      "learning_rate": 4.406065605445817e-05,
+      "loss": 2.405,
+      "step": 2398500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.405941746803209e-05,
+      "loss": 2.4159,
+      "step": 2399000
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.4058178881606004e-05,
+      "loss": 2.417,
+      "step": 2399500
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.405694277235277e-05,
+      "loss": 2.3933,
+      "step": 2400000
+    },
+    {
+      "epoch": 11.89,
+      "learning_rate": 4.405570418592669e-05,
+      "loss": 2.4296,
+      "step": 2400500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.405446559950061e-05,
+      "loss": 2.4389,
+      "step": 2401000
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.4053227013074524e-05,
+      "loss": 2.4468,
+      "step": 2401500
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.405198842664844e-05,
+      "loss": 2.4335,
+      "step": 2402000
+    },
+    {
+      "epoch": 11.9,
+      "learning_rate": 4.405074984022236e-05,
+      "loss": 2.4179,
+      "step": 2402500
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.4049511253796274e-05,
+      "loss": 2.4189,
+      "step": 2403000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.404827266737019e-05,
+      "loss": 2.4311,
+      "step": 2403500
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.40470340809441e-05,
+      "loss": 2.4236,
+      "step": 2404000
+    },
+    {
+      "epoch": 11.91,
+      "learning_rate": 4.404579549451802e-05,
+      "loss": 2.4249,
+      "step": 2404500
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.4044556908091935e-05,
+      "loss": 2.4524,
+      "step": 2405000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.404331832166585e-05,
+      "loss": 2.4291,
+      "step": 2405500
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.404207973523977e-05,
+      "loss": 2.4173,
+      "step": 2406000
+    },
+    {
+      "epoch": 11.92,
+      "learning_rate": 4.4040841148813686e-05,
+      "loss": 2.448,
+      "step": 2406500
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.4039602562387596e-05,
+      "loss": 2.4346,
+      "step": 2407000
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.403836397596151e-05,
+      "loss": 2.4152,
+      "step": 2407500
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.403712538953543e-05,
+      "loss": 2.4002,
+      "step": 2408000
+    },
+    {
+      "epoch": 11.93,
+      "learning_rate": 4.40358892802822e-05,
+      "loss": 2.419,
+      "step": 2408500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4034653171028975e-05,
+      "loss": 2.4215,
+      "step": 2409000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.403341458460289e-05,
+      "loss": 2.4047,
+      "step": 2409500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.4032178475349653e-05,
+      "loss": 2.4267,
+      "step": 2410000
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.403093988892357e-05,
+      "loss": 2.4454,
+      "step": 2410500
+    },
+    {
+      "epoch": 11.94,
+      "learning_rate": 4.402970130249749e-05,
+      "loss": 2.3954,
+      "step": 2411000
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.4028462716071404e-05,
+      "loss": 2.4412,
+      "step": 2411500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.402722412964532e-05,
+      "loss": 2.4209,
+      "step": 2412000
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.402598554321924e-05,
+      "loss": 2.4415,
+      "step": 2412500
+    },
+    {
+      "epoch": 11.95,
+      "learning_rate": 4.4024746956793155e-05,
+      "loss": 2.408,
+      "step": 2413000
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.4023508370367065e-05,
+      "loss": 2.4469,
+      "step": 2413500
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.402226978394098e-05,
+      "loss": 2.4386,
+      "step": 2414000
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.402103367468776e-05,
+      "loss": 2.4394,
+      "step": 2414500
+    },
+    {
+      "epoch": 11.96,
+      "learning_rate": 4.401979756543452e-05,
+      "loss": 2.426,
+      "step": 2415000
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401855897900844e-05,
+      "loss": 2.4471,
+      "step": 2415500
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.4017320392582354e-05,
+      "loss": 2.4293,
+      "step": 2416000
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401608180615627e-05,
+      "loss": 2.4297,
+      "step": 2416500
+    },
+    {
+      "epoch": 11.97,
+      "learning_rate": 4.401484321973019e-05,
+      "loss": 2.4321,
+      "step": 2417000
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.4013604633304104e-05,
+      "loss": 2.4261,
+      "step": 2417500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.401236604687802e-05,
+      "loss": 2.4338,
+      "step": 2418000
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.401112746045194e-05,
+      "loss": 2.4304,
+      "step": 2418500
+    },
+    {
+      "epoch": 11.98,
+      "learning_rate": 4.400989135119871e-05,
+      "loss": 2.3984,
+      "step": 2419000
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.4008652764772624e-05,
+      "loss": 2.437,
+      "step": 2419500
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.400741665551939e-05,
+      "loss": 2.4412,
+      "step": 2420000
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.400617806909331e-05,
+      "loss": 2.4605,
+      "step": 2420500
+    },
+    {
+      "epoch": 11.99,
+      "learning_rate": 4.400493948266723e-05,
+      "loss": 2.4154,
+      "step": 2421000
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.400370089624114e-05,
+      "loss": 2.4555,
+      "step": 2421500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.4002462309815054e-05,
+      "loss": 2.4047,
+      "step": 2422000
+    },
+    {
+      "epoch": 12.0,
+      "eval_accuracy": 0.6447791117813911,
+      "eval_accuracy_mlm": 0.5983567225929853,
+      "eval_accuracy_nsp": 0.8637231868653391,
+      "eval_loss": 2.398054361343384,
+      "eval_runtime": 146.0821,
+      "eval_samples_per_second": 1745.314,
+      "eval_steps_per_second": 72.726,
+      "step": 2422116
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.400122372338897e-05,
+      "loss": 2.3898,
+      "step": 2422500
+    },
+    {
+      "epoch": 12.0,
+      "learning_rate": 4.399998513696289e-05,
+      "loss": 2.4349,
+      "step": 2423000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.3998746550536804e-05,
+      "loss": 2.3875,
+      "step": 2423500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.399750796411072e-05,
+      "loss": 2.3812,
+      "step": 2424000
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.399626937768464e-05,
+      "loss": 2.3915,
+      "step": 2424500
+    },
+    {
+      "epoch": 12.01,
+      "learning_rate": 4.3995030791258555e-05,
+      "loss": 2.4071,
+      "step": 2425000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.399379220483247e-05,
+      "loss": 2.3903,
+      "step": 2425500
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.399255609557924e-05,
+      "loss": 2.3863,
+      "step": 2426000
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.399131998632601e-05,
+      "loss": 2.3889,
+      "step": 2426500
+    },
+    {
+      "epoch": 12.02,
+      "learning_rate": 4.399008139989993e-05,
+      "loss": 2.3892,
+      "step": 2427000
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.3988842813473844e-05,
+      "loss": 2.4049,
+      "step": 2427500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.398760422704776e-05,
+      "loss": 2.3953,
+      "step": 2428000
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.398636564062167e-05,
+      "loss": 2.3994,
+      "step": 2428500
+    },
+    {
+      "epoch": 12.03,
+      "learning_rate": 4.398512705419559e-05,
+      "loss": 2.3911,
+      "step": 2429000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.3983888467769505e-05,
+      "loss": 2.4036,
+      "step": 2429500
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.3982652358516273e-05,
+      "loss": 2.3935,
+      "step": 2430000
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.398141624926304e-05,
+      "loss": 2.4119,
+      "step": 2430500
+    },
+    {
+      "epoch": 12.04,
+      "learning_rate": 4.398018014000981e-05,
+      "loss": 2.4128,
+      "step": 2431000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.397894155358373e-05,
+      "loss": 2.3834,
+      "step": 2431500
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.3977702967157645e-05,
+      "loss": 2.402,
+      "step": 2432000
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.397646438073156e-05,
+      "loss": 2.3786,
+      "step": 2432500
+    },
+    {
+      "epoch": 12.05,
+      "learning_rate": 4.397522579430547e-05,
+      "loss": 2.3985,
+      "step": 2433000
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397398720787939e-05,
+      "loss": 2.3975,
+      "step": 2433500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.3972748621453306e-05,
+      "loss": 2.395,
+      "step": 2434000
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397151251220008e-05,
+      "loss": 2.3914,
+      "step": 2434500
+    },
+    {
+      "epoch": 12.06,
+      "learning_rate": 4.397027640294685e-05,
+      "loss": 2.4154,
+      "step": 2435000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.396903781652077e-05,
+      "loss": 2.4228,
+      "step": 2435500
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.396779923009468e-05,
+      "loss": 2.4234,
+      "step": 2436000
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.3966560643668594e-05,
+      "loss": 2.4103,
+      "step": 2436500
+    },
+    {
+      "epoch": 12.07,
+      "learning_rate": 4.396532205724251e-05,
+      "loss": 2.4163,
+      "step": 2437000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.396408347081643e-05,
+      "loss": 2.408,
+      "step": 2437500
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.3962844884390345e-05,
+      "loss": 2.3927,
+      "step": 2438000
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.396160629796426e-05,
+      "loss": 2.3937,
+      "step": 2438500
+    },
+    {
+      "epoch": 12.08,
+      "learning_rate": 4.396036771153817e-05,
+      "loss": 2.3874,
+      "step": 2439000
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.395912912511209e-05,
+      "loss": 2.4065,
+      "step": 2439500
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.3957890538686006e-05,
+      "loss": 2.3979,
+      "step": 2440000
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.395665442943278e-05,
+      "loss": 2.3901,
+      "step": 2440500
+    },
+    {
+      "epoch": 12.09,
+      "learning_rate": 4.39554158430067e-05,
+      "loss": 2.4009,
+      "step": 2441000
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.3954177256580615e-05,
+      "loss": 2.396,
+      "step": 2441500
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.3952941147327384e-05,
+      "loss": 2.4246,
+      "step": 2442000
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.39517025609013e-05,
+      "loss": 2.4031,
+      "step": 2442500
+    },
+    {
+      "epoch": 12.1,
+      "learning_rate": 4.395046397447521e-05,
+      "loss": 2.4032,
+      "step": 2443000
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.394922538804913e-05,
+      "loss": 2.3833,
+      "step": 2443500
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.3947986801623045e-05,
+      "loss": 2.3917,
+      "step": 2444000
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.394674821519696e-05,
+      "loss": 2.4239,
+      "step": 2444500
+    },
+    {
+      "epoch": 12.11,
+      "learning_rate": 4.394550962877088e-05,
+      "loss": 2.4095,
+      "step": 2445000
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394427104234479e-05,
+      "loss": 2.4093,
+      "step": 2445500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.3943032455918706e-05,
+      "loss": 2.4138,
+      "step": 2446000
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394179386949262e-05,
+      "loss": 2.3883,
+      "step": 2446500
+    },
+    {
+      "epoch": 12.12,
+      "learning_rate": 4.394055528306654e-05,
+      "loss": 2.3971,
+      "step": 2447000
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.393931669664046e-05,
+      "loss": 2.4231,
+      "step": 2447500
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.3938078110214374e-05,
+      "loss": 2.4131,
+      "step": 2448000
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.393683952378829e-05,
+      "loss": 2.4189,
+      "step": 2448500
+    },
+    {
+      "epoch": 12.13,
+      "learning_rate": 4.393560093736221e-05,
+      "loss": 2.3864,
+      "step": 2449000
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.3934362350936125e-05,
+      "loss": 2.4134,
+      "step": 2449500
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.393312376451004e-05,
+      "loss": 2.3885,
+      "step": 2450000
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.393188517808396e-05,
+      "loss": 2.4138,
+      "step": 2450500
+    },
+    {
+      "epoch": 12.14,
+      "learning_rate": 4.393064906883073e-05,
+      "loss": 2.42,
+      "step": 2451000
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.3929410482404644e-05,
+      "loss": 2.4087,
+      "step": 2451500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.392817189597856e-05,
+      "loss": 2.4067,
+      "step": 2452000
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.392693330955248e-05,
+      "loss": 2.4229,
+      "step": 2452500
+    },
+    {
+      "epoch": 12.15,
+      "learning_rate": 4.3925694723126395e-05,
+      "loss": 2.3739,
+      "step": 2453000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392445613670031e-05,
+      "loss": 2.3939,
+      "step": 2453500
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.3923220027447074e-05,
+      "loss": 2.4108,
+      "step": 2454000
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392198391819384e-05,
+      "loss": 2.3995,
+      "step": 2454500
+    },
+    {
+      "epoch": 12.16,
+      "learning_rate": 4.392074533176776e-05,
+      "loss": 2.396,
+      "step": 2455000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.3919506745341677e-05,
+      "loss": 2.4091,
+      "step": 2455500
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.3918268158915594e-05,
+      "loss": 2.4122,
+      "step": 2456000
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.391702957248951e-05,
+      "loss": 2.3943,
+      "step": 2456500
+    },
+    {
+      "epoch": 12.17,
+      "learning_rate": 4.391579098606343e-05,
+      "loss": 2.4066,
+      "step": 2457000
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.3914552399637344e-05,
+      "loss": 2.4244,
+      "step": 2457500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.391331381321126e-05,
+      "loss": 2.386,
+      "step": 2458000
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.391207522678518e-05,
+      "loss": 2.4081,
+      "step": 2458500
+    },
+    {
+      "epoch": 12.18,
+      "learning_rate": 4.3910836640359095e-05,
+      "loss": 2.376,
+      "step": 2459000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.390959805393301e-05,
+      "loss": 2.4178,
+      "step": 2459500
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.3908361944679774e-05,
+      "loss": 2.4064,
+      "step": 2460000
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.390712335825369e-05,
+      "loss": 2.4019,
+      "step": 2460500
+    },
+    {
+      "epoch": 12.19,
+      "learning_rate": 4.390588477182761e-05,
+      "loss": 2.3993,
+      "step": 2461000
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.3904646185401525e-05,
+      "loss": 2.3876,
+      "step": 2461500
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.390340759897544e-05,
+      "loss": 2.388,
+      "step": 2462000
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.390216901254936e-05,
+      "loss": 2.4285,
+      "step": 2462500
+    },
+    {
+      "epoch": 12.2,
+      "learning_rate": 4.390093290329613e-05,
+      "loss": 2.4094,
+      "step": 2463000
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.3899696794042896e-05,
+      "loss": 2.3865,
+      "step": 2463500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389845820761681e-05,
+      "loss": 2.4204,
+      "step": 2464000
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389721962119073e-05,
+      "loss": 2.4236,
+      "step": 2464500
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.389598103476465e-05,
+      "loss": 2.3961,
+      "step": 2465000
+    },
+    {
+      "epoch": 12.21,
+      "learning_rate": 4.3894744925511416e-05,
+      "loss": 2.397,
+      "step": 2465500
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.389350633908533e-05,
+      "loss": 2.3895,
+      "step": 2466000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.389226775265925e-05,
+      "loss": 2.4102,
+      "step": 2466500
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.389102916623316e-05,
+      "loss": 2.405,
+      "step": 2467000
+    },
+    {
+      "epoch": 12.22,
+      "learning_rate": 4.388979057980708e-05,
+      "loss": 2.4291,
+      "step": 2467500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.3888551993380994e-05,
+      "loss": 2.4213,
+      "step": 2468000
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.388731340695491e-05,
+      "loss": 2.4009,
+      "step": 2468500
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.388607482052883e-05,
+      "loss": 2.4047,
+      "step": 2469000
+    },
+    {
+      "epoch": 12.23,
+      "learning_rate": 4.3884836234102745e-05,
+      "loss": 2.385,
+      "step": 2469500
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.388360012484951e-05,
+      "loss": 2.4111,
+      "step": 2470000
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.388236153842343e-05,
+      "loss": 2.4104,
+      "step": 2470500
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.388112295199735e-05,
+      "loss": 2.4294,
+      "step": 2471000
+    },
+    {
+      "epoch": 12.24,
+      "learning_rate": 4.3879884365571264e-05,
+      "loss": 2.4056,
+      "step": 2471500
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387864577914518e-05,
+      "loss": 2.396,
+      "step": 2472000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387740719271909e-05,
+      "loss": 2.3991,
+      "step": 2472500
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387617108346587e-05,
+      "loss": 2.3729,
+      "step": 2473000
+    },
+    {
+      "epoch": 12.25,
+      "learning_rate": 4.387493249703978e-05,
+      "loss": 2.4009,
+      "step": 2473500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.3873693910613694e-05,
+      "loss": 2.4272,
+      "step": 2474000
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.387245532418761e-05,
+      "loss": 2.4015,
+      "step": 2474500
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.387121673776153e-05,
+      "loss": 2.4204,
+      "step": 2475000
+    },
+    {
+      "epoch": 12.26,
+      "learning_rate": 4.3869978151335445e-05,
+      "loss": 2.4145,
+      "step": 2475500
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.386873956490936e-05,
+      "loss": 2.4175,
+      "step": 2476000
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.386750097848328e-05,
+      "loss": 2.3936,
+      "step": 2476500
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.3866262392057195e-05,
+      "loss": 2.382,
+      "step": 2477000
+    },
+    {
+      "epoch": 12.27,
+      "learning_rate": 4.3865026282803964e-05,
+      "loss": 2.4075,
+      "step": 2477500
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.386378769637788e-05,
+      "loss": 2.3898,
+      "step": 2478000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.38625491099518e-05,
+      "loss": 2.4162,
+      "step": 2478500
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.386131052352571e-05,
+      "loss": 2.4163,
+      "step": 2479000
+    },
+    {
+      "epoch": 12.28,
+      "learning_rate": 4.3860071937099625e-05,
+      "loss": 2.4165,
+      "step": 2479500
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.385883335067354e-05,
+      "loss": 2.4039,
+      "step": 2480000
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.385759476424746e-05,
+      "loss": 2.4346,
+      "step": 2480500
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.385635865499423e-05,
+      "loss": 2.427,
+      "step": 2481000
+    },
+    {
+      "epoch": 12.29,
+      "learning_rate": 4.3855120068568145e-05,
+      "loss": 2.3986,
+      "step": 2481500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.385388148214206e-05,
+      "loss": 2.4292,
+      "step": 2482000
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.385264537288884e-05,
+      "loss": 2.4181,
+      "step": 2482500
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.385140678646275e-05,
+      "loss": 2.4078,
+      "step": 2483000
+    },
+    {
+      "epoch": 12.3,
+      "learning_rate": 4.3850168200036664e-05,
+      "loss": 2.426,
+      "step": 2483500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.384892961361058e-05,
+      "loss": 2.4333,
+      "step": 2484000
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.38476910271845e-05,
+      "loss": 2.4103,
+      "step": 2484500
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.3846452440758415e-05,
+      "loss": 2.418,
+      "step": 2485000
+    },
+    {
+      "epoch": 12.31,
+      "learning_rate": 4.384521385433233e-05,
+      "loss": 2.399,
+      "step": 2485500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.384397526790624e-05,
+      "loss": 2.408,
+      "step": 2486000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.384273668148016e-05,
+      "loss": 2.3786,
+      "step": 2486500
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.3841498095054076e-05,
+      "loss": 2.434,
+      "step": 2487000
+    },
+    {
+      "epoch": 12.32,
+      "learning_rate": 4.384025950862799e-05,
+      "loss": 2.4179,
+      "step": 2487500
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383902339937476e-05,
+      "loss": 2.4173,
+      "step": 2488000
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383778729012154e-05,
+      "loss": 2.4215,
+      "step": 2488500
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383655365804115e-05,
+      "loss": 2.4273,
+      "step": 2489000
+    },
+    {
+      "epoch": 12.33,
+      "learning_rate": 4.383531507161507e-05,
+      "loss": 2.4262,
+      "step": 2489500
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.3834076485188985e-05,
+      "loss": 2.4236,
+      "step": 2490000
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.38328378987629e-05,
+      "loss": 2.3886,
+      "step": 2490500
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.383159931233682e-05,
+      "loss": 2.4172,
+      "step": 2491000
+    },
+    {
+      "epoch": 12.34,
+      "learning_rate": 4.3830360725910736e-05,
+      "loss": 2.4084,
+      "step": 2491500
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.382912213948465e-05,
+      "loss": 2.4236,
+      "step": 2492000
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.382788355305857e-05,
+      "loss": 2.4233,
+      "step": 2492500
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.382664496663249e-05,
+      "loss": 2.4178,
+      "step": 2493000
+    },
+    {
+      "epoch": 12.35,
+      "learning_rate": 4.3825406380206404e-05,
+      "loss": 2.3815,
+      "step": 2493500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.382416779378032e-05,
+      "loss": 2.4237,
+      "step": 2494000
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.382292920735424e-05,
+      "loss": 2.397,
+      "step": 2494500
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.3821690620928154e-05,
+      "loss": 2.4008,
+      "step": 2495000
+    },
+    {
+      "epoch": 12.36,
+      "learning_rate": 4.3820452034502065e-05,
+      "loss": 2.4192,
+      "step": 2495500
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.3819215925248833e-05,
+      "loss": 2.3788,
+      "step": 2496000
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.381797733882275e-05,
+      "loss": 2.3979,
+      "step": 2496500
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.381673875239667e-05,
+      "loss": 2.3989,
+      "step": 2497000
+    },
+    {
+      "epoch": 12.37,
+      "learning_rate": 4.3815500165970584e-05,
+      "loss": 2.4017,
+      "step": 2497500
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.38142615795445e-05,
+      "loss": 2.4174,
+      "step": 2498000
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.381302299311841e-05,
+      "loss": 2.4145,
+      "step": 2498500
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.381178440669233e-05,
+      "loss": 2.4302,
+      "step": 2499000
+    },
+    {
+      "epoch": 12.38,
+      "learning_rate": 4.3810545820266245e-05,
+      "loss": 2.4266,
+      "step": 2499500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.380930971101302e-05,
+      "loss": 2.3984,
+      "step": 2500000
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.380807360175978e-05,
+      "loss": 2.4286,
+      "step": 2500500
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.38068350153337e-05,
+      "loss": 2.3976,
+      "step": 2501000
+    },
+    {
+      "epoch": 12.39,
+      "learning_rate": 4.380559890608047e-05,
+      "loss": 2.4343,
+      "step": 2501500
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.3804360319654385e-05,
+      "loss": 2.4084,
+      "step": 2502000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.38031217332283e-05,
+      "loss": 2.4167,
+      "step": 2502500
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.380188314680222e-05,
+      "loss": 2.4184,
+      "step": 2503000
+    },
+    {
+      "epoch": 12.4,
+      "learning_rate": 4.3800644560376136e-05,
+      "loss": 2.424,
+      "step": 2503500
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3799408451122905e-05,
+      "loss": 2.4085,
+      "step": 2504000
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.379816986469682e-05,
+      "loss": 2.4307,
+      "step": 2504500
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.379693127827074e-05,
+      "loss": 2.4117,
+      "step": 2505000
+    },
+    {
+      "epoch": 12.41,
+      "learning_rate": 4.3795692691844656e-05,
+      "loss": 2.3954,
+      "step": 2505500
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.379445658259142e-05,
+      "loss": 2.4061,
+      "step": 2506000
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.3793217996165335e-05,
+      "loss": 2.4096,
+      "step": 2506500
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.379197940973925e-05,
+      "loss": 2.3931,
+      "step": 2507000
+    },
+    {
+      "epoch": 12.42,
+      "learning_rate": 4.379074082331317e-05,
+      "loss": 2.4206,
+      "step": 2507500
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3789502236887086e-05,
+      "loss": 2.4262,
+      "step": 2508000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3788263650461e-05,
+      "loss": 2.3758,
+      "step": 2508500
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.378702506403492e-05,
+      "loss": 2.3879,
+      "step": 2509000
+    },
+    {
+      "epoch": 12.43,
+      "learning_rate": 4.3785786477608836e-05,
+      "loss": 2.4278,
+      "step": 2509500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3784550368355605e-05,
+      "loss": 2.4015,
+      "step": 2510000
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.378331178192952e-05,
+      "loss": 2.4261,
+      "step": 2510500
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.378207319550344e-05,
+      "loss": 2.4078,
+      "step": 2511000
+    },
+    {
+      "epoch": 12.44,
+      "learning_rate": 4.3780834609077356e-05,
+      "loss": 2.3968,
+      "step": 2511500
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.377959602265127e-05,
+      "loss": 2.3812,
+      "step": 2512000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.377835743622519e-05,
+      "loss": 2.3986,
+      "step": 2512500
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.377711884979911e-05,
+      "loss": 2.4222,
+      "step": 2513000
+    },
+    {
+      "epoch": 12.45,
+      "learning_rate": 4.377588026337302e-05,
+      "loss": 2.4022,
+      "step": 2513500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3774641676946934e-05,
+      "loss": 2.4151,
+      "step": 2514000
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.377340309052085e-05,
+      "loss": 2.4226,
+      "step": 2514500
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.377216450409477e-05,
+      "loss": 2.3848,
+      "step": 2515000
+    },
+    {
+      "epoch": 12.46,
+      "learning_rate": 4.3770925917668685e-05,
+      "loss": 2.4285,
+      "step": 2515500
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.37696873312426e-05,
+      "loss": 2.4243,
+      "step": 2516000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.376845122198937e-05,
+      "loss": 2.4008,
+      "step": 2516500
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.376721263556329e-05,
+      "loss": 2.4064,
+      "step": 2517000
+    },
+    {
+      "epoch": 12.47,
+      "learning_rate": 4.3765974049137204e-05,
+      "loss": 2.4022,
+      "step": 2517500
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376473793988397e-05,
+      "loss": 2.3923,
+      "step": 2518000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376349935345789e-05,
+      "loss": 2.4132,
+      "step": 2518500
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.376226076703181e-05,
+      "loss": 2.4178,
+      "step": 2519000
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.3761022180605724e-05,
+      "loss": 2.4013,
+      "step": 2519500
+    },
+    {
+      "epoch": 12.48,
+      "learning_rate": 4.375978359417964e-05,
+      "loss": 2.4077,
+      "step": 2520000
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.375854500775355e-05,
+      "loss": 2.4075,
+      "step": 2520500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.375730642132747e-05,
+      "loss": 2.4111,
+      "step": 2521000
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.375607031207424e-05,
+      "loss": 2.3895,
+      "step": 2521500
+    },
+    {
+      "epoch": 12.49,
+      "learning_rate": 4.3754831725648154e-05,
+      "loss": 2.3763,
+      "step": 2522000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.375359313922207e-05,
+      "loss": 2.4228,
+      "step": 2522500
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.375235455279599e-05,
+      "loss": 2.3938,
+      "step": 2523000
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.3751115966369904e-05,
+      "loss": 2.404,
+      "step": 2523500
+    },
+    {
+      "epoch": 12.5,
+      "learning_rate": 4.374987985711667e-05,
+      "loss": 2.4042,
+      "step": 2524000
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374864127069059e-05,
+      "loss": 2.4155,
+      "step": 2524500
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374740268426451e-05,
+      "loss": 2.4128,
+      "step": 2525000
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.3746164097838424e-05,
+      "loss": 2.3995,
+      "step": 2525500
+    },
+    {
+      "epoch": 12.51,
+      "learning_rate": 4.374492551141234e-05,
+      "loss": 2.3965,
+      "step": 2526000
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.374368692498626e-05,
+      "loss": 2.426,
+      "step": 2526500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3742448338560175e-05,
+      "loss": 2.4275,
+      "step": 2527000
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3741209752134085e-05,
+      "loss": 2.407,
+      "step": 2527500
+    },
+    {
+      "epoch": 12.52,
+      "learning_rate": 4.3739973642880854e-05,
+      "loss": 2.376,
+      "step": 2528000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.373873505645477e-05,
+      "loss": 2.4029,
+      "step": 2528500
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.373749647002869e-05,
+      "loss": 2.4195,
+      "step": 2529000
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.3736257883602604e-05,
+      "loss": 2.3921,
+      "step": 2529500
+    },
+    {
+      "epoch": 12.53,
+      "learning_rate": 4.373501929717652e-05,
+      "loss": 2.4134,
+      "step": 2530000
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3733785665096135e-05,
+      "loss": 2.4015,
+      "step": 2530500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.373254707867005e-05,
+      "loss": 2.4044,
+      "step": 2531000
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.373130849224397e-05,
+      "loss": 2.4226,
+      "step": 2531500
+    },
+    {
+      "epoch": 12.54,
+      "learning_rate": 4.3730069905817886e-05,
+      "loss": 2.4238,
+      "step": 2532000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.37288313193918e-05,
+      "loss": 2.3989,
+      "step": 2532500
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.372759273296572e-05,
+      "loss": 2.4198,
+      "step": 2533000
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.372635414653964e-05,
+      "loss": 2.3914,
+      "step": 2533500
+    },
+    {
+      "epoch": 12.55,
+      "learning_rate": 4.3725115560113554e-05,
+      "loss": 2.4127,
+      "step": 2534000
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372387697368747e-05,
+      "loss": 2.4067,
+      "step": 2534500
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372264086443424e-05,
+      "loss": 2.4096,
+      "step": 2535000
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.3721402278008156e-05,
+      "loss": 2.392,
+      "step": 2535500
+    },
+    {
+      "epoch": 12.56,
+      "learning_rate": 4.372016369158207e-05,
+      "loss": 2.4025,
+      "step": 2536000
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.371892510515599e-05,
+      "loss": 2.3878,
+      "step": 2536500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.371768651872991e-05,
+      "loss": 2.428,
+      "step": 2537000
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.3716447932303824e-05,
+      "loss": 2.3988,
+      "step": 2537500
+    },
+    {
+      "epoch": 12.57,
+      "learning_rate": 4.371520934587774e-05,
+      "loss": 2.4164,
+      "step": 2538000
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.37139732366245e-05,
+      "loss": 2.4146,
+      "step": 2538500
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.371273465019842e-05,
+      "loss": 2.4205,
+      "step": 2539000
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3711498540945196e-05,
+      "loss": 2.4266,
+      "step": 2539500
+    },
+    {
+      "epoch": 12.58,
+      "learning_rate": 4.3710259954519106e-05,
+      "loss": 2.405,
+      "step": 2540000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.370902384526588e-05,
+      "loss": 2.4024,
+      "step": 2540500
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.37077852588398e-05,
+      "loss": 2.4214,
+      "step": 2541000
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.370654667241371e-05,
+      "loss": 2.4359,
+      "step": 2541500
+    },
+    {
+      "epoch": 12.59,
+      "learning_rate": 4.3705308085987625e-05,
+      "loss": 2.4429,
+      "step": 2542000
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.370406949956154e-05,
+      "loss": 2.4162,
+      "step": 2542500
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.370283339030831e-05,
+      "loss": 2.4065,
+      "step": 2543000
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.370159480388223e-05,
+      "loss": 2.4123,
+      "step": 2543500
+    },
+    {
+      "epoch": 12.6,
+      "learning_rate": 4.3700356217456145e-05,
+      "loss": 2.3914,
+      "step": 2544000
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.369911763103006e-05,
+      "loss": 2.4199,
+      "step": 2544500
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.369787904460398e-05,
+      "loss": 2.4165,
+      "step": 2545000
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3696640458177896e-05,
+      "loss": 2.4082,
+      "step": 2545500
+    },
+    {
+      "epoch": 12.61,
+      "learning_rate": 4.3695401871751806e-05,
+      "loss": 2.4234,
+      "step": 2546000
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.369416576249858e-05,
+      "loss": 2.4354,
+      "step": 2546500
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.36929271760725e-05,
+      "loss": 2.434,
+      "step": 2547000
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.3691688589646415e-05,
+      "loss": 2.3977,
+      "step": 2547500
+    },
+    {
+      "epoch": 12.62,
+      "learning_rate": 4.369045000322033e-05,
+      "loss": 2.3974,
+      "step": 2548000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.368921141679424e-05,
+      "loss": 2.4089,
+      "step": 2548500
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.368797283036816e-05,
+      "loss": 2.4315,
+      "step": 2549000
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.3686734243942076e-05,
+      "loss": 2.4032,
+      "step": 2549500
+    },
+    {
+      "epoch": 12.63,
+      "learning_rate": 4.368549565751599e-05,
+      "loss": 2.4246,
+      "step": 2550000
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.368425707108991e-05,
+      "loss": 2.3949,
+      "step": 2550500
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.368302096183668e-05,
+      "loss": 2.4226,
+      "step": 2551000
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.3681782375410596e-05,
+      "loss": 2.4192,
+      "step": 2551500
+    },
+    {
+      "epoch": 12.64,
+      "learning_rate": 4.368054378898451e-05,
+      "loss": 2.4099,
+      "step": 2552000
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367930520255842e-05,
+      "loss": 2.4061,
+      "step": 2552500
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367806661613234e-05,
+      "loss": 2.4307,
+      "step": 2553000
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367683298405196e-05,
+      "loss": 2.3988,
+      "step": 2553500
+    },
+    {
+      "epoch": 12.65,
+      "learning_rate": 4.367559439762588e-05,
+      "loss": 2.3905,
+      "step": 2554000
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3674355811199794e-05,
+      "loss": 2.4154,
+      "step": 2554500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.367311722477371e-05,
+      "loss": 2.4097,
+      "step": 2555000
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.367187863834763e-05,
+      "loss": 2.4109,
+      "step": 2555500
+    },
+    {
+      "epoch": 12.66,
+      "learning_rate": 4.3670640051921545e-05,
+      "loss": 2.3961,
+      "step": 2556000
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.366940146549546e-05,
+      "loss": 2.4039,
+      "step": 2556500
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.366816287906938e-05,
+      "loss": 2.3979,
+      "step": 2557000
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.366692676981615e-05,
+      "loss": 2.4154,
+      "step": 2557500
+    },
+    {
+      "epoch": 12.67,
+      "learning_rate": 4.3665688183390065e-05,
+      "loss": 2.4105,
+      "step": 2558000
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.366444959696398e-05,
+      "loss": 2.4075,
+      "step": 2558500
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.36632110105379e-05,
+      "loss": 2.4176,
+      "step": 2559000
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.3661972424111816e-05,
+      "loss": 2.4234,
+      "step": 2559500
+    },
+    {
+      "epoch": 12.68,
+      "learning_rate": 4.366073631485858e-05,
+      "loss": 2.414,
+      "step": 2560000
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3659497728432495e-05,
+      "loss": 2.4034,
+      "step": 2560500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.365825914200641e-05,
+      "loss": 2.4229,
+      "step": 2561000
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.365702055558033e-05,
+      "loss": 2.3905,
+      "step": 2561500
+    },
+    {
+      "epoch": 12.69,
+      "learning_rate": 4.3655781969154245e-05,
+      "loss": 2.4277,
+      "step": 2562000
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.365454338272816e-05,
+      "loss": 2.4012,
+      "step": 2562500
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.365330479630208e-05,
+      "loss": 2.4179,
+      "step": 2563000
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.365206868704885e-05,
+      "loss": 2.4339,
+      "step": 2563500
+    },
+    {
+      "epoch": 12.7,
+      "learning_rate": 4.3650830100622765e-05,
+      "loss": 2.4012,
+      "step": 2564000
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.364959151419668e-05,
+      "loss": 2.4154,
+      "step": 2564500
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.36483529277706e-05,
+      "loss": 2.3908,
+      "step": 2565000
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.3647114341344516e-05,
+      "loss": 2.4239,
+      "step": 2565500
+    },
+    {
+      "epoch": 12.71,
+      "learning_rate": 4.364587575491843e-05,
+      "loss": 2.403,
+      "step": 2566000
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.364463716849235e-05,
+      "loss": 2.3735,
+      "step": 2566500
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.3643398582066267e-05,
+      "loss": 2.4058,
+      "step": 2567000
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.364216247281303e-05,
+      "loss": 2.4092,
+      "step": 2567500
+    },
+    {
+      "epoch": 12.72,
+      "learning_rate": 4.36409263635598e-05,
+      "loss": 2.4055,
+      "step": 2568000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.3639687777133714e-05,
+      "loss": 2.4084,
+      "step": 2568500
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.363844919070763e-05,
+      "loss": 2.4078,
+      "step": 2569000
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.363721060428155e-05,
+      "loss": 2.3979,
+      "step": 2569500
+    },
+    {
+      "epoch": 12.73,
+      "learning_rate": 4.363597449502832e-05,
+      "loss": 2.4249,
+      "step": 2570000
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3634735908602234e-05,
+      "loss": 2.4067,
+      "step": 2570500
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.3633499799349e-05,
+      "loss": 2.4111,
+      "step": 2571000
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.363226121292291e-05,
+      "loss": 2.3941,
+      "step": 2571500
+    },
+    {
+      "epoch": 12.74,
+      "learning_rate": 4.363102262649683e-05,
+      "loss": 2.4572,
+      "step": 2572000
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.362978404007075e-05,
+      "loss": 2.428,
+      "step": 2572500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3628545453644664e-05,
+      "loss": 2.4121,
+      "step": 2573000
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.362730686721858e-05,
+      "loss": 2.4407,
+      "step": 2573500
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.36260682807925e-05,
+      "loss": 2.4202,
+      "step": 2574000
+    },
+    {
+      "epoch": 12.75,
+      "learning_rate": 4.3624829694366414e-05,
+      "loss": 2.3971,
+      "step": 2574500
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.362359358511318e-05,
+      "loss": 2.4459,
+      "step": 2575000
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.36223549986871e-05,
+      "loss": 2.4038,
+      "step": 2575500
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.362111641226102e-05,
+      "loss": 2.4283,
+      "step": 2576000
+    },
+    {
+      "epoch": 12.76,
+      "learning_rate": 4.3619877825834934e-05,
+      "loss": 2.4301,
+      "step": 2576500
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.361863923940885e-05,
+      "loss": 2.4261,
+      "step": 2577000
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.361740313015562e-05,
+      "loss": 2.4179,
+      "step": 2577500
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.361616702090239e-05,
+      "loss": 2.4112,
+      "step": 2578000
+    },
+    {
+      "epoch": 12.77,
+      "learning_rate": 4.3614928434476305e-05,
+      "loss": 2.4111,
+      "step": 2578500
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.361368984805022e-05,
+      "loss": 2.4339,
+      "step": 2579000
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.361245126162414e-05,
+      "loss": 2.4154,
+      "step": 2579500
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.3611212675198056e-05,
+      "loss": 2.4092,
+      "step": 2580000
+    },
+    {
+      "epoch": 12.78,
+      "learning_rate": 4.360997408877197e-05,
+      "loss": 2.4017,
+      "step": 2580500
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3608737979518735e-05,
+      "loss": 2.3993,
+      "step": 2581000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.360749939309265e-05,
+      "loss": 2.4092,
+      "step": 2581500
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.360626080666657e-05,
+      "loss": 2.4148,
+      "step": 2582000
+    },
+    {
+      "epoch": 12.79,
+      "learning_rate": 4.3605022220240486e-05,
+      "loss": 2.4223,
+      "step": 2582500
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.36037836338144e-05,
+      "loss": 2.4065,
+      "step": 2583000
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.360254504738832e-05,
+      "loss": 2.4275,
+      "step": 2583500
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.360130646096223e-05,
+      "loss": 2.3997,
+      "step": 2584000
+    },
+    {
+      "epoch": 12.8,
+      "learning_rate": 4.360006787453615e-05,
+      "loss": 2.4098,
+      "step": 2584500
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3598829288110064e-05,
+      "loss": 2.4142,
+      "step": 2585000
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.359759070168398e-05,
+      "loss": 2.4112,
+      "step": 2585500
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.3596354592430756e-05,
+      "loss": 2.4065,
+      "step": 2586000
+    },
+    {
+      "epoch": 12.81,
+      "learning_rate": 4.359511600600467e-05,
+      "loss": 2.4327,
+      "step": 2586500
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.3593879896751435e-05,
+      "loss": 2.4185,
+      "step": 2587000
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.359264131032535e-05,
+      "loss": 2.3929,
+      "step": 2587500
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.359140272389927e-05,
+      "loss": 2.4129,
+      "step": 2588000
+    },
+    {
+      "epoch": 12.82,
+      "learning_rate": 4.3590164137473186e-05,
+      "loss": 2.4261,
+      "step": 2588500
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.35889255510471e-05,
+      "loss": 2.4036,
+      "step": 2589000
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.358768696462102e-05,
+      "loss": 2.4252,
+      "step": 2589500
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.358644837819494e-05,
+      "loss": 2.4257,
+      "step": 2590000
+    },
+    {
+      "epoch": 12.83,
+      "learning_rate": 4.358520979176885e-05,
+      "loss": 2.3993,
+      "step": 2590500
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.3583971205342764e-05,
+      "loss": 2.4082,
+      "step": 2591000
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.358273261891668e-05,
+      "loss": 2.417,
+      "step": 2591500
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.35814940324906e-05,
+      "loss": 2.4277,
+      "step": 2592000
+    },
+    {
+      "epoch": 12.84,
+      "learning_rate": 4.3580257923237373e-05,
+      "loss": 2.4523,
+      "step": 2592500
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.357901933681129e-05,
+      "loss": 2.4193,
+      "step": 2593000
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.35777807503852e-05,
+      "loss": 2.403,
+      "step": 2593500
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.357654216395912e-05,
+      "loss": 2.4337,
+      "step": 2594000
+    },
+    {
+      "epoch": 12.85,
+      "learning_rate": 4.3575303577533034e-05,
+      "loss": 2.4078,
+      "step": 2594500
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.35740674682798e-05,
+      "loss": 2.4119,
+      "step": 2595000
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.357282888185372e-05,
+      "loss": 2.4088,
+      "step": 2595500
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.357159029542764e-05,
+      "loss": 2.4246,
+      "step": 2596000
+    },
+    {
+      "epoch": 12.86,
+      "learning_rate": 4.357035170900155e-05,
+      "loss": 2.4001,
+      "step": 2596500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.3569113122575464e-05,
+      "loss": 2.4112,
+      "step": 2597000
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.356787701332224e-05,
+      "loss": 2.4181,
+      "step": 2597500
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.356664090406901e-05,
+      "loss": 2.43,
+      "step": 2598000
+    },
+    {
+      "epoch": 12.87,
+      "learning_rate": 4.356540479481577e-05,
+      "loss": 2.4409,
+      "step": 2598500
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.3564168685562546e-05,
+      "loss": 2.4328,
+      "step": 2599000
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.356293009913646e-05,
+      "loss": 2.4351,
+      "step": 2599500
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.356169151271038e-05,
+      "loss": 2.4067,
+      "step": 2600000
+    },
+    {
+      "epoch": 12.88,
+      "learning_rate": 4.356045292628429e-05,
+      "loss": 2.4088,
+      "step": 2600500
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.355921433985821e-05,
+      "loss": 2.4367,
+      "step": 2601000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.3557975753432124e-05,
+      "loss": 2.4095,
+      "step": 2601500
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.355673716700604e-05,
+      "loss": 2.4283,
+      "step": 2602000
+    },
+    {
+      "epoch": 12.89,
+      "learning_rate": 4.355549858057996e-05,
+      "loss": 2.4218,
+      "step": 2602500
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.3554259994153875e-05,
+      "loss": 2.4259,
+      "step": 2603000
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.355302140772779e-05,
+      "loss": 2.3895,
+      "step": 2603500
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.355178282130171e-05,
+      "loss": 2.416,
+      "step": 2604000
+    },
+    {
+      "epoch": 12.9,
+      "learning_rate": 4.3550544234875626e-05,
+      "loss": 2.3923,
+      "step": 2604500
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.354930812562239e-05,
+      "loss": 2.3966,
+      "step": 2605000
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.3548069539196305e-05,
+      "loss": 2.4156,
+      "step": 2605500
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.354683095277022e-05,
+      "loss": 2.4116,
+      "step": 2606000
+    },
+    {
+      "epoch": 12.91,
+      "learning_rate": 4.354559236634414e-05,
+      "loss": 2.4211,
+      "step": 2606500
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.3544353779918055e-05,
+      "loss": 2.4306,
+      "step": 2607000
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.354311519349197e-05,
+      "loss": 2.3948,
+      "step": 2607500
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.354187660706589e-05,
+      "loss": 2.4124,
+      "step": 2608000
+    },
+    {
+      "epoch": 12.92,
+      "learning_rate": 4.3540638020639806e-05,
+      "loss": 2.4227,
+      "step": 2608500
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.353939943421372e-05,
+      "loss": 2.4166,
+      "step": 2609000
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.353816084778764e-05,
+      "loss": 2.4322,
+      "step": 2609500
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.353692226136156e-05,
+      "loss": 2.4191,
+      "step": 2610000
+    },
+    {
+      "epoch": 12.93,
+      "learning_rate": 4.3535686152108326e-05,
+      "loss": 2.424,
+      "step": 2610500
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.353444756568224e-05,
+      "loss": 2.4131,
+      "step": 2611000
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.353320897925616e-05,
+      "loss": 2.422,
+      "step": 2611500
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.3531970392830076e-05,
+      "loss": 2.3948,
+      "step": 2612000
+    },
+    {
+      "epoch": 12.94,
+      "learning_rate": 4.3530731806403993e-05,
+      "loss": 2.4332,
+      "step": 2612500
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.352949321997791e-05,
+      "loss": 2.4331,
+      "step": 2613000
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.352825463355182e-05,
+      "loss": 2.4107,
+      "step": 2613500
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.352701604712574e-05,
+      "loss": 2.4368,
+      "step": 2614000
+    },
+    {
+      "epoch": 12.95,
+      "learning_rate": 4.3525779937872506e-05,
+      "loss": 2.4095,
+      "step": 2614500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.352454135144642e-05,
+      "loss": 2.3953,
+      "step": 2615000
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.352330276502034e-05,
+      "loss": 2.411,
+      "step": 2615500
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.352206417859426e-05,
+      "loss": 2.3921,
+      "step": 2616000
+    },
+    {
+      "epoch": 12.96,
+      "learning_rate": 4.3520828069341026e-05,
+      "loss": 2.4145,
+      "step": 2616500
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.351958948291494e-05,
+      "loss": 2.4006,
+      "step": 2617000
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.351835089648886e-05,
+      "loss": 2.4215,
+      "step": 2617500
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.3517112310062777e-05,
+      "loss": 2.4077,
+      "step": 2618000
+    },
+    {
+      "epoch": 12.97,
+      "learning_rate": 4.3515873723636694e-05,
+      "loss": 2.4019,
+      "step": 2618500
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.351463513721061e-05,
+      "loss": 2.4322,
+      "step": 2619000
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.351339655078453e-05,
+      "loss": 2.4438,
+      "step": 2619500
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.3512157964358444e-05,
+      "loss": 2.3985,
+      "step": 2620000
+    },
+    {
+      "epoch": 12.98,
+      "learning_rate": 4.3510919377932354e-05,
+      "loss": 2.4106,
+      "step": 2620500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.350968326867912e-05,
+      "loss": 2.4127,
+      "step": 2621000
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.350844468225304e-05,
+      "loss": 2.3968,
+      "step": 2621500
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.350720609582696e-05,
+      "loss": 2.4192,
+      "step": 2622000
+    },
+    {
+      "epoch": 12.99,
+      "learning_rate": 4.3505967509400874e-05,
+      "loss": 2.4349,
+      "step": 2622500
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.350472892297479e-05,
+      "loss": 2.403,
+      "step": 2623000
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.350349281372156e-05,
+      "loss": 2.4137,
+      "step": 2623500
+    },
+    {
+      "epoch": 13.0,
+      "eval_accuracy": 0.6461218698848585,
+      "eval_accuracy_mlm": 0.5996847093159505,
+      "eval_accuracy_nsp": 0.8652920665675657,
+      "eval_loss": 2.396927833557129,
+      "eval_runtime": 146.13,
+      "eval_samples_per_second": 1744.741,
+      "eval_steps_per_second": 72.702,
+      "step": 2623959
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.350225422729548e-05,
+      "loss": 2.3977,
+      "step": 2624000
+    },
+    {
+      "epoch": 13.0,
+      "learning_rate": 4.3501015640869394e-05,
+      "loss": 2.374,
+      "step": 2624500
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.3499779531616156e-05,
+      "loss": 2.3743,
+      "step": 2625000
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.349854094519007e-05,
+      "loss": 2.3774,
+      "step": 2625500
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.349730483593684e-05,
+      "loss": 2.3954,
+      "step": 2626000
+    },
+    {
+      "epoch": 13.01,
+      "learning_rate": 4.349606624951076e-05,
+      "loss": 2.3685,
+      "step": 2626500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.3494827663084675e-05,
+      "loss": 2.3846,
+      "step": 2627000
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.349358907665859e-05,
+      "loss": 2.4169,
+      "step": 2627500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.349235049023251e-05,
+      "loss": 2.3881,
+      "step": 2628000
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.3491111903806426e-05,
+      "loss": 2.3866,
+      "step": 2628500
+    },
+    {
+      "epoch": 13.02,
+      "learning_rate": 4.3489875794553195e-05,
+      "loss": 2.3958,
+      "step": 2629000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.3488639685299964e-05,
+      "loss": 2.3686,
+      "step": 2629500
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.348740109887388e-05,
+      "loss": 2.3922,
+      "step": 2630000
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.34861625124478e-05,
+      "loss": 2.3662,
+      "step": 2630500
+    },
+    {
+      "epoch": 13.03,
+      "learning_rate": 4.3484923926021714e-05,
+      "loss": 2.3793,
+      "step": 2631000
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.3483685339595625e-05,
+      "loss": 2.3712,
+      "step": 2631500
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.348244675316954e-05,
+      "loss": 2.4067,
+      "step": 2632000
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.348120816674346e-05,
+      "loss": 2.3861,
+      "step": 2632500
+    },
+    {
+      "epoch": 13.04,
+      "learning_rate": 4.3479972057490234e-05,
+      "loss": 2.4153,
+      "step": 2633000
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.347873347106415e-05,
+      "loss": 2.3681,
+      "step": 2633500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.347749488463807e-05,
+      "loss": 2.3843,
+      "step": 2634000
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.347625629821198e-05,
+      "loss": 2.3793,
+      "step": 2634500
+    },
+    {
+      "epoch": 13.05,
+      "learning_rate": 4.3475017711785895e-05,
+      "loss": 2.3923,
+      "step": 2635000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.347377912535981e-05,
+      "loss": 2.3799,
+      "step": 2635500
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.347254053893373e-05,
+      "loss": 2.3595,
+      "step": 2636000
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.3471301952507646e-05,
+      "loss": 2.4064,
+      "step": 2636500
+    },
+    {
+      "epoch": 13.06,
+      "learning_rate": 4.3470063366081556e-05,
+      "loss": 2.4083,
+      "step": 2637000
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.346882477965547e-05,
+      "loss": 2.3916,
+      "step": 2637500
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.346758619322939e-05,
+      "loss": 2.372,
+      "step": 2638000
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.346634760680331e-05,
+      "loss": 2.3781,
+      "step": 2638500
+    },
+    {
+      "epoch": 13.07,
+      "learning_rate": 4.3465109020377224e-05,
+      "loss": 2.408,
+      "step": 2639000
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.346387043395114e-05,
+      "loss": 2.4136,
+      "step": 2639500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.346263432469791e-05,
+      "loss": 2.3621,
+      "step": 2640000
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.3461398215444685e-05,
+      "loss": 2.3862,
+      "step": 2640500
+    },
+    {
+      "epoch": 13.08,
+      "learning_rate": 4.3460159629018595e-05,
+      "loss": 2.3855,
+      "step": 2641000
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.345892104259251e-05,
+      "loss": 2.4046,
+      "step": 2641500
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.345768245616643e-05,
+      "loss": 2.3887,
+      "step": 2642000
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.3456443869740346e-05,
+      "loss": 2.3702,
+      "step": 2642500
+    },
+    {
+      "epoch": 13.09,
+      "learning_rate": 4.3455207760487115e-05,
+      "loss": 2.4123,
+      "step": 2643000
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.345396917406103e-05,
+      "loss": 2.3679,
+      "step": 2643500
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.345273058763494e-05,
+      "loss": 2.3806,
+      "step": 2644000
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.345149200120886e-05,
+      "loss": 2.4071,
+      "step": 2644500
+    },
+    {
+      "epoch": 13.1,
+      "learning_rate": 4.3450253414782776e-05,
+      "loss": 2.3682,
+      "step": 2645000
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.344901482835669e-05,
+      "loss": 2.375,
+      "step": 2645500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.344777624193061e-05,
+      "loss": 2.3903,
+      "step": 2646000
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.3446537655504526e-05,
+      "loss": 2.3944,
+      "step": 2646500
+    },
+    {
+      "epoch": 13.11,
+      "learning_rate": 4.344529906907844e-05,
+      "loss": 2.3925,
+      "step": 2647000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.344406048265236e-05,
+      "loss": 2.3984,
+      "step": 2647500
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.344282189622628e-05,
+      "loss": 2.4028,
+      "step": 2648000
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.3441583309800194e-05,
+      "loss": 2.3867,
+      "step": 2648500
+    },
+    {
+      "epoch": 13.12,
+      "learning_rate": 4.344034472337411e-05,
+      "loss": 2.4064,
+      "step": 2649000
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.343910613694803e-05,
+      "loss": 2.3914,
+      "step": 2649500
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.34378700276948e-05,
+      "loss": 2.4027,
+      "step": 2650000
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.343663391844156e-05,
+      "loss": 2.4052,
+      "step": 2650500
+    },
+    {
+      "epoch": 13.13,
+      "learning_rate": 4.3435395332015476e-05,
+      "loss": 2.3882,
+      "step": 2651000
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.3434161699935096e-05,
+      "loss": 2.3821,
+      "step": 2651500
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.3432923113509013e-05,
+      "loss": 2.4076,
+      "step": 2652000
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.343168452708293e-05,
+      "loss": 2.3935,
+      "step": 2652500
+    },
+    {
+      "epoch": 13.14,
+      "learning_rate": 4.34304484178297e-05,
+      "loss": 2.3977,
+      "step": 2653000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.3429209831403616e-05,
+      "loss": 2.3899,
+      "step": 2653500
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.342797124497753e-05,
+      "loss": 2.3781,
+      "step": 2654000
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.342673265855145e-05,
+      "loss": 2.3914,
+      "step": 2654500
+    },
+    {
+      "epoch": 13.15,
+      "learning_rate": 4.342549407212537e-05,
+      "loss": 2.3926,
+      "step": 2655000
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.3424255485699284e-05,
+      "loss": 2.4267,
+      "step": 2655500
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.34230168992732e-05,
+      "loss": 2.3881,
+      "step": 2656000
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.342177831284712e-05,
+      "loss": 2.3844,
+      "step": 2656500
+    },
+    {
+      "epoch": 13.16,
+      "learning_rate": 4.3420539726421035e-05,
+      "loss": 2.4005,
+      "step": 2657000
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341930113999495e-05,
+      "loss": 2.3819,
+      "step": 2657500
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341806255356887e-05,
+      "loss": 2.395,
+      "step": 2658000
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341682644431563e-05,
+      "loss": 2.371,
+      "step": 2658500
+    },
+    {
+      "epoch": 13.17,
+      "learning_rate": 4.341558785788955e-05,
+      "loss": 2.3832,
+      "step": 2659000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.3414349271463464e-05,
+      "loss": 2.391,
+      "step": 2659500
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.341311068503738e-05,
+      "loss": 2.3821,
+      "step": 2660000
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.34118720986113e-05,
+      "loss": 2.3738,
+      "step": 2660500
+    },
+    {
+      "epoch": 13.18,
+      "learning_rate": 4.3410633512185215e-05,
+      "loss": 2.398,
+      "step": 2661000
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.340939492575913e-05,
+      "loss": 2.395,
+      "step": 2661500
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.340815633933305e-05,
+      "loss": 2.3963,
+      "step": 2662000
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.3406917752906966e-05,
+      "loss": 2.3994,
+      "step": 2662500
+    },
+    {
+      "epoch": 13.19,
+      "learning_rate": 4.3405679166480876e-05,
+      "loss": 2.3915,
+      "step": 2663000
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.340444058005479e-05,
+      "loss": 2.4018,
+      "step": 2663500
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.340320199362871e-05,
+      "loss": 2.3685,
+      "step": 2664000
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.340196340720263e-05,
+      "loss": 2.4209,
+      "step": 2664500
+    },
+    {
+      "epoch": 13.2,
+      "learning_rate": 4.3400724820776544e-05,
+      "loss": 2.406,
+      "step": 2665000
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.339948623435046e-05,
+      "loss": 2.3968,
+      "step": 2665500
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.339824764792438e-05,
+      "loss": 2.3927,
+      "step": 2666000
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.3397011538671146e-05,
+      "loss": 2.3701,
+      "step": 2666500
+    },
+    {
+      "epoch": 13.21,
+      "learning_rate": 4.339577295224506e-05,
+      "loss": 2.364,
+      "step": 2667000
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.339453436581898e-05,
+      "loss": 2.3787,
+      "step": 2667500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.33932957793929e-05,
+      "loss": 2.3985,
+      "step": 2668000
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.3392057192966814e-05,
+      "loss": 2.379,
+      "step": 2668500
+    },
+    {
+      "epoch": 13.22,
+      "learning_rate": 4.339081860654073e-05,
+      "loss": 2.3979,
+      "step": 2669000
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.338958002011465e-05,
+      "loss": 2.3876,
+      "step": 2669500
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.3388341433688565e-05,
+      "loss": 2.4065,
+      "step": 2670000
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.338710284726248e-05,
+      "loss": 2.3983,
+      "step": 2670500
+    },
+    {
+      "epoch": 13.23,
+      "learning_rate": 4.338586426083639e-05,
+      "loss": 2.4001,
+      "step": 2671000
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.338462567441031e-05,
+      "loss": 2.4228,
+      "step": 2671500
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.338338956515708e-05,
+      "loss": 2.3995,
+      "step": 2672000
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.3382153455903847e-05,
+      "loss": 2.3867,
+      "step": 2672500
+    },
+    {
+      "epoch": 13.24,
+      "learning_rate": 4.3380914869477763e-05,
+      "loss": 2.3989,
+      "step": 2673000
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.337967628305168e-05,
+      "loss": 2.4051,
+      "step": 2673500
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.33784376966256e-05,
+      "loss": 2.4195,
+      "step": 2674000
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.3377199110199514e-05,
+      "loss": 2.4004,
+      "step": 2674500
+    },
+    {
+      "epoch": 13.25,
+      "learning_rate": 4.337596052377343e-05,
+      "loss": 2.3939,
+      "step": 2675000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.337472193734735e-05,
+      "loss": 2.3849,
+      "step": 2675500
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.3373483350921265e-05,
+      "loss": 2.3963,
+      "step": 2676000
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.337224476449518e-05,
+      "loss": 2.4051,
+      "step": 2676500
+    },
+    {
+      "epoch": 13.26,
+      "learning_rate": 4.3371008655241944e-05,
+      "loss": 2.3968,
+      "step": 2677000
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.336977006881586e-05,
+      "loss": 2.4036,
+      "step": 2677500
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.336853148238978e-05,
+      "loss": 2.4113,
+      "step": 2678000
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.3367292895963695e-05,
+      "loss": 2.4127,
+      "step": 2678500
+    },
+    {
+      "epoch": 13.27,
+      "learning_rate": 4.336605430953761e-05,
+      "loss": 2.3876,
+      "step": 2679000
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.336481572311153e-05,
+      "loss": 2.4134,
+      "step": 2679500
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.33635796138583e-05,
+      "loss": 2.4021,
+      "step": 2680000
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.3362343504605066e-05,
+      "loss": 2.401,
+      "step": 2680500
+    },
+    {
+      "epoch": 13.28,
+      "learning_rate": 4.336110491817898e-05,
+      "loss": 2.3873,
+      "step": 2681000
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.335986633175289e-05,
+      "loss": 2.394,
+      "step": 2681500
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.335863022249967e-05,
+      "loss": 2.4049,
+      "step": 2682000
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.3357391636073586e-05,
+      "loss": 2.3955,
+      "step": 2682500
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.33561530496475e-05,
+      "loss": 2.4062,
+      "step": 2683000
+    },
+    {
+      "epoch": 13.29,
+      "learning_rate": 4.335491446322142e-05,
+      "loss": 2.427,
+      "step": 2683500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.335367587679534e-05,
+      "loss": 2.4073,
+      "step": 2684000
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.3352437290369254e-05,
+      "loss": 2.4097,
+      "step": 2684500
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.3351198703943164e-05,
+      "loss": 2.3962,
+      "step": 2685000
+    },
+    {
+      "epoch": 13.3,
+      "learning_rate": 4.334996011751708e-05,
+      "loss": 2.3851,
+      "step": 2685500
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.3348721531091e-05,
+      "loss": 2.4041,
+      "step": 2686000
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.3347482944664914e-05,
+      "loss": 2.3944,
+      "step": 2686500
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.334624435823883e-05,
+      "loss": 2.4004,
+      "step": 2687000
+    },
+    {
+      "epoch": 13.31,
+      "learning_rate": 4.334501072615845e-05,
+      "loss": 2.3767,
+      "step": 2687500
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.334377213973237e-05,
+      "loss": 2.3833,
+      "step": 2688000
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.3342533553306286e-05,
+      "loss": 2.3789,
+      "step": 2688500
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.3341297444053055e-05,
+      "loss": 2.4131,
+      "step": 2689000
+    },
+    {
+      "epoch": 13.32,
+      "learning_rate": 4.334005885762697e-05,
+      "loss": 2.404,
+      "step": 2689500
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.333882027120089e-05,
+      "loss": 2.4251,
+      "step": 2690000
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.3337581684774806e-05,
+      "loss": 2.4047,
+      "step": 2690500
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.333634309834872e-05,
+      "loss": 2.384,
+      "step": 2691000
+    },
+    {
+      "epoch": 13.33,
+      "learning_rate": 4.333510451192264e-05,
+      "loss": 2.4168,
+      "step": 2691500
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.333386592549655e-05,
+      "loss": 2.3786,
+      "step": 2692000
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.3332627339070467e-05,
+      "loss": 2.3896,
+      "step": 2692500
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.3331388752644383e-05,
+      "loss": 2.3878,
+      "step": 2693000
+    },
+    {
+      "epoch": 13.34,
+      "learning_rate": 4.33301501662183e-05,
+      "loss": 2.3996,
+      "step": 2693500
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.332891157979222e-05,
+      "loss": 2.4324,
+      "step": 2694000
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.3327675470538986e-05,
+      "loss": 2.3555,
+      "step": 2694500
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.33264368841129e-05,
+      "loss": 2.4025,
+      "step": 2695000
+    },
+    {
+      "epoch": 13.35,
+      "learning_rate": 4.332520077485967e-05,
+      "loss": 2.4122,
+      "step": 2695500
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.332396218843359e-05,
+      "loss": 2.3975,
+      "step": 2696000
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.3322723602007506e-05,
+      "loss": 2.3986,
+      "step": 2696500
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.332148501558142e-05,
+      "loss": 2.378,
+      "step": 2697000
+    },
+    {
+      "epoch": 13.36,
+      "learning_rate": 4.332024642915534e-05,
+      "loss": 2.4197,
+      "step": 2697500
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.3319007842729256e-05,
+      "loss": 2.4088,
+      "step": 2698000
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.3317769256303173e-05,
+      "loss": 2.427,
+      "step": 2698500
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.3316530669877084e-05,
+      "loss": 2.4063,
+      "step": 2699000
+    },
+    {
+      "epoch": 13.37,
+      "learning_rate": 4.3315292083451e-05,
+      "loss": 2.4056,
+      "step": 2699500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.331405349702492e-05,
+      "loss": 2.3971,
+      "step": 2700000
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.3312814910598834e-05,
+      "loss": 2.4082,
+      "step": 2700500
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.331157632417275e-05,
+      "loss": 2.401,
+      "step": 2701000
+    },
+    {
+      "epoch": 13.38,
+      "learning_rate": 4.331033773774666e-05,
+      "loss": 2.3885,
+      "step": 2701500
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.330909915132058e-05,
+      "loss": 2.3773,
+      "step": 2702000
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.3307860564894495e-05,
+      "loss": 2.441,
+      "step": 2702500
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.330662197846841e-05,
+      "loss": 2.4094,
+      "step": 2703000
+    },
+    {
+      "epoch": 13.39,
+      "learning_rate": 4.330538339204233e-05,
+      "loss": 2.4056,
+      "step": 2703500
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.3304144805616246e-05,
+      "loss": 2.4129,
+      "step": 2704000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.3302908696363015e-05,
+      "loss": 2.3796,
+      "step": 2704500
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.330167010993693e-05,
+      "loss": 2.3743,
+      "step": 2705000
+    },
+    {
+      "epoch": 13.4,
+      "learning_rate": 4.330043152351085e-05,
+      "loss": 2.4091,
+      "step": 2705500
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.329919789143047e-05,
+      "loss": 2.3977,
+      "step": 2706000
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.3297959305004386e-05,
+      "loss": 2.3935,
+      "step": 2706500
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.32967207185783e-05,
+      "loss": 2.4121,
+      "step": 2707000
+    },
+    {
+      "epoch": 13.41,
+      "learning_rate": 4.329548213215222e-05,
+      "loss": 2.4118,
+      "step": 2707500
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.329424354572614e-05,
+      "loss": 2.4029,
+      "step": 2708000
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.3293007436472906e-05,
+      "loss": 2.3758,
+      "step": 2708500
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.329176885004682e-05,
+      "loss": 2.3919,
+      "step": 2709000
+    },
+    {
+      "epoch": 13.42,
+      "learning_rate": 4.329053026362074e-05,
+      "loss": 2.408,
+      "step": 2709500
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.328929167719466e-05,
+      "loss": 2.3962,
+      "step": 2710000
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.3288053090768574e-05,
+      "loss": 2.3936,
+      "step": 2710500
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.328681450434249e-05,
+      "loss": 2.402,
+      "step": 2711000
+    },
+    {
+      "epoch": 13.43,
+      "learning_rate": 4.328557839508925e-05,
+      "loss": 2.3929,
+      "step": 2711500
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.328433980866317e-05,
+      "loss": 2.3941,
+      "step": 2712000
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.3283101222237086e-05,
+      "loss": 2.396,
+      "step": 2712500
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.3281862635811003e-05,
+      "loss": 2.4242,
+      "step": 2713000
+    },
+    {
+      "epoch": 13.44,
+      "learning_rate": 4.328062404938492e-05,
+      "loss": 2.4014,
+      "step": 2713500
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.327938546295884e-05,
+      "loss": 2.3839,
+      "step": 2714000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.3278146876532754e-05,
+      "loss": 2.4211,
+      "step": 2714500
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.327691076727952e-05,
+      "loss": 2.3848,
+      "step": 2715000
+    },
+    {
+      "epoch": 13.45,
+      "learning_rate": 4.327567218085344e-05,
+      "loss": 2.4099,
+      "step": 2715500
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.327443359442736e-05,
+      "loss": 2.4064,
+      "step": 2716000
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.3273195008001274e-05,
+      "loss": 2.4105,
+      "step": 2716500
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.327195642157519e-05,
+      "loss": 2.4022,
+      "step": 2717000
+    },
+    {
+      "epoch": 13.46,
+      "learning_rate": 4.327071783514911e-05,
+      "loss": 2.3729,
+      "step": 2717500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3269479248723025e-05,
+      "loss": 2.4031,
+      "step": 2718000
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.326824066229694e-05,
+      "loss": 2.3839,
+      "step": 2718500
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.3267004553043704e-05,
+      "loss": 2.4294,
+      "step": 2719000
+    },
+    {
+      "epoch": 13.47,
+      "learning_rate": 4.326576596661762e-05,
+      "loss": 2.4108,
+      "step": 2719500
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.326452738019154e-05,
+      "loss": 2.4364,
+      "step": 2720000
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.3263288793765454e-05,
+      "loss": 2.4068,
+      "step": 2720500
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.326205020733937e-05,
+      "loss": 2.4131,
+      "step": 2721000
+    },
+    {
+      "epoch": 13.48,
+      "learning_rate": 4.326081162091329e-05,
+      "loss": 2.3866,
+      "step": 2721500
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.325957551166006e-05,
+      "loss": 2.388,
+      "step": 2722000
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.3258336925233974e-05,
+      "loss": 2.3977,
+      "step": 2722500
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.3257100815980736e-05,
+      "loss": 2.4153,
+      "step": 2723000
+    },
+    {
+      "epoch": 13.49,
+      "learning_rate": 4.325586222955465e-05,
+      "loss": 2.3835,
+      "step": 2723500
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.325462364312857e-05,
+      "loss": 2.4049,
+      "step": 2724000
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.325338505670249e-05,
+      "loss": 2.4043,
+      "step": 2724500
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.3252146470276404e-05,
+      "loss": 2.4012,
+      "step": 2725000
+    },
+    {
+      "epoch": 13.5,
+      "learning_rate": 4.325090788385032e-05,
+      "loss": 2.3883,
+      "step": 2725500
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.324966929742424e-05,
+      "loss": 2.4236,
+      "step": 2726000
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.3248430710998154e-05,
+      "loss": 2.3839,
+      "step": 2726500
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.324719212457207e-05,
+      "loss": 2.4157,
+      "step": 2727000
+    },
+    {
+      "epoch": 13.51,
+      "learning_rate": 4.324595353814599e-05,
+      "loss": 2.3927,
+      "step": 2727500
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.324471742889276e-05,
+      "loss": 2.388,
+      "step": 2728000
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.3243481319639526e-05,
+      "loss": 2.3996,
+      "step": 2728500
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.3242242733213436e-05,
+      "loss": 2.4144,
+      "step": 2729000
+    },
+    {
+      "epoch": 13.52,
+      "learning_rate": 4.324100414678735e-05,
+      "loss": 2.4069,
+      "step": 2729500
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323976556036127e-05,
+      "loss": 2.4112,
+      "step": 2730000
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323852697393519e-05,
+      "loss": 2.4076,
+      "step": 2730500
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.3237288387509104e-05,
+      "loss": 2.4015,
+      "step": 2731000
+    },
+    {
+      "epoch": 13.53,
+      "learning_rate": 4.323604980108302e-05,
+      "loss": 2.4073,
+      "step": 2731500
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323481369182979e-05,
+      "loss": 2.428,
+      "step": 2732000
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.3233575105403706e-05,
+      "loss": 2.3869,
+      "step": 2732500
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323233651897762e-05,
+      "loss": 2.411,
+      "step": 2733000
+    },
+    {
+      "epoch": 13.54,
+      "learning_rate": 4.323109793255154e-05,
+      "loss": 2.3869,
+      "step": 2733500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.322985934612546e-05,
+      "loss": 2.4143,
+      "step": 2734000
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.3228620759699374e-05,
+      "loss": 2.4027,
+      "step": 2734500
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.322738217327329e-05,
+      "loss": 2.3911,
+      "step": 2735000
+    },
+    {
+      "epoch": 13.55,
+      "learning_rate": 4.322614606402006e-05,
+      "loss": 2.4308,
+      "step": 2735500
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.322490747759397e-05,
+      "loss": 2.4117,
+      "step": 2736000
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.322366889116789e-05,
+      "loss": 2.4093,
+      "step": 2736500
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.3222430304741804e-05,
+      "loss": 2.372,
+      "step": 2737000
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.322119171831572e-05,
+      "loss": 2.3907,
+      "step": 2737500
+    },
+    {
+      "epoch": 13.56,
+      "learning_rate": 4.321995313188964e-05,
+      "loss": 2.3867,
+      "step": 2738000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.3218714545463555e-05,
+      "loss": 2.3667,
+      "step": 2738500
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.3217478436210323e-05,
+      "loss": 2.3809,
+      "step": 2739000
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.321623984978424e-05,
+      "loss": 2.4062,
+      "step": 2739500
+    },
+    {
+      "epoch": 13.57,
+      "learning_rate": 4.321500126335816e-05,
+      "loss": 2.396,
+      "step": 2740000
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.3213762676932074e-05,
+      "loss": 2.4295,
+      "step": 2740500
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.321252409050599e-05,
+      "loss": 2.4223,
+      "step": 2741000
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.321128550407991e-05,
+      "loss": 2.4312,
+      "step": 2741500
+    },
+    {
+      "epoch": 13.58,
+      "learning_rate": 4.3210046917653825e-05,
+      "loss": 2.408,
+      "step": 2742000
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320880833122774e-05,
+      "loss": 2.391,
+      "step": 2742500
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320756974480166e-05,
+      "loss": 2.4148,
+      "step": 2743000
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320633363554842e-05,
+      "loss": 2.3847,
+      "step": 2743500
+    },
+    {
+      "epoch": 13.59,
+      "learning_rate": 4.320509504912234e-05,
+      "loss": 2.4125,
+      "step": 2744000
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.3203856462696255e-05,
+      "loss": 2.4068,
+      "step": 2744500
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.320261787627017e-05,
+      "loss": 2.3831,
+      "step": 2745000
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.320137928984409e-05,
+      "loss": 2.4067,
+      "step": 2745500
+    },
+    {
+      "epoch": 13.6,
+      "learning_rate": 4.3200140703418006e-05,
+      "loss": 2.3968,
+      "step": 2746000
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.319890211699192e-05,
+      "loss": 2.4172,
+      "step": 2746500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.319766353056584e-05,
+      "loss": 2.4011,
+      "step": 2747000
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.319642494413975e-05,
+      "loss": 2.4083,
+      "step": 2747500
+    },
+    {
+      "epoch": 13.61,
+      "learning_rate": 4.3195188834886525e-05,
+      "loss": 2.4248,
+      "step": 2748000
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.319395024846044e-05,
+      "loss": 2.3911,
+      "step": 2748500
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.319271166203436e-05,
+      "loss": 2.3895,
+      "step": 2749000
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.3191473075608276e-05,
+      "loss": 2.4364,
+      "step": 2749500
+    },
+    {
+      "epoch": 13.62,
+      "learning_rate": 4.319023448918219e-05,
+      "loss": 2.4193,
+      "step": 2750000
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.3188998379928955e-05,
+      "loss": 2.4033,
+      "step": 2750500
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.318775979350287e-05,
+      "loss": 2.4104,
+      "step": 2751000
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.318652120707679e-05,
+      "loss": 2.4298,
+      "step": 2751500
+    },
+    {
+      "epoch": 13.63,
+      "learning_rate": 4.3185282620650706e-05,
+      "loss": 2.3979,
+      "step": 2752000
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318404403422462e-05,
+      "loss": 2.39,
+      "step": 2752500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318280792497139e-05,
+      "loss": 2.4094,
+      "step": 2753000
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.318156933854531e-05,
+      "loss": 2.4306,
+      "step": 2753500
+    },
+    {
+      "epoch": 13.64,
+      "learning_rate": 4.3180330752119225e-05,
+      "loss": 2.4091,
+      "step": 2754000
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.317909216569314e-05,
+      "loss": 2.4295,
+      "step": 2754500
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.317785357926706e-05,
+      "loss": 2.4031,
+      "step": 2755000
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.317661747001383e-05,
+      "loss": 2.4377,
+      "step": 2755500
+    },
+    {
+      "epoch": 13.65,
+      "learning_rate": 4.3175378883587745e-05,
+      "loss": 2.3773,
+      "step": 2756000
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.3174140297161655e-05,
+      "loss": 2.4163,
+      "step": 2756500
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.317290171073557e-05,
+      "loss": 2.4139,
+      "step": 2757000
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.317166312430949e-05,
+      "loss": 2.4116,
+      "step": 2757500
+    },
+    {
+      "epoch": 13.66,
+      "learning_rate": 4.3170424537883406e-05,
+      "loss": 2.391,
+      "step": 2758000
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316918595145732e-05,
+      "loss": 2.4302,
+      "step": 2758500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316794984220409e-05,
+      "loss": 2.4052,
+      "step": 2759000
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316671373295086e-05,
+      "loss": 2.4154,
+      "step": 2759500
+    },
+    {
+      "epoch": 13.67,
+      "learning_rate": 4.316547514652478e-05,
+      "loss": 2.3856,
+      "step": 2760000
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.3164236560098694e-05,
+      "loss": 2.4001,
+      "step": 2760500
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.316299797367261e-05,
+      "loss": 2.4079,
+      "step": 2761000
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.316175938724653e-05,
+      "loss": 2.3876,
+      "step": 2761500
+    },
+    {
+      "epoch": 13.68,
+      "learning_rate": 4.3160520800820445e-05,
+      "loss": 2.4033,
+      "step": 2762000
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.315928221439436e-05,
+      "loss": 2.4024,
+      "step": 2762500
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.3158046105141124e-05,
+      "loss": 2.3802,
+      "step": 2763000
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.315680751871504e-05,
+      "loss": 2.4104,
+      "step": 2763500
+    },
+    {
+      "epoch": 13.69,
+      "learning_rate": 4.315556893228896e-05,
+      "loss": 2.3841,
+      "step": 2764000
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.315433282303573e-05,
+      "loss": 2.3901,
+      "step": 2764500
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.3153094236609644e-05,
+      "loss": 2.4044,
+      "step": 2765000
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.315185565018356e-05,
+      "loss": 2.4209,
+      "step": 2765500
+    },
+    {
+      "epoch": 13.7,
+      "learning_rate": 4.315061706375748e-05,
+      "loss": 2.3937,
+      "step": 2766000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.3149380954504246e-05,
+      "loss": 2.41,
+      "step": 2766500
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.3148142368078156e-05,
+      "loss": 2.4149,
+      "step": 2767000
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.314690378165207e-05,
+      "loss": 2.4024,
+      "step": 2767500
+    },
+    {
+      "epoch": 13.71,
+      "learning_rate": 4.314566519522599e-05,
+      "loss": 2.4355,
+      "step": 2768000
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.314442660879991e-05,
+      "loss": 2.4085,
+      "step": 2768500
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.3143188022373824e-05,
+      "loss": 2.3709,
+      "step": 2769000
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.314194943594774e-05,
+      "loss": 2.397,
+      "step": 2769500
+    },
+    {
+      "epoch": 13.72,
+      "learning_rate": 4.314071084952166e-05,
+      "loss": 2.3901,
+      "step": 2770000
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.3139472263095575e-05,
+      "loss": 2.4026,
+      "step": 2770500
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.313823367666949e-05,
+      "loss": 2.4013,
+      "step": 2771000
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.313699756741626e-05,
+      "loss": 2.3871,
+      "step": 2771500
+    },
+    {
+      "epoch": 13.73,
+      "learning_rate": 4.313575898099018e-05,
+      "loss": 2.4078,
+      "step": 2772000
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.3134520394564094e-05,
+      "loss": 2.4151,
+      "step": 2772500
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.313328180813801e-05,
+      "loss": 2.4024,
+      "step": 2773000
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.313204322171193e-05,
+      "loss": 2.4182,
+      "step": 2773500
+    },
+    {
+      "epoch": 13.74,
+      "learning_rate": 4.3130804635285845e-05,
+      "loss": 2.3841,
+      "step": 2774000
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.312956604885976e-05,
+      "loss": 2.3987,
+      "step": 2774500
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.312832746243368e-05,
+      "loss": 2.3907,
+      "step": 2775000
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.3127088876007596e-05,
+      "loss": 2.4168,
+      "step": 2775500
+    },
+    {
+      "epoch": 13.75,
+      "learning_rate": 4.312585028958151e-05,
+      "loss": 2.4074,
+      "step": 2776000
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.312461170315542e-05,
+      "loss": 2.3894,
+      "step": 2776500
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.312337311672934e-05,
+      "loss": 2.3936,
+      "step": 2777000
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.312213453030326e-05,
+      "loss": 2.3887,
+      "step": 2777500
+    },
+    {
+      "epoch": 13.76,
+      "learning_rate": 4.3120895943877174e-05,
+      "loss": 2.4014,
+      "step": 2778000
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.311965735745109e-05,
+      "loss": 2.4004,
+      "step": 2778500
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.3118418771025e-05,
+      "loss": 2.4207,
+      "step": 2779000
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.3117182661771777e-05,
+      "loss": 2.4038,
+      "step": 2779500
+    },
+    {
+      "epoch": 13.77,
+      "learning_rate": 4.3115944075345694e-05,
+      "loss": 2.417,
+      "step": 2780000
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311470548891961e-05,
+      "loss": 2.4107,
+      "step": 2780500
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311346690249353e-05,
+      "loss": 2.4136,
+      "step": 2781000
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.3112228316067444e-05,
+      "loss": 2.3845,
+      "step": 2781500
+    },
+    {
+      "epoch": 13.78,
+      "learning_rate": 4.311099220681421e-05,
+      "loss": 2.43,
+      "step": 2782000
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.3109756097560975e-05,
+      "loss": 2.4391,
+      "step": 2782500
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.3108519988307744e-05,
+      "loss": 2.417,
+      "step": 2783000
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.310728387905452e-05,
+      "loss": 2.4119,
+      "step": 2783500
+    },
+    {
+      "epoch": 13.79,
+      "learning_rate": 4.310604529262843e-05,
+      "loss": 2.4281,
+      "step": 2784000
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.31048091833752e-05,
+      "loss": 2.4254,
+      "step": 2784500
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.3103570596949115e-05,
+      "loss": 2.3977,
+      "step": 2785000
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.310233201052303e-05,
+      "loss": 2.4255,
+      "step": 2785500
+    },
+    {
+      "epoch": 13.8,
+      "learning_rate": 4.310109342409695e-05,
+      "loss": 2.393,
+      "step": 2786000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.3099854837670866e-05,
+      "loss": 2.4038,
+      "step": 2786500
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.309861625124478e-05,
+      "loss": 2.4045,
+      "step": 2787000
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.30973776648187e-05,
+      "loss": 2.4029,
+      "step": 2787500
+    },
+    {
+      "epoch": 13.81,
+      "learning_rate": 4.309613907839262e-05,
+      "loss": 2.4214,
+      "step": 2788000
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.3094902969139386e-05,
+      "loss": 2.3809,
+      "step": 2788500
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.30936643827133e-05,
+      "loss": 2.402,
+      "step": 2789000
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.309242579628722e-05,
+      "loss": 2.3671,
+      "step": 2789500
+    },
+    {
+      "epoch": 13.82,
+      "learning_rate": 4.3091187209861137e-05,
+      "loss": 2.391,
+      "step": 2790000
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.30899511006079e-05,
+      "loss": 2.4029,
+      "step": 2790500
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.3088712514181816e-05,
+      "loss": 2.3973,
+      "step": 2791000
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.308747392775573e-05,
+      "loss": 2.3823,
+      "step": 2791500
+    },
+    {
+      "epoch": 13.83,
+      "learning_rate": 4.308623534132965e-05,
+      "loss": 2.4324,
+      "step": 2792000
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.3084996754903566e-05,
+      "loss": 2.4068,
+      "step": 2792500
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.308375816847748e-05,
+      "loss": 2.4183,
+      "step": 2793000
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.30825195820514e-05,
+      "loss": 2.4176,
+      "step": 2793500
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.308128099562532e-05,
+      "loss": 2.41,
+      "step": 2794000
+    },
+    {
+      "epoch": 13.84,
+      "learning_rate": 4.3080042409199234e-05,
+      "loss": 2.3977,
+      "step": 2794500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.3078806299946e-05,
+      "loss": 2.3854,
+      "step": 2795000
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.3077570190692765e-05,
+      "loss": 2.4014,
+      "step": 2795500
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.307633160426668e-05,
+      "loss": 2.432,
+      "step": 2796000
+    },
+    {
+      "epoch": 13.85,
+      "learning_rate": 4.30750930178406e-05,
+      "loss": 2.4016,
+      "step": 2796500
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.3073854431414516e-05,
+      "loss": 2.4061,
+      "step": 2797000
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.307261584498843e-05,
+      "loss": 2.3972,
+      "step": 2797500
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.307137725856235e-05,
+      "loss": 2.3841,
+      "step": 2798000
+    },
+    {
+      "epoch": 13.86,
+      "learning_rate": 4.307014114930912e-05,
+      "loss": 2.3905,
+      "step": 2798500
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.3068902562883035e-05,
+      "loss": 2.412,
+      "step": 2799000
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.306766397645695e-05,
+      "loss": 2.3943,
+      "step": 2799500
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.306642539003087e-05,
+      "loss": 2.3898,
+      "step": 2800000
+    },
+    {
+      "epoch": 13.87,
+      "learning_rate": 4.3065186803604786e-05,
+      "loss": 2.3945,
+      "step": 2800500
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.30639482171787e-05,
+      "loss": 2.4051,
+      "step": 2801000
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.306270963075262e-05,
+      "loss": 2.3813,
+      "step": 2801500
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.306147104432654e-05,
+      "loss": 2.3979,
+      "step": 2802000
+    },
+    {
+      "epoch": 13.88,
+      "learning_rate": 4.3060232457900454e-05,
+      "loss": 2.4034,
+      "step": 2802500
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.305899387147437e-05,
+      "loss": 2.399,
+      "step": 2803000
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.305775528504829e-05,
+      "loss": 2.3926,
+      "step": 2803500
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.3056516698622205e-05,
+      "loss": 2.3909,
+      "step": 2804000
+    },
+    {
+      "epoch": 13.89,
+      "learning_rate": 4.3055278112196115e-05,
+      "loss": 2.4082,
+      "step": 2804500
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.305403952577003e-05,
+      "loss": 2.3697,
+      "step": 2805000
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.30528034165168e-05,
+      "loss": 2.4134,
+      "step": 2805500
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.305156483009072e-05,
+      "loss": 2.3691,
+      "step": 2806000
+    },
+    {
+      "epoch": 13.9,
+      "learning_rate": 4.3050326243664634e-05,
+      "loss": 2.388,
+      "step": 2806500
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.304908765723855e-05,
+      "loss": 2.4274,
+      "step": 2807000
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.304784907081246e-05,
+      "loss": 2.4141,
+      "step": 2807500
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.304661048438638e-05,
+      "loss": 2.3835,
+      "step": 2808000
+    },
+    {
+      "epoch": 13.91,
+      "learning_rate": 4.3045371897960295e-05,
+      "loss": 2.4,
+      "step": 2808500
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.304413331153421e-05,
+      "loss": 2.4225,
+      "step": 2809000
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.304289472510813e-05,
+      "loss": 2.3942,
+      "step": 2809500
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.3041658615854905e-05,
+      "loss": 2.3977,
+      "step": 2810000
+    },
+    {
+      "epoch": 13.92,
+      "learning_rate": 4.304042250660167e-05,
+      "loss": 2.4204,
+      "step": 2810500
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.3039183920175584e-05,
+      "loss": 2.4132,
+      "step": 2811000
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.30379453337495e-05,
+      "loss": 2.4255,
+      "step": 2811500
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.303670674732342e-05,
+      "loss": 2.4183,
+      "step": 2812000
+    },
+    {
+      "epoch": 13.93,
+      "learning_rate": 4.3035468160897334e-05,
+      "loss": 2.3954,
+      "step": 2812500
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.30342320516441e-05,
+      "loss": 2.4053,
+      "step": 2813000
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.303299594239087e-05,
+      "loss": 2.3931,
+      "step": 2813500
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.303175735596479e-05,
+      "loss": 2.4367,
+      "step": 2814000
+    },
+    {
+      "epoch": 13.94,
+      "learning_rate": 4.30305187695387e-05,
+      "loss": 2.4039,
+      "step": 2814500
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.3029280183112616e-05,
+      "loss": 2.3991,
+      "step": 2815000
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.302804159668653e-05,
+      "loss": 2.4093,
+      "step": 2815500
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.302680301026045e-05,
+      "loss": 2.3957,
+      "step": 2816000
+    },
+    {
+      "epoch": 13.95,
+      "learning_rate": 4.302556442383437e-05,
+      "loss": 2.4145,
+      "step": 2816500
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.3024325837408284e-05,
+      "loss": 2.4256,
+      "step": 2817000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.30230872509822e-05,
+      "loss": 2.412,
+      "step": 2817500
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.302184866455612e-05,
+      "loss": 2.412,
+      "step": 2818000
+    },
+    {
+      "epoch": 13.96,
+      "learning_rate": 4.3020612555302886e-05,
+      "loss": 2.4011,
+      "step": 2818500
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.3019376446049655e-05,
+      "loss": 2.4118,
+      "step": 2819000
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.301813785962357e-05,
+      "loss": 2.4149,
+      "step": 2819500
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.301689927319749e-05,
+      "loss": 2.4283,
+      "step": 2820000
+    },
+    {
+      "epoch": 13.97,
+      "learning_rate": 4.3015660686771406e-05,
+      "loss": 2.3776,
+      "step": 2820500
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.301442457751817e-05,
+      "loss": 2.4164,
+      "step": 2821000
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.3013185991092085e-05,
+      "loss": 2.3833,
+      "step": 2821500
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.3011947404666e-05,
+      "loss": 2.384,
+      "step": 2822000
+    },
+    {
+      "epoch": 13.98,
+      "learning_rate": 4.301070881823992e-05,
+      "loss": 2.4177,
+      "step": 2822500
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.3009470231813836e-05,
+      "loss": 2.4214,
+      "step": 2823000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.300823164538775e-05,
+      "loss": 2.4005,
+      "step": 2823500
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.300699305896167e-05,
+      "loss": 2.3818,
+      "step": 2824000
+    },
+    {
+      "epoch": 13.99,
+      "learning_rate": 4.3005754472535587e-05,
+      "loss": 2.3987,
+      "step": 2824500
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.3004515886109503e-05,
+      "loss": 2.4102,
+      "step": 2825000
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.300327729968342e-05,
+      "loss": 2.4001,
+      "step": 2825500
+    },
+    {
+      "epoch": 14.0,
+      "eval_accuracy": 0.6461582624983472,
+      "eval_accuracy_mlm": 0.5998228536990704,
+      "eval_accuracy_nsp": 0.8650881122062763,
+      "eval_loss": 2.389559268951416,
+      "eval_runtime": 145.7903,
+      "eval_samples_per_second": 1748.807,
+      "eval_steps_per_second": 72.872,
+      "step": 2825802
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.300204119043019e-05,
+      "loss": 2.3643,
+      "step": 2826000
+    },
+    {
+      "epoch": 14.0,
+      "learning_rate": 4.3000802604004106e-05,
+      "loss": 2.3761,
+      "step": 2826500
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.299956401757802e-05,
+      "loss": 2.3749,
+      "step": 2827000
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.299832543115194e-05,
+      "loss": 2.3892,
+      "step": 2827500
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.299708684472585e-05,
+      "loss": 2.3667,
+      "step": 2828000
+    },
+    {
+      "epoch": 14.01,
+      "learning_rate": 4.299585073547262e-05,
+      "loss": 2.364,
+      "step": 2828500
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.2994612149046536e-05,
+      "loss": 2.3782,
+      "step": 2829000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.299337356262045e-05,
+      "loss": 2.371,
+      "step": 2829500
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.299213497619437e-05,
+      "loss": 2.349,
+      "step": 2830000
+    },
+    {
+      "epoch": 14.02,
+      "learning_rate": 4.299089638976829e-05,
+      "loss": 2.3903,
+      "step": 2830500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.2989657803342204e-05,
+      "loss": 2.374,
+      "step": 2831000
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.298842169408897e-05,
+      "loss": 2.378,
+      "step": 2831500
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.298718310766289e-05,
+      "loss": 2.3867,
+      "step": 2832000
+    },
+    {
+      "epoch": 14.03,
+      "learning_rate": 4.2985944521236806e-05,
+      "loss": 2.366,
+      "step": 2832500
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.298470593481072e-05,
+      "loss": 2.3584,
+      "step": 2833000
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.298346734838464e-05,
+      "loss": 2.3585,
+      "step": 2833500
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.298222876195856e-05,
+      "loss": 2.3845,
+      "step": 2834000
+    },
+    {
+      "epoch": 14.04,
+      "learning_rate": 4.2980990175532474e-05,
+      "loss": 2.3867,
+      "step": 2834500
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.2979751589106384e-05,
+      "loss": 2.3844,
+      "step": 2835000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.29785130026803e-05,
+      "loss": 2.3945,
+      "step": 2835500
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.297727689342707e-05,
+      "loss": 2.3819,
+      "step": 2836000
+    },
+    {
+      "epoch": 14.05,
+      "learning_rate": 4.297603830700099e-05,
+      "loss": 2.3963,
+      "step": 2836500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.2974799720574904e-05,
+      "loss": 2.3649,
+      "step": 2837000
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.297356113414882e-05,
+      "loss": 2.3759,
+      "step": 2837500
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.297232254772274e-05,
+      "loss": 2.3773,
+      "step": 2838000
+    },
+    {
+      "epoch": 14.06,
+      "learning_rate": 4.2971083961296655e-05,
+      "loss": 2.3649,
+      "step": 2838500
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.296984785204342e-05,
+      "loss": 2.3764,
+      "step": 2839000
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.296860926561734e-05,
+      "loss": 2.3793,
+      "step": 2839500
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.29673731563641e-05,
+      "loss": 2.4066,
+      "step": 2840000
+    },
+    {
+      "epoch": 14.07,
+      "learning_rate": 4.296613704711088e-05,
+      "loss": 2.377,
+      "step": 2840500
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.2964898460684795e-05,
+      "loss": 2.3808,
+      "step": 2841000
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.296365987425871e-05,
+      "loss": 2.4138,
+      "step": 2841500
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.296242128783263e-05,
+      "loss": 2.3763,
+      "step": 2842000
+    },
+    {
+      "epoch": 14.08,
+      "learning_rate": 4.296118270140654e-05,
+      "loss": 2.3747,
+      "step": 2842500
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.2959944114980456e-05,
+      "loss": 2.3829,
+      "step": 2843000
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.295870552855437e-05,
+      "loss": 2.4047,
+      "step": 2843500
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.295746694212829e-05,
+      "loss": 2.3814,
+      "step": 2844000
+    },
+    {
+      "epoch": 14.09,
+      "learning_rate": 4.2956228355702207e-05,
+      "loss": 2.387,
+      "step": 2844500
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.2954989769276123e-05,
+      "loss": 2.3661,
+      "step": 2845000
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.295375118285004e-05,
+      "loss": 2.4036,
+      "step": 2845500
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.29525150735968e-05,
+      "loss": 2.3828,
+      "step": 2846000
+    },
+    {
+      "epoch": 14.1,
+      "learning_rate": 4.295127648717072e-05,
+      "loss": 2.3651,
+      "step": 2846500
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.2950037900744636e-05,
+      "loss": 2.3889,
+      "step": 2847000
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.294879931431855e-05,
+      "loss": 2.3733,
+      "step": 2847500
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.294756072789247e-05,
+      "loss": 2.3593,
+      "step": 2848000
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.294632214146639e-05,
+      "loss": 2.3742,
+      "step": 2848500
+    },
+    {
+      "epoch": 14.11,
+      "learning_rate": 4.2945083555040304e-05,
+      "loss": 2.3745,
+      "step": 2849000
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.294384496861422e-05,
+      "loss": 2.3674,
+      "step": 2849500
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.294260638218814e-05,
+      "loss": 2.3607,
+      "step": 2850000
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.294137027293491e-05,
+      "loss": 2.383,
+      "step": 2850500
+    },
+    {
+      "epoch": 14.12,
+      "learning_rate": 4.2940131686508824e-05,
+      "loss": 2.3761,
+      "step": 2851000
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.293889310008274e-05,
+      "loss": 2.3966,
+      "step": 2851500
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.293765451365666e-05,
+      "loss": 2.3912,
+      "step": 2852000
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.293641840440342e-05,
+      "loss": 2.3719,
+      "step": 2852500
+    },
+    {
+      "epoch": 14.13,
+      "learning_rate": 4.2935179817977336e-05,
+      "loss": 2.3608,
+      "step": 2853000
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.293394123155125e-05,
+      "loss": 2.3762,
+      "step": 2853500
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.293270264512517e-05,
+      "loss": 2.3916,
+      "step": 2854000
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.293146405869909e-05,
+      "loss": 2.3773,
+      "step": 2854500
+    },
+    {
+      "epoch": 14.14,
+      "learning_rate": 4.2930225472273004e-05,
+      "loss": 2.3872,
+      "step": 2855000
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.292898688584692e-05,
+      "loss": 2.3776,
+      "step": 2855500
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.292775077659369e-05,
+      "loss": 2.4074,
+      "step": 2856000
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.292651219016761e-05,
+      "loss": 2.379,
+      "step": 2856500
+    },
+    {
+      "epoch": 14.15,
+      "learning_rate": 4.2925273603741524e-05,
+      "loss": 2.3803,
+      "step": 2857000
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.292403501731544e-05,
+      "loss": 2.3807,
+      "step": 2857500
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.292279643088936e-05,
+      "loss": 2.3591,
+      "step": 2858000
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.2921557844463274e-05,
+      "loss": 2.3725,
+      "step": 2858500
+    },
+    {
+      "epoch": 14.16,
+      "learning_rate": 4.292031925803719e-05,
+      "loss": 2.3891,
+      "step": 2859000
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.291908067161111e-05,
+      "loss": 2.3872,
+      "step": 2859500
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.291784456235787e-05,
+      "loss": 2.3672,
+      "step": 2860000
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.2916608453104646e-05,
+      "loss": 2.3655,
+      "step": 2860500
+    },
+    {
+      "epoch": 14.17,
+      "learning_rate": 4.2915369866678556e-05,
+      "loss": 2.403,
+      "step": 2861000
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.291413375742533e-05,
+      "loss": 2.4014,
+      "step": 2861500
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.291289517099925e-05,
+      "loss": 2.3719,
+      "step": 2862000
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.2911656584573166e-05,
+      "loss": 2.377,
+      "step": 2862500
+    },
+    {
+      "epoch": 14.18,
+      "learning_rate": 4.2910417998147076e-05,
+      "loss": 2.3801,
+      "step": 2863000
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.290917941172099e-05,
+      "loss": 2.3656,
+      "step": 2863500
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.290794082529491e-05,
+      "loss": 2.4083,
+      "step": 2864000
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.2906702238868826e-05,
+      "loss": 2.3794,
+      "step": 2864500
+    },
+    {
+      "epoch": 14.19,
+      "learning_rate": 4.2905463652442743e-05,
+      "loss": 2.3929,
+      "step": 2865000
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.2904225066016654e-05,
+      "loss": 2.3909,
+      "step": 2865500
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.290298647959057e-05,
+      "loss": 2.4093,
+      "step": 2866000
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.290174789316449e-05,
+      "loss": 2.3792,
+      "step": 2866500
+    },
+    {
+      "epoch": 14.2,
+      "learning_rate": 4.2900509306738404e-05,
+      "loss": 2.3973,
+      "step": 2867000
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.289927072031232e-05,
+      "loss": 2.3682,
+      "step": 2867500
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.289803213388624e-05,
+      "loss": 2.3672,
+      "step": 2868000
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.2896793547460155e-05,
+      "loss": 2.3825,
+      "step": 2868500
+    },
+    {
+      "epoch": 14.21,
+      "learning_rate": 4.289555496103407e-05,
+      "loss": 2.3766,
+      "step": 2869000
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.289431885178084e-05,
+      "loss": 2.3758,
+      "step": 2869500
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.289308026535476e-05,
+      "loss": 2.37,
+      "step": 2870000
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.2891844156101527e-05,
+      "loss": 2.3608,
+      "step": 2870500
+    },
+    {
+      "epoch": 14.22,
+      "learning_rate": 4.2890605569675444e-05,
+      "loss": 2.3819,
+      "step": 2871000
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.288936698324936e-05,
+      "loss": 2.3681,
+      "step": 2871500
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.288813087399613e-05,
+      "loss": 2.3852,
+      "step": 2872000
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.2886892287570046e-05,
+      "loss": 2.3943,
+      "step": 2872500
+    },
+    {
+      "epoch": 14.23,
+      "learning_rate": 4.288565370114396e-05,
+      "loss": 2.3918,
+      "step": 2873000
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.288441511471787e-05,
+      "loss": 2.363,
+      "step": 2873500
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.288317652829179e-05,
+      "loss": 2.3689,
+      "step": 2874000
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.2881940419038566e-05,
+      "loss": 2.3561,
+      "step": 2874500
+    },
+    {
+      "epoch": 14.24,
+      "learning_rate": 4.288070183261248e-05,
+      "loss": 2.3556,
+      "step": 2875000
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.28794632461864e-05,
+      "loss": 2.3698,
+      "step": 2875500
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.2878224659760317e-05,
+      "loss": 2.3819,
+      "step": 2876000
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.287698607333423e-05,
+      "loss": 2.3529,
+      "step": 2876500
+    },
+    {
+      "epoch": 14.25,
+      "learning_rate": 4.2875747486908144e-05,
+      "loss": 2.3704,
+      "step": 2877000
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.287450890048206e-05,
+      "loss": 2.399,
+      "step": 2877500
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.287327031405598e-05,
+      "loss": 2.3887,
+      "step": 2878000
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.2872031727629894e-05,
+      "loss": 2.3905,
+      "step": 2878500
+    },
+    {
+      "epoch": 14.26,
+      "learning_rate": 4.2870793141203805e-05,
+      "loss": 2.3566,
+      "step": 2879000
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.286955703195058e-05,
+      "loss": 2.4,
+      "step": 2879500
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.286831844552449e-05,
+      "loss": 2.3926,
+      "step": 2880000
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.286707985909841e-05,
+      "loss": 2.3815,
+      "step": 2880500
+    },
+    {
+      "epoch": 14.27,
+      "learning_rate": 4.2865841272672324e-05,
+      "loss": 2.369,
+      "step": 2881000
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.286460268624624e-05,
+      "loss": 2.3836,
+      "step": 2881500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.286336409982016e-05,
+      "loss": 2.4035,
+      "step": 2882000
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.2862127990566934e-05,
+      "loss": 2.3799,
+      "step": 2882500
+    },
+    {
+      "epoch": 14.28,
+      "learning_rate": 4.2860889404140844e-05,
+      "loss": 2.4322,
+      "step": 2883000
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.285965081771476e-05,
+      "loss": 2.4124,
+      "step": 2883500
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.285841223128868e-05,
+      "loss": 2.3728,
+      "step": 2884000
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.2857173644862595e-05,
+      "loss": 2.3791,
+      "step": 2884500
+    },
+    {
+      "epoch": 14.29,
+      "learning_rate": 4.285593505843651e-05,
+      "loss": 2.3921,
+      "step": 2885000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.285469647201042e-05,
+      "loss": 2.3687,
+      "step": 2885500
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.285346036275719e-05,
+      "loss": 2.3897,
+      "step": 2886000
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.285222177633111e-05,
+      "loss": 2.3735,
+      "step": 2886500
+    },
+    {
+      "epoch": 14.3,
+      "learning_rate": 4.2850983189905024e-05,
+      "loss": 2.3695,
+      "step": 2887000
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.284974460347894e-05,
+      "loss": 2.3833,
+      "step": 2887500
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.284850601705286e-05,
+      "loss": 2.3874,
+      "step": 2888000
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.2847267430626775e-05,
+      "loss": 2.4175,
+      "step": 2888500
+    },
+    {
+      "epoch": 14.31,
+      "learning_rate": 4.284603132137355e-05,
+      "loss": 2.3706,
+      "step": 2889000
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.2844797689293165e-05,
+      "loss": 2.3967,
+      "step": 2889500
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.2843561580039933e-05,
+      "loss": 2.3934,
+      "step": 2890000
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.284232299361385e-05,
+      "loss": 2.3733,
+      "step": 2890500
+    },
+    {
+      "epoch": 14.32,
+      "learning_rate": 4.284108440718777e-05,
+      "loss": 2.3728,
+      "step": 2891000
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.2839845820761684e-05,
+      "loss": 2.3741,
+      "step": 2891500
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.28386072343356e-05,
+      "loss": 2.406,
+      "step": 2892000
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.283736864790952e-05,
+      "loss": 2.3878,
+      "step": 2892500
+    },
+    {
+      "epoch": 14.33,
+      "learning_rate": 4.283613006148343e-05,
+      "loss": 2.3857,
+      "step": 2893000
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.2834891475057345e-05,
+      "loss": 2.3929,
+      "step": 2893500
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.2833655365804114e-05,
+      "loss": 2.3873,
+      "step": 2894000
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.283241925655089e-05,
+      "loss": 2.3801,
+      "step": 2894500
+    },
+    {
+      "epoch": 14.34,
+      "learning_rate": 4.2831180670124806e-05,
+      "loss": 2.3904,
+      "step": 2895000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.282994208369872e-05,
+      "loss": 2.4079,
+      "step": 2895500
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.2828703497272634e-05,
+      "loss": 2.418,
+      "step": 2896000
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.282746491084655e-05,
+      "loss": 2.3692,
+      "step": 2896500
+    },
+    {
+      "epoch": 14.35,
+      "learning_rate": 4.282622632442047e-05,
+      "loss": 2.377,
+      "step": 2897000
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.2824987737994384e-05,
+      "loss": 2.3925,
+      "step": 2897500
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.28237491515683e-05,
+      "loss": 2.3754,
+      "step": 2898000
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.282251056514222e-05,
+      "loss": 2.3814,
+      "step": 2898500
+    },
+    {
+      "epoch": 14.36,
+      "learning_rate": 4.2821271978716135e-05,
+      "loss": 2.3872,
+      "step": 2899000
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.282003339229005e-05,
+      "loss": 2.3753,
+      "step": 2899500
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.281879480586396e-05,
+      "loss": 2.4065,
+      "step": 2900000
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.281755621943788e-05,
+      "loss": 2.3957,
+      "step": 2900500
+    },
+    {
+      "epoch": 14.37,
+      "learning_rate": 4.2816317633011796e-05,
+      "loss": 2.4159,
+      "step": 2901000
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.281507904658571e-05,
+      "loss": 2.4126,
+      "step": 2901500
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.281384046015963e-05,
+      "loss": 2.3595,
+      "step": 2902000
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.281260187373355e-05,
+      "loss": 2.3574,
+      "step": 2902500
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.281136824165317e-05,
+      "loss": 2.3769,
+      "step": 2903000
+    },
+    {
+      "epoch": 14.38,
+      "learning_rate": 4.2810129655227084e-05,
+      "loss": 2.395,
+      "step": 2903500
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.2808891068801e-05,
+      "loss": 2.4074,
+      "step": 2904000
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.280765495954777e-05,
+      "loss": 2.3847,
+      "step": 2904500
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.280641637312169e-05,
+      "loss": 2.4074,
+      "step": 2905000
+    },
+    {
+      "epoch": 14.39,
+      "learning_rate": 4.28051777866956e-05,
+      "loss": 2.3715,
+      "step": 2905500
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.2803939200269514e-05,
+      "loss": 2.3871,
+      "step": 2906000
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.280270061384343e-05,
+      "loss": 2.4013,
+      "step": 2906500
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.280146202741735e-05,
+      "loss": 2.3838,
+      "step": 2907000
+    },
+    {
+      "epoch": 14.4,
+      "learning_rate": 4.2800223440991265e-05,
+      "loss": 2.4154,
+      "step": 2907500
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.279898485456518e-05,
+      "loss": 2.3687,
+      "step": 2908000
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.27977462681391e-05,
+      "loss": 2.396,
+      "step": 2908500
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.2796507681713016e-05,
+      "loss": 2.3786,
+      "step": 2909000
+    },
+    {
+      "epoch": 14.41,
+      "learning_rate": 4.2795271572459785e-05,
+      "loss": 2.3984,
+      "step": 2909500
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.279403546320655e-05,
+      "loss": 2.3943,
+      "step": 2910000
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.279279687678047e-05,
+      "loss": 2.3614,
+      "step": 2910500
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.279155829035439e-05,
+      "loss": 2.3771,
+      "step": 2911000
+    },
+    {
+      "epoch": 14.42,
+      "learning_rate": 4.2790319703928304e-05,
+      "loss": 2.3691,
+      "step": 2911500
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.2789081117502214e-05,
+      "loss": 2.392,
+      "step": 2912000
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.278784253107613e-05,
+      "loss": 2.4161,
+      "step": 2912500
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.278660394465005e-05,
+      "loss": 2.3881,
+      "step": 2913000
+    },
+    {
+      "epoch": 14.43,
+      "learning_rate": 4.2785365358223965e-05,
+      "loss": 2.3568,
+      "step": 2913500
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.278412677179788e-05,
+      "loss": 2.4166,
+      "step": 2914000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.278289066254466e-05,
+      "loss": 2.4008,
+      "step": 2914500
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.278165207611857e-05,
+      "loss": 2.3719,
+      "step": 2915000
+    },
+    {
+      "epoch": 14.44,
+      "learning_rate": 4.2780413489692485e-05,
+      "loss": 2.4054,
+      "step": 2915500
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.27791749032664e-05,
+      "loss": 2.3981,
+      "step": 2916000
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.277793631684032e-05,
+      "loss": 2.4088,
+      "step": 2916500
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.2776697730414235e-05,
+      "loss": 2.4193,
+      "step": 2917000
+    },
+    {
+      "epoch": 14.45,
+      "learning_rate": 4.277545914398815e-05,
+      "loss": 2.3723,
+      "step": 2917500
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.277422055756207e-05,
+      "loss": 2.3896,
+      "step": 2918000
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.2772981971135986e-05,
+      "loss": 2.3889,
+      "step": 2918500
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.27717433847099e-05,
+      "loss": 2.3773,
+      "step": 2919000
+    },
+    {
+      "epoch": 14.46,
+      "learning_rate": 4.277050479828382e-05,
+      "loss": 2.4043,
+      "step": 2919500
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.276927116620344e-05,
+      "loss": 2.3883,
+      "step": 2920000
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.276803257977736e-05,
+      "loss": 2.3812,
+      "step": 2920500
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.276679894769697e-05,
+      "loss": 2.4031,
+      "step": 2921000
+    },
+    {
+      "epoch": 14.47,
+      "learning_rate": 4.276556036127089e-05,
+      "loss": 2.3853,
+      "step": 2921500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.2764321774844806e-05,
+      "loss": 2.4171,
+      "step": 2922000
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.276308318841872e-05,
+      "loss": 2.3994,
+      "step": 2922500
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.276184460199264e-05,
+      "loss": 2.3933,
+      "step": 2923000
+    },
+    {
+      "epoch": 14.48,
+      "learning_rate": 4.276060849273941e-05,
+      "loss": 2.3946,
+      "step": 2923500
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.2759369906313325e-05,
+      "loss": 2.4314,
+      "step": 2924000
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.275813131988724e-05,
+      "loss": 2.3782,
+      "step": 2924500
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.275689273346116e-05,
+      "loss": 2.3963,
+      "step": 2925000
+    },
+    {
+      "epoch": 14.49,
+      "learning_rate": 4.2755654147035076e-05,
+      "loss": 2.4168,
+      "step": 2925500
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.275441556060899e-05,
+      "loss": 2.3955,
+      "step": 2926000
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.275317697418291e-05,
+      "loss": 2.4109,
+      "step": 2926500
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.275193838775683e-05,
+      "loss": 2.3643,
+      "step": 2927000
+    },
+    {
+      "epoch": 14.5,
+      "learning_rate": 4.2750699801330744e-05,
+      "loss": 2.3692,
+      "step": 2927500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.2749461214904654e-05,
+      "loss": 2.4234,
+      "step": 2928000
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.274822262847857e-05,
+      "loss": 2.4136,
+      "step": 2928500
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.274698404205249e-05,
+      "loss": 2.3882,
+      "step": 2929000
+    },
+    {
+      "epoch": 14.51,
+      "learning_rate": 4.2745745455626405e-05,
+      "loss": 2.3713,
+      "step": 2929500
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.274450686920032e-05,
+      "loss": 2.3898,
+      "step": 2930000
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.274326828277423e-05,
+      "loss": 2.4151,
+      "step": 2930500
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.274202969634815e-05,
+      "loss": 2.3902,
+      "step": 2931000
+    },
+    {
+      "epoch": 14.52,
+      "learning_rate": 4.2740793587094924e-05,
+      "loss": 2.3814,
+      "step": 2931500
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.273955500066884e-05,
+      "loss": 2.3855,
+      "step": 2932000
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.273831641424276e-05,
+      "loss": 2.4006,
+      "step": 2932500
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.2737077827816675e-05,
+      "loss": 2.3651,
+      "step": 2933000
+    },
+    {
+      "epoch": 14.53,
+      "learning_rate": 4.2735839241390585e-05,
+      "loss": 2.3846,
+      "step": 2933500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.273460313213736e-05,
+      "loss": 2.3763,
+      "step": 2934000
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.273336454571127e-05,
+      "loss": 2.382,
+      "step": 2934500
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.273212595928519e-05,
+      "loss": 2.3847,
+      "step": 2935000
+    },
+    {
+      "epoch": 14.54,
+      "learning_rate": 4.2730887372859105e-05,
+      "loss": 2.3934,
+      "step": 2935500
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.2729651263605873e-05,
+      "loss": 2.3807,
+      "step": 2936000
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.272841515435264e-05,
+      "loss": 2.3737,
+      "step": 2936500
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.272717656792656e-05,
+      "loss": 2.396,
+      "step": 2937000
+    },
+    {
+      "epoch": 14.55,
+      "learning_rate": 4.2725937981500476e-05,
+      "loss": 2.397,
+      "step": 2937500
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272469939507439e-05,
+      "loss": 2.3834,
+      "step": 2938000
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272346080864831e-05,
+      "loss": 2.383,
+      "step": 2938500
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.272222222222223e-05,
+      "loss": 2.3652,
+      "step": 2939000
+    },
+    {
+      "epoch": 14.56,
+      "learning_rate": 4.2720983635796144e-05,
+      "loss": 2.3763,
+      "step": 2939500
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.271974504937006e-05,
+      "loss": 2.4087,
+      "step": 2940000
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.271850646294398e-05,
+      "loss": 2.3843,
+      "step": 2940500
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.2717267876517895e-05,
+      "loss": 2.3971,
+      "step": 2941000
+    },
+    {
+      "epoch": 14.57,
+      "learning_rate": 4.2716029290091805e-05,
+      "loss": 2.3815,
+      "step": 2941500
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.271479070366572e-05,
+      "loss": 2.4035,
+      "step": 2942000
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.271355211723964e-05,
+      "loss": 2.3709,
+      "step": 2942500
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.2712313530813556e-05,
+      "loss": 2.3951,
+      "step": 2943000
+    },
+    {
+      "epoch": 14.58,
+      "learning_rate": 4.271107494438747e-05,
+      "loss": 2.3758,
+      "step": 2943500
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.270983635796138e-05,
+      "loss": 2.4007,
+      "step": 2944000
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.270860024870816e-05,
+      "loss": 2.4097,
+      "step": 2944500
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.2707361662282075e-05,
+      "loss": 2.4025,
+      "step": 2945000
+    },
+    {
+      "epoch": 14.59,
+      "learning_rate": 4.2706125553028844e-05,
+      "loss": 2.3816,
+      "step": 2945500
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.270488696660276e-05,
+      "loss": 2.3982,
+      "step": 2946000
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.270364838017668e-05,
+      "loss": 2.3793,
+      "step": 2946500
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.2702409793750595e-05,
+      "loss": 2.389,
+      "step": 2947000
+    },
+    {
+      "epoch": 14.6,
+      "learning_rate": 4.270117120732451e-05,
+      "loss": 2.362,
+      "step": 2947500
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.2699935098071274e-05,
+      "loss": 2.3914,
+      "step": 2948000
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.269869651164519e-05,
+      "loss": 2.3902,
+      "step": 2948500
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.269745792521911e-05,
+      "loss": 2.3968,
+      "step": 2949000
+    },
+    {
+      "epoch": 14.61,
+      "learning_rate": 4.2696219338793024e-05,
+      "loss": 2.3966,
+      "step": 2949500
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.269498075236694e-05,
+      "loss": 2.395,
+      "step": 2950000
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.269374216594086e-05,
+      "loss": 2.3799,
+      "step": 2950500
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.2692503579514775e-05,
+      "loss": 2.3837,
+      "step": 2951000
+    },
+    {
+      "epoch": 14.62,
+      "learning_rate": 4.2691267470261544e-05,
+      "loss": 2.3736,
+      "step": 2951500
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.269002888383546e-05,
+      "loss": 2.3967,
+      "step": 2952000
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.268879029740938e-05,
+      "loss": 2.4221,
+      "step": 2952500
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.2687551710983295e-05,
+      "loss": 2.3803,
+      "step": 2953000
+    },
+    {
+      "epoch": 14.63,
+      "learning_rate": 4.268631560173006e-05,
+      "loss": 2.3863,
+      "step": 2953500
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.2685077015303974e-05,
+      "loss": 2.391,
+      "step": 2954000
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.268383842887789e-05,
+      "loss": 2.374,
+      "step": 2954500
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.268259984245181e-05,
+      "loss": 2.3949,
+      "step": 2955000
+    },
+    {
+      "epoch": 14.64,
+      "learning_rate": 4.2681363733198577e-05,
+      "loss": 2.3759,
+      "step": 2955500
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.2680125146772493e-05,
+      "loss": 2.387,
+      "step": 2956000
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.267888656034641e-05,
+      "loss": 2.3696,
+      "step": 2956500
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.267764797392033e-05,
+      "loss": 2.4066,
+      "step": 2957000
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.2676409387494244e-05,
+      "loss": 2.4101,
+      "step": 2957500
+    },
+    {
+      "epoch": 14.65,
+      "learning_rate": 4.267517327824101e-05,
+      "loss": 2.3882,
+      "step": 2958000
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.267393716898778e-05,
+      "loss": 2.4137,
+      "step": 2958500
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.26726985825617e-05,
+      "loss": 2.3827,
+      "step": 2959000
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.267145999613561e-05,
+      "loss": 2.4004,
+      "step": 2959500
+    },
+    {
+      "epoch": 14.66,
+      "learning_rate": 4.2670221409709526e-05,
+      "loss": 2.3833,
+      "step": 2960000
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.26689853004563e-05,
+      "loss": 2.402,
+      "step": 2960500
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.266774671403022e-05,
+      "loss": 2.4208,
+      "step": 2961000
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.2666508127604135e-05,
+      "loss": 2.4217,
+      "step": 2961500
+    },
+    {
+      "epoch": 14.67,
+      "learning_rate": 4.266526954117805e-05,
+      "loss": 2.3977,
+      "step": 2962000
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.266403095475196e-05,
+      "loss": 2.398,
+      "step": 2962500
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.266279236832588e-05,
+      "loss": 2.3978,
+      "step": 2963000
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.2661553781899796e-05,
+      "loss": 2.4067,
+      "step": 2963500
+    },
+    {
+      "epoch": 14.68,
+      "learning_rate": 4.266031519547371e-05,
+      "loss": 2.4094,
+      "step": 2964000
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265907660904763e-05,
+      "loss": 2.3881,
+      "step": 2964500
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265783802262154e-05,
+      "loss": 2.3953,
+      "step": 2965000
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.265659943619546e-05,
+      "loss": 2.3919,
+      "step": 2965500
+    },
+    {
+      "epoch": 14.69,
+      "learning_rate": 4.2655363326942226e-05,
+      "loss": 2.3778,
+      "step": 2966000
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265412474051614e-05,
+      "loss": 2.3833,
+      "step": 2966500
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265288615409006e-05,
+      "loss": 2.3947,
+      "step": 2967000
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.265164756766398e-05,
+      "loss": 2.3884,
+      "step": 2967500
+    },
+    {
+      "epoch": 14.7,
+      "learning_rate": 4.2650408981237894e-05,
+      "loss": 2.3881,
+      "step": 2968000
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.264917039481181e-05,
+      "loss": 2.4038,
+      "step": 2968500
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.264793180838573e-05,
+      "loss": 2.412,
+      "step": 2969000
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.2646693221959644e-05,
+      "loss": 2.3893,
+      "step": 2969500
+    },
+    {
+      "epoch": 14.71,
+      "learning_rate": 4.264545463553356e-05,
+      "loss": 2.38,
+      "step": 2970000
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.264421604910748e-05,
+      "loss": 2.4093,
+      "step": 2970500
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.264297993985425e-05,
+      "loss": 2.3932,
+      "step": 2971000
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.2641741353428164e-05,
+      "loss": 2.4133,
+      "step": 2971500
+    },
+    {
+      "epoch": 14.72,
+      "learning_rate": 4.2640502767002074e-05,
+      "loss": 2.3936,
+      "step": 2972000
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.263926418057599e-05,
+      "loss": 2.405,
+      "step": 2972500
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.263802559414991e-05,
+      "loss": 2.3882,
+      "step": 2973000
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.2636787007723825e-05,
+      "loss": 2.3862,
+      "step": 2973500
+    },
+    {
+      "epoch": 14.73,
+      "learning_rate": 4.263554842129774e-05,
+      "loss": 2.3741,
+      "step": 2974000
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.263430983487166e-05,
+      "loss": 2.4191,
+      "step": 2974500
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.2633071248445576e-05,
+      "loss": 2.4142,
+      "step": 2975000
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.263183266201949e-05,
+      "loss": 2.3749,
+      "step": 2975500
+    },
+    {
+      "epoch": 14.74,
+      "learning_rate": 4.263059407559341e-05,
+      "loss": 2.3808,
+      "step": 2976000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.262935796634018e-05,
+      "loss": 2.3935,
+      "step": 2976500
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.2628119379914095e-05,
+      "loss": 2.3867,
+      "step": 2977000
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.2626883270660864e-05,
+      "loss": 2.3977,
+      "step": 2977500
+    },
+    {
+      "epoch": 14.75,
+      "learning_rate": 4.262564468423478e-05,
+      "loss": 2.3814,
+      "step": 2978000
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.262440609780869e-05,
+      "loss": 2.3813,
+      "step": 2978500
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.262316751138261e-05,
+      "loss": 2.3986,
+      "step": 2979000
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.262193140212938e-05,
+      "loss": 2.3936,
+      "step": 2979500
+    },
+    {
+      "epoch": 14.76,
+      "learning_rate": 4.2620692815703294e-05,
+      "loss": 2.403,
+      "step": 2980000
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.261945422927721e-05,
+      "loss": 2.391,
+      "step": 2980500
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.261821564285113e-05,
+      "loss": 2.4022,
+      "step": 2981000
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.2616977056425045e-05,
+      "loss": 2.3994,
+      "step": 2981500
+    },
+    {
+      "epoch": 14.77,
+      "learning_rate": 4.261573846999896e-05,
+      "loss": 2.4071,
+      "step": 2982000
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.261449988357288e-05,
+      "loss": 2.402,
+      "step": 2982500
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.2613261297146795e-05,
+      "loss": 2.377,
+      "step": 2983000
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.261202271072071e-05,
+      "loss": 2.3869,
+      "step": 2983500
+    },
+    {
+      "epoch": 14.78,
+      "learning_rate": 4.261078412429463e-05,
+      "loss": 2.3832,
+      "step": 2984000
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.26095480150414e-05,
+      "loss": 2.4119,
+      "step": 2984500
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.2608309428615315e-05,
+      "loss": 2.4104,
+      "step": 2985000
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.2607070842189225e-05,
+      "loss": 2.3959,
+      "step": 2985500
+    },
+    {
+      "epoch": 14.79,
+      "learning_rate": 4.260583225576314e-05,
+      "loss": 2.4221,
+      "step": 2986000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.260459366933706e-05,
+      "loss": 2.4011,
+      "step": 2986500
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.2603355082910976e-05,
+      "loss": 2.3937,
+      "step": 2987000
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.260211649648489e-05,
+      "loss": 2.3794,
+      "step": 2987500
+    },
+    {
+      "epoch": 14.8,
+      "learning_rate": 4.260088038723166e-05,
+      "loss": 2.4179,
+      "step": 2988000
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.259964180080558e-05,
+      "loss": 2.4193,
+      "step": 2988500
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.2598403214379496e-05,
+      "loss": 2.3833,
+      "step": 2989000
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.2597167105126264e-05,
+      "loss": 2.3923,
+      "step": 2989500
+    },
+    {
+      "epoch": 14.81,
+      "learning_rate": 4.259592851870018e-05,
+      "loss": 2.394,
+      "step": 2990000
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.25946899322741e-05,
+      "loss": 2.3941,
+      "step": 2990500
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.2593451345848015e-05,
+      "loss": 2.4062,
+      "step": 2991000
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.259221275942193e-05,
+      "loss": 2.3848,
+      "step": 2991500
+    },
+    {
+      "epoch": 14.82,
+      "learning_rate": 4.259097417299584e-05,
+      "loss": 2.3924,
+      "step": 2992000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.258973558656976e-05,
+      "loss": 2.4166,
+      "step": 2992500
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.2588497000143676e-05,
+      "loss": 2.3901,
+      "step": 2993000
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.258725841371759e-05,
+      "loss": 2.4053,
+      "step": 2993500
+    },
+    {
+      "epoch": 14.83,
+      "learning_rate": 4.258601982729151e-05,
+      "loss": 2.3861,
+      "step": 2994000
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258478124086543e-05,
+      "loss": 2.3877,
+      "step": 2994500
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.2583542654439344e-05,
+      "loss": 2.4219,
+      "step": 2995000
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258230654518611e-05,
+      "loss": 2.4288,
+      "step": 2995500
+    },
+    {
+      "epoch": 14.84,
+      "learning_rate": 4.258106795876003e-05,
+      "loss": 2.388,
+      "step": 2996000
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.2579829372333947e-05,
+      "loss": 2.391,
+      "step": 2996500
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.2578590785907863e-05,
+      "loss": 2.4086,
+      "step": 2997000
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.257735219948178e-05,
+      "loss": 2.3844,
+      "step": 2997500
+    },
+    {
+      "epoch": 14.85,
+      "learning_rate": 4.25761136130557e-05,
+      "loss": 2.3956,
+      "step": 2998000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.2574875026629614e-05,
+      "loss": 2.3777,
+      "step": 2998500
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.257363644020353e-05,
+      "loss": 2.3943,
+      "step": 2999000
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.2572402808123145e-05,
+      "loss": 2.3853,
+      "step": 2999500
+    },
+    {
+      "epoch": 14.86,
+      "learning_rate": 4.257116422169706e-05,
+      "loss": 2.388,
+      "step": 3000000
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.256992563527098e-05,
+      "loss": 2.4066,
+      "step": 3000500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.2568687048844896e-05,
+      "loss": 2.3979,
+      "step": 3001000
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.256744846241881e-05,
+      "loss": 2.392,
+      "step": 3001500
+    },
+    {
+      "epoch": 14.87,
+      "learning_rate": 4.256620987599273e-05,
+      "loss": 2.4233,
+      "step": 3002000
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.256497128956665e-05,
+      "loss": 2.3834,
+      "step": 3002500
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.2563732703140564e-05,
+      "loss": 2.3635,
+      "step": 3003000
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.256249659388733e-05,
+      "loss": 2.4039,
+      "step": 3003500
+    },
+    {
+      "epoch": 14.88,
+      "learning_rate": 4.256125800746125e-05,
+      "loss": 2.3884,
+      "step": 3004000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.2560019421035166e-05,
+      "loss": 2.4051,
+      "step": 3004500
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255878083460908e-05,
+      "loss": 2.3943,
+      "step": 3005000
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255754224818299e-05,
+      "loss": 2.4152,
+      "step": 3005500
+    },
+    {
+      "epoch": 14.89,
+      "learning_rate": 4.255630613892976e-05,
+      "loss": 2.3811,
+      "step": 3006000
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.255506755250368e-05,
+      "loss": 2.393,
+      "step": 3006500
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.2553828966077596e-05,
+      "loss": 2.4022,
+      "step": 3007000
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.255259037965151e-05,
+      "loss": 2.3587,
+      "step": 3007500
+    },
+    {
+      "epoch": 14.9,
+      "learning_rate": 4.255135179322543e-05,
+      "loss": 2.4104,
+      "step": 3008000
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.255011320679935e-05,
+      "loss": 2.3866,
+      "step": 3008500
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.2548874620373264e-05,
+      "loss": 2.4138,
+      "step": 3009000
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.254763603394718e-05,
+      "loss": 2.3889,
+      "step": 3009500
+    },
+    {
+      "epoch": 14.91,
+      "learning_rate": 4.254639992469395e-05,
+      "loss": 2.4053,
+      "step": 3010000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.2545161338267866e-05,
+      "loss": 2.3929,
+      "step": 3010500
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.254392275184178e-05,
+      "loss": 2.401,
+      "step": 3011000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.25426841654157e-05,
+      "loss": 2.3921,
+      "step": 3011500
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.254144557898962e-05,
+      "loss": 2.4238,
+      "step": 3012000
+    },
+    {
+      "epoch": 14.92,
+      "learning_rate": 4.254020699256353e-05,
+      "loss": 2.3746,
+      "step": 3012500
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.2538968406137444e-05,
+      "loss": 2.3839,
+      "step": 3013000
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.253773229688421e-05,
+      "loss": 2.3947,
+      "step": 3013500
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.253649371045813e-05,
+      "loss": 2.402,
+      "step": 3014000
+    },
+    {
+      "epoch": 14.93,
+      "learning_rate": 4.253525512403205e-05,
+      "loss": 2.3758,
+      "step": 3014500
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.2534016537605964e-05,
+      "loss": 2.3933,
+      "step": 3015000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.253278290552558e-05,
+      "loss": 2.3876,
+      "step": 3015500
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.2531544319099495e-05,
+      "loss": 2.3971,
+      "step": 3016000
+    },
+    {
+      "epoch": 14.94,
+      "learning_rate": 4.253030573267341e-05,
+      "loss": 2.4107,
+      "step": 3016500
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.252906714624733e-05,
+      "loss": 2.4113,
+      "step": 3017000
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.2527831036994104e-05,
+      "loss": 2.3868,
+      "step": 3017500
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.252659245056802e-05,
+      "loss": 2.3966,
+      "step": 3018000
+    },
+    {
+      "epoch": 14.95,
+      "learning_rate": 4.252535386414194e-05,
+      "loss": 2.3974,
+      "step": 3018500
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252411527771585e-05,
+      "loss": 2.3664,
+      "step": 3019000
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.2522876691289765e-05,
+      "loss": 2.3885,
+      "step": 3019500
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252163810486368e-05,
+      "loss": 2.3988,
+      "step": 3020000
+    },
+    {
+      "epoch": 14.96,
+      "learning_rate": 4.252040199561045e-05,
+      "loss": 2.3894,
+      "step": 3020500
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.251916340918437e-05,
+      "loss": 2.3897,
+      "step": 3021000
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.2517924822758285e-05,
+      "loss": 2.3855,
+      "step": 3021500
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.2516688713505053e-05,
+      "loss": 2.3951,
+      "step": 3022000
+    },
+    {
+      "epoch": 14.97,
+      "learning_rate": 4.251545260425182e-05,
+      "loss": 2.438,
+      "step": 3022500
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.251421401782574e-05,
+      "loss": 2.3995,
+      "step": 3023000
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.2512975431399656e-05,
+      "loss": 2.4106,
+      "step": 3023500
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.251173684497357e-05,
+      "loss": 2.3756,
+      "step": 3024000
+    },
+    {
+      "epoch": 14.98,
+      "learning_rate": 4.2510500735720335e-05,
+      "loss": 2.3789,
+      "step": 3024500
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.250926214929425e-05,
+      "loss": 2.4049,
+      "step": 3025000
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.250802356286817e-05,
+      "loss": 2.363,
+      "step": 3025500
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.2506784976442086e-05,
+      "loss": 2.4208,
+      "step": 3026000
+    },
+    {
+      "epoch": 14.99,
+      "learning_rate": 4.2505546390016e-05,
+      "loss": 2.3791,
+      "step": 3026500
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.250430780358992e-05,
+      "loss": 2.3832,
+      "step": 3027000
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.250306921716384e-05,
+      "loss": 2.3903,
+      "step": 3027500
+    },
+    {
+      "epoch": 15.0,
+      "eval_accuracy": 0.6470159402014201,
+      "eval_accuracy_mlm": 0.6010111516244826,
+      "eval_accuracy_nsp": 0.8638683082377951,
+      "eval_loss": 2.399909496307373,
+      "eval_runtime": 145.758,
+      "eval_samples_per_second": 1749.194,
+      "eval_steps_per_second": 72.888,
+      "step": 3027645
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.2501830630737754e-05,
+      "loss": 2.3627,
+      "step": 3028000
+    },
+    {
+      "epoch": 15.0,
+      "learning_rate": 4.250059204431167e-05,
+      "loss": 2.3456,
+      "step": 3028500
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.249935593505844e-05,
+      "loss": 2.3451,
+      "step": 3029000
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.2498117348632356e-05,
+      "loss": 2.3651,
+      "step": 3029500
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.249687876220627e-05,
+      "loss": 2.3518,
+      "step": 3030000
+    },
+    {
+      "epoch": 15.01,
+      "learning_rate": 4.249564017578019e-05,
+      "loss": 2.3489,
+      "step": 3030500
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 4.249440158935411e-05,
+      "loss": 2.3564,
+      "step": 3031000
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 4.2493163002928024e-05,
+      "loss": 2.3755,
+      "step": 3031500
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 4.249192441650194e-05,
+      "loss": 2.3875,
+      "step": 3032000
+    },
+    {
+      "epoch": 15.02,
+      "learning_rate": 4.249068583007586e-05,
+      "loss": 2.3706,
+      "step": 3032500
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 4.2489447243649775e-05,
+      "loss": 2.3564,
+      "step": 3033000
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 4.2488208657223685e-05,
+      "loss": 2.3672,
+      "step": 3033500
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 4.24869700707976e-05,
+      "loss": 2.3532,
+      "step": 3034000
+    },
+    {
+      "epoch": 15.03,
+      "learning_rate": 4.248573148437152e-05,
+      "loss": 2.3531,
+      "step": 3034500
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 4.2484492897945436e-05,
+      "loss": 2.3578,
+      "step": 3035000
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 4.2483256788692204e-05,
+      "loss": 2.4104,
+      "step": 3035500
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 4.248201820226612e-05,
+      "loss": 2.3724,
+      "step": 3036000
+    },
+    {
+      "epoch": 15.04,
+      "learning_rate": 4.248077961584004e-05,
+      "loss": 2.3303,
+      "step": 3036500
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 4.2479541029413955e-05,
+      "loss": 2.3409,
+      "step": 3037000
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 4.2478302442987865e-05,
+      "loss": 2.3438,
+      "step": 3037500
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 4.247706633373464e-05,
+      "loss": 2.3988,
+      "step": 3038000
+    },
+    {
+      "epoch": 15.05,
+      "learning_rate": 4.247582774730856e-05,
+      "loss": 2.3621,
+      "step": 3038500
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 4.247459163805532e-05,
+      "loss": 2.4002,
+      "step": 3039000
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 4.247335305162924e-05,
+      "loss": 2.3571,
+      "step": 3039500
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 4.2472114465203154e-05,
+      "loss": 2.3597,
+      "step": 3040000
+    },
+    {
+      "epoch": 15.06,
+      "learning_rate": 4.247087587877707e-05,
+      "loss": 2.3572,
+      "step": 3040500
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 4.246963729235099e-05,
+      "loss": 2.391,
+      "step": 3041000
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 4.2468398705924905e-05,
+      "loss": 2.3717,
+      "step": 3041500
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 4.246716011949882e-05,
+      "loss": 2.3924,
+      "step": 3042000
+    },
+    {
+      "epoch": 15.07,
+      "learning_rate": 4.246592153307274e-05,
+      "loss": 2.3618,
+      "step": 3042500
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 4.2464682946646655e-05,
+      "loss": 2.3624,
+      "step": 3043000
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 4.246344436022057e-05,
+      "loss": 2.368,
+      "step": 3043500
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 4.246220577379448e-05,
+      "loss": 2.3958,
+      "step": 3044000
+    },
+    {
+      "epoch": 15.08,
+      "learning_rate": 4.24609671873684e-05,
+      "loss": 2.3759,
+      "step": 3044500
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 4.2459728600942316e-05,
+      "loss": 2.3554,
+      "step": 3045000
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 4.245849496886194e-05,
+      "loss": 2.3552,
+      "step": 3045500
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 4.2457256382435854e-05,
+      "loss": 2.3811,
+      "step": 3046000
+    },
+    {
+      "epoch": 15.09,
+      "learning_rate": 4.245601779600977e-05,
+      "loss": 2.381,
+      "step": 3046500
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 4.245477920958369e-05,
+      "loss": 2.3725,
+      "step": 3047000
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 4.245354310033046e-05,
+      "loss": 2.3783,
+      "step": 3047500
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 4.2452304513904374e-05,
+      "loss": 2.3869,
+      "step": 3048000
+    },
+    {
+      "epoch": 15.1,
+      "learning_rate": 4.245106592747829e-05,
+      "loss": 2.3689,
+      "step": 3048500
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 4.244982734105221e-05,
+      "loss": 2.3686,
+      "step": 3049000
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 4.2448588754626124e-05,
+      "loss": 2.3877,
+      "step": 3049500
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 4.244735016820004e-05,
+      "loss": 2.3803,
+      "step": 3050000
+    },
+    {
+      "epoch": 15.11,
+      "learning_rate": 4.244611158177396e-05,
+      "loss": 2.3927,
+      "step": 3050500
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 4.2444872995347875e-05,
+      "loss": 2.3387,
+      "step": 3051000
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 4.244363688609464e-05,
+      "loss": 2.3727,
+      "step": 3051500
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 4.2442398299668554e-05,
+      "loss": 2.3679,
+      "step": 3052000
+    },
+    {
+      "epoch": 15.12,
+      "learning_rate": 4.244115971324247e-05,
+      "loss": 2.3728,
+      "step": 3052500
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 4.243992112681639e-05,
+      "loss": 2.3888,
+      "step": 3053000
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 4.2438682540390305e-05,
+      "loss": 2.3701,
+      "step": 3053500
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 4.243744395396422e-05,
+      "loss": 2.363,
+      "step": 3054000
+    },
+    {
+      "epoch": 15.13,
+      "learning_rate": 4.243620784471099e-05,
+      "loss": 2.3638,
+      "step": 3054500
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 4.243496925828491e-05,
+      "loss": 2.3703,
+      "step": 3055000
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 4.2433730671858824e-05,
+      "loss": 2.3799,
+      "step": 3055500
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 4.243249208543274e-05,
+      "loss": 2.3817,
+      "step": 3056000
+    },
+    {
+      "epoch": 15.14,
+      "learning_rate": 4.243125349900666e-05,
+      "loss": 2.3796,
+      "step": 3056500
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 4.243001738975342e-05,
+      "loss": 2.3622,
+      "step": 3057000
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 4.242877880332734e-05,
+      "loss": 2.3801,
+      "step": 3057500
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 4.2427542694074106e-05,
+      "loss": 2.3878,
+      "step": 3058000
+    },
+    {
+      "epoch": 15.15,
+      "learning_rate": 4.242630410764802e-05,
+      "loss": 2.3574,
+      "step": 3058500
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 4.242506552122194e-05,
+      "loss": 2.3479,
+      "step": 3059000
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 4.242382693479586e-05,
+      "loss": 2.3529,
+      "step": 3059500
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 4.2422588348369774e-05,
+      "loss": 2.3862,
+      "step": 3060000
+    },
+    {
+      "epoch": 15.16,
+      "learning_rate": 4.242134976194369e-05,
+      "loss": 2.3654,
+      "step": 3060500
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 4.242011117551761e-05,
+      "loss": 2.3518,
+      "step": 3061000
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 4.2418875066264376e-05,
+      "loss": 2.374,
+      "step": 3061500
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 4.241763647983829e-05,
+      "loss": 2.3655,
+      "step": 3062000
+    },
+    {
+      "epoch": 15.17,
+      "learning_rate": 4.241639789341221e-05,
+      "loss": 2.3658,
+      "step": 3062500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 4.241515930698613e-05,
+      "loss": 2.3602,
+      "step": 3063000
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 4.2413920720560044e-05,
+      "loss": 2.3609,
+      "step": 3063500
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 4.2412682134133954e-05,
+      "loss": 2.367,
+      "step": 3064000
+    },
+    {
+      "epoch": 15.18,
+      "learning_rate": 4.241144354770787e-05,
+      "loss": 2.3744,
+      "step": 3064500
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.241020496128179e-05,
+      "loss": 2.3696,
+      "step": 3065000
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.2408966374855705e-05,
+      "loss": 2.3808,
+      "step": 3065500
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.2407730265602474e-05,
+      "loss": 2.3886,
+      "step": 3066000
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.240649167917639e-05,
+      "loss": 2.3662,
+      "step": 3066500
+    },
+    {
+      "epoch": 15.19,
+      "learning_rate": 4.240525309275031e-05,
+      "loss": 2.3931,
+      "step": 3067000
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 4.2404014506324225e-05,
+      "loss": 2.3764,
+      "step": 3067500
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 4.240277591989814e-05,
+      "loss": 2.3745,
+      "step": 3068000
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 4.240153733347206e-05,
+      "loss": 2.3827,
+      "step": 3068500
+    },
+    {
+      "epoch": 15.2,
+      "learning_rate": 4.240030122421883e-05,
+      "loss": 2.3736,
+      "step": 3069000
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 4.2399062637792744e-05,
+      "loss": 2.3926,
+      "step": 3069500
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 4.239782405136666e-05,
+      "loss": 2.3844,
+      "step": 3070000
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 4.239658546494057e-05,
+      "loss": 2.3614,
+      "step": 3070500
+    },
+    {
+      "epoch": 15.21,
+      "learning_rate": 4.239534687851449e-05,
+      "loss": 2.3672,
+      "step": 3071000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.2394108292088405e-05,
+      "loss": 2.3647,
+      "step": 3071500
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.239286970566232e-05,
+      "loss": 2.3832,
+      "step": 3072000
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.239163111923624e-05,
+      "loss": 2.3962,
+      "step": 3072500
+    },
+    {
+      "epoch": 15.22,
+      "learning_rate": 4.239039500998301e-05,
+      "loss": 2.361,
+      "step": 3073000
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 4.238915890072978e-05,
+      "loss": 2.3841,
+      "step": 3073500
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 4.2387920314303694e-05,
+      "loss": 2.3729,
+      "step": 3074000
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 4.238668172787761e-05,
+      "loss": 2.3962,
+      "step": 3074500
+    },
+    {
+      "epoch": 15.23,
+      "learning_rate": 4.238544314145153e-05,
+      "loss": 2.3965,
+      "step": 3075000
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 4.2384204555025444e-05,
+      "loss": 2.3928,
+      "step": 3075500
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 4.238296596859936e-05,
+      "loss": 2.3878,
+      "step": 3076000
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 4.238172985934612e-05,
+      "loss": 2.3677,
+      "step": 3076500
+    },
+    {
+      "epoch": 15.24,
+      "learning_rate": 4.238049127292004e-05,
+      "loss": 2.3718,
+      "step": 3077000
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 4.237925268649396e-05,
+      "loss": 2.3981,
+      "step": 3077500
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 4.2378014100067874e-05,
+      "loss": 2.3709,
+      "step": 3078000
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 4.237677551364179e-05,
+      "loss": 2.4063,
+      "step": 3078500
+    },
+    {
+      "epoch": 15.25,
+      "learning_rate": 4.237553940438856e-05,
+      "loss": 2.3766,
+      "step": 3079000
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 4.237430081796248e-05,
+      "loss": 2.3836,
+      "step": 3079500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 4.2373064708709246e-05,
+      "loss": 2.3741,
+      "step": 3080000
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 4.237182612228316e-05,
+      "loss": 2.3866,
+      "step": 3080500
+    },
+    {
+      "epoch": 15.26,
+      "learning_rate": 4.237058753585708e-05,
+      "loss": 2.3761,
+      "step": 3081000
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 4.2369348949430996e-05,
+      "loss": 2.3606,
+      "step": 3081500
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 4.2368110363004907e-05,
+      "loss": 2.3544,
+      "step": 3082000
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 4.2366871776578824e-05,
+      "loss": 2.3828,
+      "step": 3082500
+    },
+    {
+      "epoch": 15.27,
+      "learning_rate": 4.236563319015274e-05,
+      "loss": 2.3494,
+      "step": 3083000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 4.236439460372666e-05,
+      "loss": 2.3749,
+      "step": 3083500
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 4.2363156017300574e-05,
+      "loss": 2.384,
+      "step": 3084000
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 4.236191990804735e-05,
+      "loss": 2.3721,
+      "step": 3084500
+    },
+    {
+      "epoch": 15.28,
+      "learning_rate": 4.236068132162126e-05,
+      "loss": 2.367,
+      "step": 3085000
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 4.235944273519518e-05,
+      "loss": 2.3508,
+      "step": 3085500
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 4.2358204148769094e-05,
+      "loss": 2.3788,
+      "step": 3086000
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 4.235696556234301e-05,
+      "loss": 2.3785,
+      "step": 3086500
+    },
+    {
+      "epoch": 15.29,
+      "learning_rate": 4.235572697591693e-05,
+      "loss": 2.3908,
+      "step": 3087000
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 4.2354488389490845e-05,
+      "loss": 2.3881,
+      "step": 3087500
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 4.235324980306476e-05,
+      "loss": 2.3729,
+      "step": 3088000
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 4.2352013693811524e-05,
+      "loss": 2.3611,
+      "step": 3088500
+    },
+    {
+      "epoch": 15.3,
+      "learning_rate": 4.235077510738544e-05,
+      "loss": 2.3854,
+      "step": 3089000
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 4.2349538998132216e-05,
+      "loss": 2.3698,
+      "step": 3089500
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 4.234830041170613e-05,
+      "loss": 2.3766,
+      "step": 3090000
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 4.234706182528005e-05,
+      "loss": 2.366,
+      "step": 3090500
+    },
+    {
+      "epoch": 15.31,
+      "learning_rate": 4.234582323885397e-05,
+      "loss": 2.3723,
+      "step": 3091000
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 4.234458465242788e-05,
+      "loss": 2.378,
+      "step": 3091500
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 4.2343346066001794e-05,
+      "loss": 2.3886,
+      "step": 3092000
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 4.234210747957571e-05,
+      "loss": 2.3767,
+      "step": 3092500
+    },
+    {
+      "epoch": 15.32,
+      "learning_rate": 4.234087137032248e-05,
+      "loss": 2.3718,
+      "step": 3093000
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.23396327838964e-05,
+      "loss": 2.3711,
+      "step": 3093500
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.2338394197470314e-05,
+      "loss": 2.3911,
+      "step": 3094000
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.2337155611044224e-05,
+      "loss": 2.3506,
+      "step": 3094500
+    },
+    {
+      "epoch": 15.33,
+      "learning_rate": 4.2335919501791e-05,
+      "loss": 2.3763,
+      "step": 3095000
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 4.2334680915364916e-05,
+      "loss": 2.3951,
+      "step": 3095500
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 4.233344232893883e-05,
+      "loss": 2.3904,
+      "step": 3096000
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 4.233220374251275e-05,
+      "loss": 2.3658,
+      "step": 3096500
+    },
+    {
+      "epoch": 15.34,
+      "learning_rate": 4.233096515608667e-05,
+      "loss": 2.3671,
+      "step": 3097000
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 4.232972656966058e-05,
+      "loss": 2.3921,
+      "step": 3097500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 4.2328487983234494e-05,
+      "loss": 2.3858,
+      "step": 3098000
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 4.232724939680841e-05,
+      "loss": 2.3708,
+      "step": 3098500
+    },
+    {
+      "epoch": 15.35,
+      "learning_rate": 4.232601081038233e-05,
+      "loss": 2.3545,
+      "step": 3099000
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 4.2324772223956245e-05,
+      "loss": 2.3699,
+      "step": 3099500
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 4.232353363753016e-05,
+      "loss": 2.3685,
+      "step": 3100000
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 4.232229505110408e-05,
+      "loss": 2.3597,
+      "step": 3100500
+    },
+    {
+      "epoch": 15.36,
+      "learning_rate": 4.2321056464677996e-05,
+      "loss": 2.4052,
+      "step": 3101000
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 4.231981787825191e-05,
+      "loss": 2.3693,
+      "step": 3101500
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 4.231857929182583e-05,
+      "loss": 2.3889,
+      "step": 3102000
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 4.2317340705399746e-05,
+      "loss": 2.3852,
+      "step": 3102500
+    },
+    {
+      "epoch": 15.37,
+      "learning_rate": 4.2316102118973663e-05,
+      "loss": 2.3749,
+      "step": 3103000
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 4.2314866009720425e-05,
+      "loss": 2.3614,
+      "step": 3103500
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 4.231362742329434e-05,
+      "loss": 2.3763,
+      "step": 3104000
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 4.231238883686826e-05,
+      "loss": 2.3818,
+      "step": 3104500
+    },
+    {
+      "epoch": 15.38,
+      "learning_rate": 4.2311150250442176e-05,
+      "loss": 2.3593,
+      "step": 3105000
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 4.2309914141188945e-05,
+      "loss": 2.3944,
+      "step": 3105500
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 4.2308678031935714e-05,
+      "loss": 2.3794,
+      "step": 3106000
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 4.230743944550963e-05,
+      "loss": 2.3482,
+      "step": 3106500
+    },
+    {
+      "epoch": 15.39,
+      "learning_rate": 4.230620085908355e-05,
+      "loss": 2.3852,
+      "step": 3107000
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 4.2304962272657465e-05,
+      "loss": 2.3996,
+      "step": 3107500
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 4.2303723686231375e-05,
+      "loss": 2.3387,
+      "step": 3108000
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 4.230248509980529e-05,
+      "loss": 2.3473,
+      "step": 3108500
+    },
+    {
+      "epoch": 15.4,
+      "learning_rate": 4.230124899055207e-05,
+      "loss": 2.3551,
+      "step": 3109000
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 4.2300012881298836e-05,
+      "loss": 2.3676,
+      "step": 3109500
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 4.229877429487275e-05,
+      "loss": 2.3606,
+      "step": 3110000
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 4.229753570844667e-05,
+      "loss": 2.3807,
+      "step": 3110500
+    },
+    {
+      "epoch": 15.41,
+      "learning_rate": 4.229629712202059e-05,
+      "loss": 2.3688,
+      "step": 3111000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 4.2295058535594504e-05,
+      "loss": 2.4027,
+      "step": 3111500
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 4.2293819949168414e-05,
+      "loss": 2.3676,
+      "step": 3112000
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 4.229258136274233e-05,
+      "loss": 2.3845,
+      "step": 3112500
+    },
+    {
+      "epoch": 15.42,
+      "learning_rate": 4.229134277631625e-05,
+      "loss": 2.3844,
+      "step": 3113000
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.2290104189890165e-05,
+      "loss": 2.3842,
+      "step": 3113500
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.2288868080636934e-05,
+      "loss": 2.3777,
+      "step": 3114000
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.228762949421085e-05,
+      "loss": 2.3698,
+      "step": 3114500
+    },
+    {
+      "epoch": 15.43,
+      "learning_rate": 4.228639090778477e-05,
+      "loss": 2.3828,
+      "step": 3115000
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 4.2285154798531536e-05,
+      "loss": 2.3369,
+      "step": 3115500
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 4.228391621210545e-05,
+      "loss": 2.3914,
+      "step": 3116000
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 4.228267762567937e-05,
+      "loss": 2.3863,
+      "step": 3116500
+    },
+    {
+      "epoch": 15.44,
+      "learning_rate": 4.228144151642613e-05,
+      "loss": 2.3908,
+      "step": 3117000
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 4.228020293000005e-05,
+      "loss": 2.3652,
+      "step": 3117500
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 4.2278964343573966e-05,
+      "loss": 2.3764,
+      "step": 3118000
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 4.227772575714788e-05,
+      "loss": 2.4005,
+      "step": 3118500
+    },
+    {
+      "epoch": 15.45,
+      "learning_rate": 4.22764871707218e-05,
+      "loss": 2.3834,
+      "step": 3119000
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 4.227525106146857e-05,
+      "loss": 2.3881,
+      "step": 3119500
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 4.2274012475042486e-05,
+      "loss": 2.3676,
+      "step": 3120000
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 4.22727738886164e-05,
+      "loss": 2.3671,
+      "step": 3120500
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 4.227153530219032e-05,
+      "loss": 2.3702,
+      "step": 3121000
+    },
+    {
+      "epoch": 15.46,
+      "learning_rate": 4.2270296715764236e-05,
+      "loss": 2.3961,
+      "step": 3121500
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 4.226905812933815e-05,
+      "loss": 2.3922,
+      "step": 3122000
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 4.226781954291207e-05,
+      "loss": 2.3649,
+      "step": 3122500
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 4.226658095648599e-05,
+      "loss": 2.3649,
+      "step": 3123000
+    },
+    {
+      "epoch": 15.47,
+      "learning_rate": 4.2265342370059904e-05,
+      "loss": 2.393,
+      "step": 3123500
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 4.226410378363382e-05,
+      "loss": 2.3866,
+      "step": 3124000
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 4.226286519720774e-05,
+      "loss": 2.3817,
+      "step": 3124500
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 4.2261626610781655e-05,
+      "loss": 2.3549,
+      "step": 3125000
+    },
+    {
+      "epoch": 15.48,
+      "learning_rate": 4.2260388024355565e-05,
+      "loss": 2.3664,
+      "step": 3125500
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 4.225914943792948e-05,
+      "loss": 2.3663,
+      "step": 3126000
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 4.22579158058491e-05,
+      "loss": 2.3709,
+      "step": 3126500
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 4.225667721942302e-05,
+      "loss": 2.368,
+      "step": 3127000
+    },
+    {
+      "epoch": 15.49,
+      "learning_rate": 4.2255438632996936e-05,
+      "loss": 2.3716,
+      "step": 3127500
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.2254200046570853e-05,
+      "loss": 2.4181,
+      "step": 3128000
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.225296146014477e-05,
+      "loss": 2.3836,
+      "step": 3128500
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.225172287371869e-05,
+      "loss": 2.3661,
+      "step": 3129000
+    },
+    {
+      "epoch": 15.5,
+      "learning_rate": 4.225048676446545e-05,
+      "loss": 2.3753,
+      "step": 3129500
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.2249248178039366e-05,
+      "loss": 2.3983,
+      "step": 3130000
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.224800959161328e-05,
+      "loss": 2.3943,
+      "step": 3130500
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.22467710051872e-05,
+      "loss": 2.3889,
+      "step": 3131000
+    },
+    {
+      "epoch": 15.51,
+      "learning_rate": 4.224553241876112e-05,
+      "loss": 2.3742,
+      "step": 3131500
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.2244293832335034e-05,
+      "loss": 2.3783,
+      "step": 3132000
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.224305524590895e-05,
+      "loss": 2.4068,
+      "step": 3132500
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.224181665948287e-05,
+      "loss": 2.3924,
+      "step": 3133000
+    },
+    {
+      "epoch": 15.52,
+      "learning_rate": 4.2240580550229637e-05,
+      "loss": 2.386,
+      "step": 3133500
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.2239341963803554e-05,
+      "loss": 2.3976,
+      "step": 3134000
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.223810337737747e-05,
+      "loss": 2.369,
+      "step": 3134500
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.223686479095139e-05,
+      "loss": 2.3854,
+      "step": 3135000
+    },
+    {
+      "epoch": 15.53,
+      "learning_rate": 4.2235626204525304e-05,
+      "loss": 2.392,
+      "step": 3135500
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.223438761809922e-05,
+      "loss": 2.3792,
+      "step": 3136000
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.223314903167314e-05,
+      "loss": 2.357,
+      "step": 3136500
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.2231910445247055e-05,
+      "loss": 2.3669,
+      "step": 3137000
+    },
+    {
+      "epoch": 15.54,
+      "learning_rate": 4.223067185882097e-05,
+      "loss": 2.3775,
+      "step": 3137500
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.2229435749567734e-05,
+      "loss": 2.3768,
+      "step": 3138000
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.222819716314165e-05,
+      "loss": 2.3868,
+      "step": 3138500
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.222695857671557e-05,
+      "loss": 2.3805,
+      "step": 3139000
+    },
+    {
+      "epoch": 15.55,
+      "learning_rate": 4.2225719990289485e-05,
+      "loss": 2.3855,
+      "step": 3139500
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.22244814038634e-05,
+      "loss": 2.364,
+      "step": 3140000
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.222324281743732e-05,
+      "loss": 2.3673,
+      "step": 3140500
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.222200670818409e-05,
+      "loss": 2.3848,
+      "step": 3141000
+    },
+    {
+      "epoch": 15.56,
+      "learning_rate": 4.2220768121758004e-05,
+      "loss": 2.3857,
+      "step": 3141500
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.221953201250477e-05,
+      "loss": 2.3872,
+      "step": 3142000
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.2218293426078683e-05,
+      "loss": 2.3764,
+      "step": 3142500
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.22170548396526e-05,
+      "loss": 2.3927,
+      "step": 3143000
+    },
+    {
+      "epoch": 15.57,
+      "learning_rate": 4.221581625322652e-05,
+      "loss": 2.3675,
+      "step": 3143500
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.2214577666800434e-05,
+      "loss": 2.3898,
+      "step": 3144000
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.221333908037435e-05,
+      "loss": 2.3677,
+      "step": 3144500
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.221210049394827e-05,
+      "loss": 2.4042,
+      "step": 3145000
+    },
+    {
+      "epoch": 15.58,
+      "learning_rate": 4.2210861907522185e-05,
+      "loss": 2.3829,
+      "step": 3145500
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.2209625798268954e-05,
+      "loss": 2.3929,
+      "step": 3146000
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.220838721184287e-05,
+      "loss": 2.3803,
+      "step": 3146500
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.220714862541679e-05,
+      "loss": 2.3651,
+      "step": 3147000
+    },
+    {
+      "epoch": 15.59,
+      "learning_rate": 4.2205910038990705e-05,
+      "loss": 2.4064,
+      "step": 3147500
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.220467392973747e-05,
+      "loss": 2.3791,
+      "step": 3148000
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.220343534331139e-05,
+      "loss": 2.402,
+      "step": 3148500
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.22021967568853e-05,
+      "loss": 2.3714,
+      "step": 3149000
+    },
+    {
+      "epoch": 15.6,
+      "learning_rate": 4.220095817045922e-05,
+      "loss": 2.3842,
+      "step": 3149500
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.2199719584033134e-05,
+      "loss": 2.3855,
+      "step": 3150000
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.21984834747799e-05,
+      "loss": 2.4002,
+      "step": 3150500
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.219724488835382e-05,
+      "loss": 2.3798,
+      "step": 3151000
+    },
+    {
+      "epoch": 15.61,
+      "learning_rate": 4.219600630192774e-05,
+      "loss": 2.3679,
+      "step": 3151500
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.2194770192674506e-05,
+      "loss": 2.3694,
+      "step": 3152000
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.219353160624842e-05,
+      "loss": 2.3997,
+      "step": 3152500
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.219229301982234e-05,
+      "loss": 2.3808,
+      "step": 3153000
+    },
+    {
+      "epoch": 15.62,
+      "learning_rate": 4.2191054433396257e-05,
+      "loss": 2.3822,
+      "step": 3153500
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.2189815846970173e-05,
+      "loss": 2.3799,
+      "step": 3154000
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.218857726054409e-05,
+      "loss": 2.3874,
+      "step": 3154500
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.218733867411801e-05,
+      "loss": 2.387,
+      "step": 3155000
+    },
+    {
+      "epoch": 15.63,
+      "learning_rate": 4.2186100087691924e-05,
+      "loss": 2.3953,
+      "step": 3155500
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.2184861501265834e-05,
+      "loss": 2.3654,
+      "step": 3156000
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.218362291483975e-05,
+      "loss": 2.3823,
+      "step": 3156500
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.218238432841367e-05,
+      "loss": 2.3889,
+      "step": 3157000
+    },
+    {
+      "epoch": 15.64,
+      "learning_rate": 4.218114821916044e-05,
+      "loss": 2.3922,
+      "step": 3157500
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.2179909632734354e-05,
+      "loss": 2.395,
+      "step": 3158000
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.217867352348112e-05,
+      "loss": 2.3761,
+      "step": 3158500
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.217743493705504e-05,
+      "loss": 2.4031,
+      "step": 3159000
+    },
+    {
+      "epoch": 15.65,
+      "learning_rate": 4.217619635062896e-05,
+      "loss": 2.3753,
+      "step": 3159500
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.2174957764202874e-05,
+      "loss": 2.3735,
+      "step": 3160000
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.217371917777679e-05,
+      "loss": 2.3785,
+      "step": 3160500
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.217248059135071e-05,
+      "loss": 2.3862,
+      "step": 3161000
+    },
+    {
+      "epoch": 15.66,
+      "learning_rate": 4.217124448209747e-05,
+      "loss": 2.3566,
+      "step": 3161500
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.2170008372844245e-05,
+      "loss": 2.3831,
+      "step": 3162000
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.216876978641816e-05,
+      "loss": 2.3836,
+      "step": 3162500
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.216753119999208e-05,
+      "loss": 2.3862,
+      "step": 3163000
+    },
+    {
+      "epoch": 15.67,
+      "learning_rate": 4.2166292613565996e-05,
+      "loss": 2.3876,
+      "step": 3163500
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.2165054027139906e-05,
+      "loss": 2.3782,
+      "step": 3164000
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.216381544071382e-05,
+      "loss": 2.3743,
+      "step": 3164500
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.216257685428774e-05,
+      "loss": 2.3789,
+      "step": 3165000
+    },
+    {
+      "epoch": 15.68,
+      "learning_rate": 4.216133826786166e-05,
+      "loss": 2.3919,
+      "step": 3165500
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.2160099681435574e-05,
+      "loss": 2.3701,
+      "step": 3166000
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.215886109500949e-05,
+      "loss": 2.3719,
+      "step": 3166500
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.215762250858341e-05,
+      "loss": 2.3928,
+      "step": 3167000
+    },
+    {
+      "epoch": 15.69,
+      "learning_rate": 4.2156383922157325e-05,
+      "loss": 2.3984,
+      "step": 3167500
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.215514533573124e-05,
+      "loss": 2.3644,
+      "step": 3168000
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.215390674930516e-05,
+      "loss": 2.3855,
+      "step": 3168500
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.2152668162879075e-05,
+      "loss": 2.3751,
+      "step": 3169000
+    },
+    {
+      "epoch": 15.7,
+      "learning_rate": 4.2151429576452985e-05,
+      "loss": 2.3593,
+      "step": 3169500
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.2150193467199754e-05,
+      "loss": 2.3574,
+      "step": 3170000
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.214895488077367e-05,
+      "loss": 2.3706,
+      "step": 3170500
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.214771629434759e-05,
+      "loss": 2.3809,
+      "step": 3171000
+    },
+    {
+      "epoch": 15.71,
+      "learning_rate": 4.214648018509436e-05,
+      "loss": 2.3843,
+      "step": 3171500
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.2145241598668274e-05,
+      "loss": 2.3904,
+      "step": 3172000
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.214400301224219e-05,
+      "loss": 2.38,
+      "step": 3172500
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.214276442581611e-05,
+      "loss": 2.3769,
+      "step": 3173000
+    },
+    {
+      "epoch": 15.72,
+      "learning_rate": 4.2141525839390025e-05,
+      "loss": 2.3672,
+      "step": 3173500
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.214028725296394e-05,
+      "loss": 2.4197,
+      "step": 3174000
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.213904866653786e-05,
+      "loss": 2.3741,
+      "step": 3174500
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.2137810080111775e-05,
+      "loss": 2.3853,
+      "step": 3175000
+    },
+    {
+      "epoch": 15.73,
+      "learning_rate": 4.213657149368569e-05,
+      "loss": 2.3921,
+      "step": 3175500
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.213533290725961e-05,
+      "loss": 2.3486,
+      "step": 3176000
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.213409679800637e-05,
+      "loss": 2.3594,
+      "step": 3176500
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.213285821158029e-05,
+      "loss": 2.3819,
+      "step": 3177000
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.2131619625154205e-05,
+      "loss": 2.3789,
+      "step": 3177500
+    },
+    {
+      "epoch": 15.74,
+      "learning_rate": 4.213038103872812e-05,
+      "loss": 2.3747,
+      "step": 3178000
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.212914245230204e-05,
+      "loss": 2.3711,
+      "step": 3178500
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.2127903865875956e-05,
+      "loss": 2.3928,
+      "step": 3179000
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.2126667756622725e-05,
+      "loss": 2.3843,
+      "step": 3179500
+    },
+    {
+      "epoch": 15.75,
+      "learning_rate": 4.212543164736949e-05,
+      "loss": 2.4232,
+      "step": 3180000
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.2124193060943404e-05,
+      "loss": 2.3803,
+      "step": 3180500
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.212295447451732e-05,
+      "loss": 2.3913,
+      "step": 3181000
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.212171588809124e-05,
+      "loss": 2.3512,
+      "step": 3181500
+    },
+    {
+      "epoch": 15.76,
+      "learning_rate": 4.2120477301665155e-05,
+      "loss": 2.3715,
+      "step": 3182000
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.211923871523907e-05,
+      "loss": 2.3803,
+      "step": 3182500
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.211800012881299e-05,
+      "loss": 2.3765,
+      "step": 3183000
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.211676401955976e-05,
+      "loss": 2.3654,
+      "step": 3183500
+    },
+    {
+      "epoch": 15.77,
+      "learning_rate": 4.2115525433133674e-05,
+      "loss": 2.3893,
+      "step": 3184000
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.211428684670759e-05,
+      "loss": 2.3841,
+      "step": 3184500
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.211304826028151e-05,
+      "loss": 2.3901,
+      "step": 3185000
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.2111809673855425e-05,
+      "loss": 2.3614,
+      "step": 3185500
+    },
+    {
+      "epoch": 15.78,
+      "learning_rate": 4.211057108742934e-05,
+      "loss": 2.3858,
+      "step": 3186000
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.210933250100326e-05,
+      "loss": 2.3817,
+      "step": 3186500
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.210809886892288e-05,
+      "loss": 2.3566,
+      "step": 3187000
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.2106860282496796e-05,
+      "loss": 2.3557,
+      "step": 3187500
+    },
+    {
+      "epoch": 15.79,
+      "learning_rate": 4.210562169607071e-05,
+      "loss": 2.3717,
+      "step": 3188000
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.210438310964463e-05,
+      "loss": 2.3864,
+      "step": 3188500
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.210314452321854e-05,
+      "loss": 2.3699,
+      "step": 3189000
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.210190593679246e-05,
+      "loss": 2.4,
+      "step": 3189500
+    },
+    {
+      "epoch": 15.8,
+      "learning_rate": 4.2100667350366374e-05,
+      "loss": 2.3853,
+      "step": 3190000
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.209942876394029e-05,
+      "loss": 2.3952,
+      "step": 3190500
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.209819017751421e-05,
+      "loss": 2.3948,
+      "step": 3191000
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.2096951591088125e-05,
+      "loss": 2.3522,
+      "step": 3191500
+    },
+    {
+      "epoch": 15.81,
+      "learning_rate": 4.209571300466204e-05,
+      "loss": 2.3915,
+      "step": 3192000
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.209447441823596e-05,
+      "loss": 2.3876,
+      "step": 3192500
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.2093235831809876e-05,
+      "loss": 2.4019,
+      "step": 3193000
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.209199724538379e-05,
+      "loss": 2.3881,
+      "step": 3193500
+    },
+    {
+      "epoch": 15.82,
+      "learning_rate": 4.209075865895771e-05,
+      "loss": 2.4025,
+      "step": 3194000
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.208952254970447e-05,
+      "loss": 2.3753,
+      "step": 3194500
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.208828644045124e-05,
+      "loss": 2.4118,
+      "step": 3195000
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.208704785402516e-05,
+      "loss": 2.3851,
+      "step": 3195500
+    },
+    {
+      "epoch": 15.83,
+      "learning_rate": 4.2085809267599074e-05,
+      "loss": 2.3801,
+      "step": 3196000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.208457315834585e-05,
+      "loss": 2.394,
+      "step": 3196500
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.208333457191977e-05,
+      "loss": 2.4099,
+      "step": 3197000
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.208209598549368e-05,
+      "loss": 2.3875,
+      "step": 3197500
+    },
+    {
+      "epoch": 15.84,
+      "learning_rate": 4.2080857399067594e-05,
+      "loss": 2.3896,
+      "step": 3198000
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.207961881264151e-05,
+      "loss": 2.3944,
+      "step": 3198500
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.207838022621543e-05,
+      "loss": 2.3874,
+      "step": 3199000
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.2077141639789345e-05,
+      "loss": 2.3929,
+      "step": 3199500
+    },
+    {
+      "epoch": 15.85,
+      "learning_rate": 4.2075903053363255e-05,
+      "loss": 2.3987,
+      "step": 3200000
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.207466694411003e-05,
+      "loss": 2.3844,
+      "step": 3200500
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.207342835768395e-05,
+      "loss": 2.3909,
+      "step": 3201000
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.207218977125786e-05,
+      "loss": 2.3669,
+      "step": 3201500
+    },
+    {
+      "epoch": 15.86,
+      "learning_rate": 4.2070951184831774e-05,
+      "loss": 2.3722,
+      "step": 3202000
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.206971259840569e-05,
+      "loss": 2.3922,
+      "step": 3202500
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.206847401197961e-05,
+      "loss": 2.3647,
+      "step": 3203000
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.2067235425553525e-05,
+      "loss": 2.3798,
+      "step": 3203500
+    },
+    {
+      "epoch": 15.87,
+      "learning_rate": 4.206599683912744e-05,
+      "loss": 2.3789,
+      "step": 3204000
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.206475825270136e-05,
+      "loss": 2.3835,
+      "step": 3204500
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.2063519666275276e-05,
+      "loss": 2.4027,
+      "step": 3205000
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.206228107984919e-05,
+      "loss": 2.3981,
+      "step": 3205500
+    },
+    {
+      "epoch": 15.88,
+      "learning_rate": 4.206104249342311e-05,
+      "loss": 2.3786,
+      "step": 3206000
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.205980390699703e-05,
+      "loss": 2.3917,
+      "step": 3206500
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.2058565320570944e-05,
+      "loss": 2.381,
+      "step": 3207000
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.2057329211317706e-05,
+      "loss": 2.4097,
+      "step": 3207500
+    },
+    {
+      "epoch": 15.89,
+      "learning_rate": 4.205609062489162e-05,
+      "loss": 2.37,
+      "step": 3208000
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.205485203846554e-05,
+      "loss": 2.3886,
+      "step": 3208500
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.2053613452039457e-05,
+      "loss": 2.3526,
+      "step": 3209000
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.2052374865613374e-05,
+      "loss": 2.3746,
+      "step": 3209500
+    },
+    {
+      "epoch": 15.9,
+      "learning_rate": 4.205113875636014e-05,
+      "loss": 2.397,
+      "step": 3210000
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.204990016993406e-05,
+      "loss": 2.3853,
+      "step": 3210500
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.2048661583507976e-05,
+      "loss": 2.4041,
+      "step": 3211000
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.204742299708189e-05,
+      "loss": 2.3686,
+      "step": 3211500
+    },
+    {
+      "epoch": 15.91,
+      "learning_rate": 4.204618441065581e-05,
+      "loss": 2.3785,
+      "step": 3212000
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.204494582422973e-05,
+      "loss": 2.4122,
+      "step": 3212500
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.2043707237803644e-05,
+      "loss": 2.3736,
+      "step": 3213000
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.204246865137756e-05,
+      "loss": 2.3955,
+      "step": 3213500
+    },
+    {
+      "epoch": 15.92,
+      "learning_rate": 4.204123006495148e-05,
+      "loss": 2.4139,
+      "step": 3214000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.2039991478525395e-05,
+      "loss": 2.385,
+      "step": 3214500
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.203875536927216e-05,
+      "loss": 2.3506,
+      "step": 3215000
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.2037516782846074e-05,
+      "loss": 2.3841,
+      "step": 3215500
+    },
+    {
+      "epoch": 15.93,
+      "learning_rate": 4.203627819641999e-05,
+      "loss": 2.3781,
+      "step": 3216000
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.203503960999391e-05,
+      "loss": 2.3828,
+      "step": 3216500
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.2033801023567824e-05,
+      "loss": 2.402,
+      "step": 3217000
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.203256243714174e-05,
+      "loss": 2.3728,
+      "step": 3217500
+    },
+    {
+      "epoch": 15.94,
+      "learning_rate": 4.203132632788851e-05,
+      "loss": 2.3832,
+      "step": 3218000
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.203008774146243e-05,
+      "loss": 2.3773,
+      "step": 3218500
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.2028849155036344e-05,
+      "loss": 2.3783,
+      "step": 3219000
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.202761056861026e-05,
+      "loss": 2.401,
+      "step": 3219500
+    },
+    {
+      "epoch": 15.95,
+      "learning_rate": 4.202637445935702e-05,
+      "loss": 2.4113,
+      "step": 3220000
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.202513587293094e-05,
+      "loss": 2.3905,
+      "step": 3220500
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.202389728650486e-05,
+      "loss": 2.3798,
+      "step": 3221000
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.2022658700078774e-05,
+      "loss": 2.3802,
+      "step": 3221500
+    },
+    {
+      "epoch": 15.96,
+      "learning_rate": 4.202142011365269e-05,
+      "loss": 2.3822,
+      "step": 3222000
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.202018152722661e-05,
+      "loss": 2.3768,
+      "step": 3222500
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.2018942940800525e-05,
+      "loss": 2.373,
+      "step": 3223000
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.201770683154729e-05,
+      "loss": 2.3979,
+      "step": 3223500
+    },
+    {
+      "epoch": 15.97,
+      "learning_rate": 4.201646824512121e-05,
+      "loss": 2.4065,
+      "step": 3224000
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.201522965869513e-05,
+      "loss": 2.3761,
+      "step": 3224500
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.2013991072269044e-05,
+      "loss": 2.3741,
+      "step": 3225000
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.201275248584296e-05,
+      "loss": 2.354,
+      "step": 3225500
+    },
+    {
+      "epoch": 15.98,
+      "learning_rate": 4.201151637658973e-05,
+      "loss": 2.3871,
+      "step": 3226000
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.201027779016365e-05,
+      "loss": 2.3685,
+      "step": 3226500
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.200903920373756e-05,
+      "loss": 2.3788,
+      "step": 3227000
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.2007805571657184e-05,
+      "loss": 2.365,
+      "step": 3227500
+    },
+    {
+      "epoch": 15.99,
+      "learning_rate": 4.20065669852311e-05,
+      "loss": 2.3865,
+      "step": 3228000
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.200532839880502e-05,
+      "loss": 2.3693,
+      "step": 3228500
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.200409228955178e-05,
+      "loss": 2.3664,
+      "step": 3229000
+    },
+    {
+      "epoch": 16.0,
+      "eval_accuracy": 0.6492868439902124,
+      "eval_accuracy_mlm": 0.6034055016175841,
+      "eval_accuracy_nsp": 0.8654450323385329,
+      "eval_loss": 2.379127025604248,
+      "eval_runtime": 145.9487,
+      "eval_samples_per_second": 1746.908,
+      "eval_steps_per_second": 72.793,
+      "step": 3229488
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.20028537031257e-05,
+      "loss": 2.3991,
+      "step": 3229500
+    },
+    {
+      "epoch": 16.0,
+      "learning_rate": 4.2001615116699614e-05,
+      "loss": 2.3311,
+      "step": 3230000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.200037653027353e-05,
+      "loss": 2.379,
+      "step": 3230500
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.199913794384745e-05,
+      "loss": 2.3483,
+      "step": 3231000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.1997899357421365e-05,
+      "loss": 2.3625,
+      "step": 3231500
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.199666077099528e-05,
+      "loss": 2.3739,
+      "step": 3232000
+    },
+    {
+      "epoch": 16.01,
+      "learning_rate": 4.199542218456919e-05,
+      "loss": 2.3595,
+      "step": 3232500
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.199418359814311e-05,
+      "loss": 2.3672,
+      "step": 3233000
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.1992947488889885e-05,
+      "loss": 2.3369,
+      "step": 3233500
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.19917089024638e-05,
+      "loss": 2.3675,
+      "step": 3234000
+    },
+    {
+      "epoch": 16.02,
+      "learning_rate": 4.199047031603772e-05,
+      "loss": 2.3456,
+      "step": 3234500
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.1989231729611635e-05,
+      "loss": 2.345,
+      "step": 3235000
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.1987993143185545e-05,
+      "loss": 2.3621,
+      "step": 3235500
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.198675455675946e-05,
+      "loss": 2.3692,
+      "step": 3236000
+    },
+    {
+      "epoch": 16.03,
+      "learning_rate": 4.198551597033338e-05,
+      "loss": 2.3583,
+      "step": 3236500
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.1984277383907296e-05,
+      "loss": 2.3604,
+      "step": 3237000
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.198303879748121e-05,
+      "loss": 2.3367,
+      "step": 3237500
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.198180268822798e-05,
+      "loss": 2.3517,
+      "step": 3238000
+    },
+    {
+      "epoch": 16.04,
+      "learning_rate": 4.19805641018019e-05,
+      "loss": 2.3562,
+      "step": 3238500
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.197932551537581e-05,
+      "loss": 2.3516,
+      "step": 3239000
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.1978086928949726e-05,
+      "loss": 2.3629,
+      "step": 3239500
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.197684834252364e-05,
+      "loss": 2.3508,
+      "step": 3240000
+    },
+    {
+      "epoch": 16.05,
+      "learning_rate": 4.197560975609756e-05,
+      "loss": 2.3546,
+      "step": 3240500
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.1974373646844335e-05,
+      "loss": 2.3317,
+      "step": 3241000
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.19731375375911e-05,
+      "loss": 2.3723,
+      "step": 3241500
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.1971898951165014e-05,
+      "loss": 2.3662,
+      "step": 3242000
+    },
+    {
+      "epoch": 16.06,
+      "learning_rate": 4.197066036473893e-05,
+      "loss": 2.3793,
+      "step": 3242500
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.196942177831285e-05,
+      "loss": 2.3432,
+      "step": 3243000
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.196818566905962e-05,
+      "loss": 2.357,
+      "step": 3243500
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.1966947082633534e-05,
+      "loss": 2.3268,
+      "step": 3244000
+    },
+    {
+      "epoch": 16.07,
+      "learning_rate": 4.196570849620745e-05,
+      "loss": 2.3552,
+      "step": 3244500
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.196446990978137e-05,
+      "loss": 2.3761,
+      "step": 3245000
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.1963231323355285e-05,
+      "loss": 2.3717,
+      "step": 3245500
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.19619927369292e-05,
+      "loss": 2.3475,
+      "step": 3246000
+    },
+    {
+      "epoch": 16.08,
+      "learning_rate": 4.196075415050312e-05,
+      "loss": 2.3493,
+      "step": 3246500
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.1959515564077036e-05,
+      "loss": 2.3632,
+      "step": 3247000
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.1958279454823804e-05,
+      "loss": 2.3624,
+      "step": 3247500
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.1957040868397715e-05,
+      "loss": 2.3572,
+      "step": 3248000
+    },
+    {
+      "epoch": 16.09,
+      "learning_rate": 4.195580228197163e-05,
+      "loss": 2.3751,
+      "step": 3248500
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.195456369554555e-05,
+      "loss": 2.3589,
+      "step": 3249000
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.195332758629232e-05,
+      "loss": 2.3459,
+      "step": 3249500
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.1952088999866234e-05,
+      "loss": 2.3791,
+      "step": 3250000
+    },
+    {
+      "epoch": 16.1,
+      "learning_rate": 4.195085041344015e-05,
+      "loss": 2.3528,
+      "step": 3250500
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.194961182701407e-05,
+      "loss": 2.3381,
+      "step": 3251000
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.1948373240587985e-05,
+      "loss": 2.3431,
+      "step": 3251500
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.19471346541619e-05,
+      "loss": 2.3478,
+      "step": 3252000
+    },
+    {
+      "epoch": 16.11,
+      "learning_rate": 4.194589606773582e-05,
+      "loss": 2.3681,
+      "step": 3252500
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.1944657481309736e-05,
+      "loss": 2.3455,
+      "step": 3253000
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.194341889488365e-05,
+      "loss": 2.3379,
+      "step": 3253500
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.194218030845757e-05,
+      "loss": 2.3439,
+      "step": 3254000
+    },
+    {
+      "epoch": 16.12,
+      "learning_rate": 4.194094172203148e-05,
+      "loss": 2.3919,
+      "step": 3254500
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.193970561277825e-05,
+      "loss": 2.3615,
+      "step": 3255000
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.1938467026352165e-05,
+      "loss": 2.3455,
+      "step": 3255500
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.193722843992608e-05,
+      "loss": 2.3646,
+      "step": 3256000
+    },
+    {
+      "epoch": 16.13,
+      "learning_rate": 4.19359898535e-05,
+      "loss": 2.3457,
+      "step": 3256500
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.1934751267073916e-05,
+      "loss": 2.364,
+      "step": 3257000
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.1933512680647826e-05,
+      "loss": 2.3615,
+      "step": 3257500
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.193227409422174e-05,
+      "loss": 2.3757,
+      "step": 3258000
+    },
+    {
+      "epoch": 16.14,
+      "learning_rate": 4.193104046214137e-05,
+      "loss": 2.3399,
+      "step": 3258500
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.192980187571529e-05,
+      "loss": 2.3593,
+      "step": 3259000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.192856576646205e-05,
+      "loss": 2.378,
+      "step": 3259500
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.192732718003597e-05,
+      "loss": 2.362,
+      "step": 3260000
+    },
+    {
+      "epoch": 16.15,
+      "learning_rate": 4.1926088593609884e-05,
+      "loss": 2.381,
+      "step": 3260500
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.19248500071838e-05,
+      "loss": 2.3708,
+      "step": 3261000
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.192361142075772e-05,
+      "loss": 2.3763,
+      "step": 3261500
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.1922372834331634e-05,
+      "loss": 2.3587,
+      "step": 3262000
+    },
+    {
+      "epoch": 16.16,
+      "learning_rate": 4.192113424790555e-05,
+      "loss": 2.3468,
+      "step": 3262500
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.191989566147947e-05,
+      "loss": 2.3629,
+      "step": 3263000
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.1918657075053385e-05,
+      "loss": 2.3905,
+      "step": 3263500
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.1917423442973006e-05,
+      "loss": 2.3469,
+      "step": 3264000
+    },
+    {
+      "epoch": 16.17,
+      "learning_rate": 4.191618485654692e-05,
+      "loss": 2.3685,
+      "step": 3264500
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.191494627012083e-05,
+      "loss": 2.3798,
+      "step": 3265000
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.191370768369475e-05,
+      "loss": 2.368,
+      "step": 3265500
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.191246909726867e-05,
+      "loss": 2.3569,
+      "step": 3266000
+    },
+    {
+      "epoch": 16.18,
+      "learning_rate": 4.1911230510842584e-05,
+      "loss": 2.3642,
+      "step": 3266500
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.19099919244165e-05,
+      "loss": 2.339,
+      "step": 3267000
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.190875333799042e-05,
+      "loss": 2.392,
+      "step": 3267500
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.1907514751564335e-05,
+      "loss": 2.3535,
+      "step": 3268000
+    },
+    {
+      "epoch": 16.19,
+      "learning_rate": 4.190627616513825e-05,
+      "loss": 2.3608,
+      "step": 3268500
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.190504005588502e-05,
+      "loss": 2.3881,
+      "step": 3269000
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.190380146945894e-05,
+      "loss": 2.354,
+      "step": 3269500
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.1902562883032854e-05,
+      "loss": 2.3587,
+      "step": 3270000
+    },
+    {
+      "epoch": 16.2,
+      "learning_rate": 4.190132429660677e-05,
+      "loss": 2.375,
+      "step": 3270500
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.190008571018069e-05,
+      "loss": 2.3826,
+      "step": 3271000
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.1898847123754605e-05,
+      "loss": 2.3526,
+      "step": 3271500
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.189760853732852e-05,
+      "loss": 2.3849,
+      "step": 3272000
+    },
+    {
+      "epoch": 16.21,
+      "learning_rate": 4.189636995090244e-05,
+      "loss": 2.3827,
+      "step": 3272500
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.1895131364476356e-05,
+      "loss": 2.3505,
+      "step": 3273000
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.189389277805027e-05,
+      "loss": 2.3436,
+      "step": 3273500
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.189265419162419e-05,
+      "loss": 2.3404,
+      "step": 3274000
+    },
+    {
+      "epoch": 16.22,
+      "learning_rate": 4.1891415605198106e-05,
+      "loss": 2.3904,
+      "step": 3274500
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.189017701877202e-05,
+      "loss": 2.3878,
+      "step": 3275000
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.188894338669164e-05,
+      "loss": 2.372,
+      "step": 3275500
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.1887704800265554e-05,
+      "loss": 2.3541,
+      "step": 3276000
+    },
+    {
+      "epoch": 16.23,
+      "learning_rate": 4.188646621383947e-05,
+      "loss": 2.3633,
+      "step": 3276500
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.188522762741339e-05,
+      "loss": 2.3915,
+      "step": 3277000
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.1883989040987305e-05,
+      "loss": 2.3363,
+      "step": 3277500
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.188275045456122e-05,
+      "loss": 2.3568,
+      "step": 3278000
+    },
+    {
+      "epoch": 16.24,
+      "learning_rate": 4.188151186813514e-05,
+      "loss": 2.3588,
+      "step": 3278500
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.18802757588819e-05,
+      "loss": 2.3645,
+      "step": 3279000
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.187903717245582e-05,
+      "loss": 2.3591,
+      "step": 3279500
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.1877798586029735e-05,
+      "loss": 2.3448,
+      "step": 3280000
+    },
+    {
+      "epoch": 16.25,
+      "learning_rate": 4.187655999960365e-05,
+      "loss": 2.3665,
+      "step": 3280500
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.187532389035042e-05,
+      "loss": 2.3366,
+      "step": 3281000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.187408530392434e-05,
+      "loss": 2.3512,
+      "step": 3281500
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.1872846717498254e-05,
+      "loss": 2.3426,
+      "step": 3282000
+    },
+    {
+      "epoch": 16.26,
+      "learning_rate": 4.187160813107217e-05,
+      "loss": 2.3714,
+      "step": 3282500
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.187036954464609e-05,
+      "loss": 2.3823,
+      "step": 3283000
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.186913343539286e-05,
+      "loss": 2.3549,
+      "step": 3283500
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.1867894848966774e-05,
+      "loss": 2.3734,
+      "step": 3284000
+    },
+    {
+      "epoch": 16.27,
+      "learning_rate": 4.186665626254069e-05,
+      "loss": 2.372,
+      "step": 3284500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.18654176761146e-05,
+      "loss": 2.3525,
+      "step": 3285000
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.186417908968852e-05,
+      "loss": 2.3635,
+      "step": 3285500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.1862940503262435e-05,
+      "loss": 2.3709,
+      "step": 3286000
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.186170191683635e-05,
+      "loss": 2.364,
+      "step": 3286500
+    },
+    {
+      "epoch": 16.28,
+      "learning_rate": 4.186046333041027e-05,
+      "loss": 2.3777,
+      "step": 3287000
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.185922722115704e-05,
+      "loss": 2.3548,
+      "step": 3287500
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.1857988634730954e-05,
+      "loss": 2.3503,
+      "step": 3288000
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.185675004830487e-05,
+      "loss": 2.3702,
+      "step": 3288500
+    },
+    {
+      "epoch": 16.29,
+      "learning_rate": 4.185551146187879e-05,
+      "loss": 2.3839,
+      "step": 3289000
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.185427535262556e-05,
+      "loss": 2.3612,
+      "step": 3289500
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.1853039243372326e-05,
+      "loss": 2.3803,
+      "step": 3290000
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.185180065694624e-05,
+      "loss": 2.3574,
+      "step": 3290500
+    },
+    {
+      "epoch": 16.3,
+      "learning_rate": 4.185056207052016e-05,
+      "loss": 2.3609,
+      "step": 3291000
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.184932348409408e-05,
+      "loss": 2.3628,
+      "step": 3291500
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.1848087374840846e-05,
+      "loss": 2.3595,
+      "step": 3292000
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.1846851265587614e-05,
+      "loss": 2.3629,
+      "step": 3292500
+    },
+    {
+      "epoch": 16.31,
+      "learning_rate": 4.1845612679161525e-05,
+      "loss": 2.3337,
+      "step": 3293000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.184437409273544e-05,
+      "loss": 2.3881,
+      "step": 3293500
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.184313550630936e-05,
+      "loss": 2.3534,
+      "step": 3294000
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.184189939705613e-05,
+      "loss": 2.4006,
+      "step": 3294500
+    },
+    {
+      "epoch": 16.32,
+      "learning_rate": 4.1840660810630044e-05,
+      "loss": 2.351,
+      "step": 3295000
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.183942222420396e-05,
+      "loss": 2.3661,
+      "step": 3295500
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.183818363777788e-05,
+      "loss": 2.3705,
+      "step": 3296000
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.1836945051351795e-05,
+      "loss": 2.3444,
+      "step": 3296500
+    },
+    {
+      "epoch": 16.33,
+      "learning_rate": 4.183570646492571e-05,
+      "loss": 2.3706,
+      "step": 3297000
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.183446787849963e-05,
+      "loss": 2.3554,
+      "step": 3297500
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.1833229292073546e-05,
+      "loss": 2.372,
+      "step": 3298000
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.183199070564746e-05,
+      "loss": 2.3748,
+      "step": 3298500
+    },
+    {
+      "epoch": 16.34,
+      "learning_rate": 4.183075211922138e-05,
+      "loss": 2.3779,
+      "step": 3299000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.1829513532795296e-05,
+      "loss": 2.3628,
+      "step": 3299500
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.182827494636921e-05,
+      "loss": 2.3664,
+      "step": 3300000
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.182703635994313e-05,
+      "loss": 2.3823,
+      "step": 3300500
+    },
+    {
+      "epoch": 16.35,
+      "learning_rate": 4.182579777351705e-05,
+      "loss": 2.3607,
+      "step": 3301000
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.1824559187090964e-05,
+      "loss": 2.3748,
+      "step": 3301500
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.1823320600664874e-05,
+      "loss": 2.3491,
+      "step": 3302000
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.182208201423879e-05,
+      "loss": 2.3577,
+      "step": 3302500
+    },
+    {
+      "epoch": 16.36,
+      "learning_rate": 4.182084342781271e-05,
+      "loss": 2.3337,
+      "step": 3303000
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.1819604841386625e-05,
+      "loss": 2.3717,
+      "step": 3303500
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.181836625496054e-05,
+      "loss": 2.3637,
+      "step": 3304000
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.181712766853446e-05,
+      "loss": 2.3689,
+      "step": 3304500
+    },
+    {
+      "epoch": 16.37,
+      "learning_rate": 4.181589155928122e-05,
+      "loss": 2.3704,
+      "step": 3305000
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.181465297285514e-05,
+      "loss": 2.3826,
+      "step": 3305500
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.1813414386429055e-05,
+      "loss": 2.3344,
+      "step": 3306000
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.181217580000297e-05,
+      "loss": 2.3534,
+      "step": 3306500
+    },
+    {
+      "epoch": 16.38,
+      "learning_rate": 4.181093721357689e-05,
+      "loss": 2.3628,
+      "step": 3307000
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.1809698627150806e-05,
+      "loss": 2.3458,
+      "step": 3307500
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.180846251789758e-05,
+      "loss": 2.3597,
+      "step": 3308000
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.180722640864434e-05,
+      "loss": 2.3612,
+      "step": 3308500
+    },
+    {
+      "epoch": 16.39,
+      "learning_rate": 4.180598782221826e-05,
+      "loss": 2.348,
+      "step": 3309000
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.180474923579218e-05,
+      "loss": 2.3849,
+      "step": 3309500
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.1803510649366094e-05,
+      "loss": 2.3843,
+      "step": 3310000
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.180227206294001e-05,
+      "loss": 2.3771,
+      "step": 3310500
+    },
+    {
+      "epoch": 16.4,
+      "learning_rate": 4.180103347651393e-05,
+      "loss": 2.3771,
+      "step": 3311000
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.179979489008784e-05,
+      "loss": 2.3497,
+      "step": 3311500
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.1798556303661755e-05,
+      "loss": 2.348,
+      "step": 3312000
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.179732019440853e-05,
+      "loss": 2.3576,
+      "step": 3312500
+    },
+    {
+      "epoch": 16.41,
+      "learning_rate": 4.179608160798245e-05,
+      "loss": 2.3733,
+      "step": 3313000
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.1794843021556364e-05,
+      "loss": 2.3712,
+      "step": 3313500
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.179360443513028e-05,
+      "loss": 2.3685,
+      "step": 3314000
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.179236832587704e-05,
+      "loss": 2.3851,
+      "step": 3314500
+    },
+    {
+      "epoch": 16.42,
+      "learning_rate": 4.179112973945096e-05,
+      "loss": 2.3961,
+      "step": 3315000
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.178989115302488e-05,
+      "loss": 2.4049,
+      "step": 3315500
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.1788652566598794e-05,
+      "loss": 2.3543,
+      "step": 3316000
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.178741398017271e-05,
+      "loss": 2.3707,
+      "step": 3316500
+    },
+    {
+      "epoch": 16.43,
+      "learning_rate": 4.178617539374663e-05,
+      "loss": 2.3482,
+      "step": 3317000
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.178493680732054e-05,
+      "loss": 2.3832,
+      "step": 3317500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.1783700698067314e-05,
+      "loss": 2.4093,
+      "step": 3318000
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.178246458881408e-05,
+      "loss": 2.3686,
+      "step": 3318500
+    },
+    {
+      "epoch": 16.44,
+      "learning_rate": 4.1781226002388e-05,
+      "loss": 2.3364,
+      "step": 3319000
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.177998989313476e-05,
+      "loss": 2.3651,
+      "step": 3319500
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.177875130670868e-05,
+      "loss": 2.3806,
+      "step": 3320000
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.1777512720282595e-05,
+      "loss": 2.3776,
+      "step": 3320500
+    },
+    {
+      "epoch": 16.45,
+      "learning_rate": 4.177627413385651e-05,
+      "loss": 2.3673,
+      "step": 3321000
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.177503554743043e-05,
+      "loss": 2.3572,
+      "step": 3321500
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.1773796961004346e-05,
+      "loss": 2.3464,
+      "step": 3322000
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.177255837457826e-05,
+      "loss": 2.3861,
+      "step": 3322500
+    },
+    {
+      "epoch": 16.46,
+      "learning_rate": 4.177131978815218e-05,
+      "loss": 2.3693,
+      "step": 3323000
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.177008367889895e-05,
+      "loss": 2.362,
+      "step": 3323500
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.1768845092472866e-05,
+      "loss": 2.3583,
+      "step": 3324000
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.176760650604678e-05,
+      "loss": 2.4084,
+      "step": 3324500
+    },
+    {
+      "epoch": 16.47,
+      "learning_rate": 4.17663679196207e-05,
+      "loss": 2.3874,
+      "step": 3325000
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.1765129333194617e-05,
+      "loss": 2.4014,
+      "step": 3325500
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.1763890746768533e-05,
+      "loss": 2.3739,
+      "step": 3326000
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.1762652160342444e-05,
+      "loss": 2.362,
+      "step": 3326500
+    },
+    {
+      "epoch": 16.48,
+      "learning_rate": 4.176141357391636e-05,
+      "loss": 2.3732,
+      "step": 3327000
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 4.176017498749028e-05,
+      "loss": 2.3626,
+      "step": 3327500
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 4.1758936401064194e-05,
+      "loss": 2.3443,
+      "step": 3328000
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 4.175770029181096e-05,
+      "loss": 2.3804,
+      "step": 3328500
+    },
+    {
+      "epoch": 16.49,
+      "learning_rate": 4.175646170538488e-05,
+      "loss": 2.3758,
+      "step": 3329000
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.17552231189588e-05,
+      "loss": 2.3673,
+      "step": 3329500
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.1753984532532714e-05,
+      "loss": 2.3549,
+      "step": 3330000
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.175274594610663e-05,
+      "loss": 2.362,
+      "step": 3330500
+    },
+    {
+      "epoch": 16.5,
+      "learning_rate": 4.175150735968055e-05,
+      "loss": 2.3912,
+      "step": 3331000
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 4.1750268773254465e-05,
+      "loss": 2.3859,
+      "step": 3331500
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 4.174903018682838e-05,
+      "loss": 2.3581,
+      "step": 3332000
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 4.17477916004023e-05,
+      "loss": 2.3842,
+      "step": 3332500
+    },
+    {
+      "epoch": 16.51,
+      "learning_rate": 4.1746553013976216e-05,
+      "loss": 2.3812,
+      "step": 3333000
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 4.1745314427550126e-05,
+      "loss": 2.3868,
+      "step": 3333500
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 4.174407584112404e-05,
+      "loss": 2.3852,
+      "step": 3334000
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 4.174283725469796e-05,
+      "loss": 2.3561,
+      "step": 3334500
+    },
+    {
+      "epoch": 16.52,
+      "learning_rate": 4.1741598668271877e-05,
+      "loss": 2.3751,
+      "step": 3335000
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 4.1740362559018645e-05,
+      "loss": 2.3717,
+      "step": 3335500
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 4.173912397259256e-05,
+      "loss": 2.3852,
+      "step": 3336000
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 4.173788786333933e-05,
+      "loss": 2.4165,
+      "step": 3336500
+    },
+    {
+      "epoch": 16.53,
+      "learning_rate": 4.173664927691325e-05,
+      "loss": 2.3534,
+      "step": 3337000
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 4.1735410690487165e-05,
+      "loss": 2.3673,
+      "step": 3337500
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 4.173417210406108e-05,
+      "loss": 2.3468,
+      "step": 3338000
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 4.1732933517635e-05,
+      "loss": 2.3575,
+      "step": 3338500
+    },
+    {
+      "epoch": 16.54,
+      "learning_rate": 4.1731694931208916e-05,
+      "loss": 2.4043,
+      "step": 3339000
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.1730456344782826e-05,
+      "loss": 2.3744,
+      "step": 3339500
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.172921775835674e-05,
+      "loss": 2.3968,
+      "step": 3340000
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.172797917193066e-05,
+      "loss": 2.389,
+      "step": 3340500
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.172674306267743e-05,
+      "loss": 2.3549,
+      "step": 3341000
+    },
+    {
+      "epoch": 16.55,
+      "learning_rate": 4.17255069534242e-05,
+      "loss": 2.3684,
+      "step": 3341500
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 4.1724268366998114e-05,
+      "loss": 2.379,
+      "step": 3342000
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 4.172302978057203e-05,
+      "loss": 2.3729,
+      "step": 3342500
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 4.172179119414595e-05,
+      "loss": 2.3568,
+      "step": 3343000
+    },
+    {
+      "epoch": 16.56,
+      "learning_rate": 4.172055508489272e-05,
+      "loss": 2.3773,
+      "step": 3343500
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 4.1719316498466634e-05,
+      "loss": 2.3581,
+      "step": 3344000
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 4.171807791204055e-05,
+      "loss": 2.368,
+      "step": 3344500
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 4.171683932561447e-05,
+      "loss": 2.3668,
+      "step": 3345000
+    },
+    {
+      "epoch": 16.57,
+      "learning_rate": 4.1715600739188385e-05,
+      "loss": 2.3714,
+      "step": 3345500
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 4.17143621527623e-05,
+      "loss": 2.3788,
+      "step": 3346000
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 4.171312356633622e-05,
+      "loss": 2.3737,
+      "step": 3346500
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 4.171188497991013e-05,
+      "loss": 2.3877,
+      "step": 3347000
+    },
+    {
+      "epoch": 16.58,
+      "learning_rate": 4.1710646393484046e-05,
+      "loss": 2.359,
+      "step": 3347500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 4.170940780705796e-05,
+      "loss": 2.3579,
+      "step": 3348000
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 4.170817169780473e-05,
+      "loss": 2.3688,
+      "step": 3348500
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 4.170693311137865e-05,
+      "loss": 2.3799,
+      "step": 3349000
+    },
+    {
+      "epoch": 16.59,
+      "learning_rate": 4.170569700212542e-05,
+      "loss": 2.3597,
+      "step": 3349500
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 4.1704458415699334e-05,
+      "loss": 2.3678,
+      "step": 3350000
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 4.170321982927325e-05,
+      "loss": 2.3693,
+      "step": 3350500
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 4.170198124284717e-05,
+      "loss": 2.3535,
+      "step": 3351000
+    },
+    {
+      "epoch": 16.6,
+      "learning_rate": 4.1700742656421085e-05,
+      "loss": 2.3655,
+      "step": 3351500
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 4.169950654716785e-05,
+      "loss": 2.3376,
+      "step": 3352000
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 4.1698267960741764e-05,
+      "loss": 2.3532,
+      "step": 3352500
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 4.169702937431568e-05,
+      "loss": 2.3747,
+      "step": 3353000
+    },
+    {
+      "epoch": 16.61,
+      "learning_rate": 4.16957907878896e-05,
+      "loss": 2.3603,
+      "step": 3353500
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 4.1694552201463515e-05,
+      "loss": 2.377,
+      "step": 3354000
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 4.169331361503743e-05,
+      "loss": 2.3917,
+      "step": 3354500
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 4.16920775057842e-05,
+      "loss": 2.3773,
+      "step": 3355000
+    },
+    {
+      "epoch": 16.62,
+      "learning_rate": 4.169083891935812e-05,
+      "loss": 2.3885,
+      "step": 3355500
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 4.1689600332932034e-05,
+      "loss": 2.3602,
+      "step": 3356000
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 4.16883642236788e-05,
+      "loss": 2.3732,
+      "step": 3356500
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 4.168712563725271e-05,
+      "loss": 2.3674,
+      "step": 3357000
+    },
+    {
+      "epoch": 16.63,
+      "learning_rate": 4.168588705082663e-05,
+      "loss": 2.3854,
+      "step": 3357500
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 4.168464846440055e-05,
+      "loss": 2.3712,
+      "step": 3358000
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 4.1683409877974464e-05,
+      "loss": 2.3552,
+      "step": 3358500
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 4.168217129154838e-05,
+      "loss": 2.3662,
+      "step": 3359000
+    },
+    {
+      "epoch": 16.64,
+      "learning_rate": 4.168093518229515e-05,
+      "loss": 2.3816,
+      "step": 3359500
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 4.1679696595869067e-05,
+      "loss": 2.3567,
+      "step": 3360000
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 4.1678458009442983e-05,
+      "loss": 2.3709,
+      "step": 3360500
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 4.16772194230169e-05,
+      "loss": 2.3764,
+      "step": 3361000
+    },
+    {
+      "epoch": 16.65,
+      "learning_rate": 4.167598083659082e-05,
+      "loss": 2.375,
+      "step": 3361500
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 4.1674742250164734e-05,
+      "loss": 2.3625,
+      "step": 3362000
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 4.167350366373865e-05,
+      "loss": 2.3537,
+      "step": 3362500
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 4.167226507731257e-05,
+      "loss": 2.3476,
+      "step": 3363000
+    },
+    {
+      "epoch": 16.66,
+      "learning_rate": 4.1671026490886485e-05,
+      "loss": 2.4053,
+      "step": 3363500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 4.16697879044604e-05,
+      "loss": 2.3621,
+      "step": 3364000
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 4.166854931803432e-05,
+      "loss": 2.3565,
+      "step": 3364500
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 4.1667310731608236e-05,
+      "loss": 2.4182,
+      "step": 3365000
+    },
+    {
+      "epoch": 16.67,
+      "learning_rate": 4.1666074622355e-05,
+      "loss": 2.378,
+      "step": 3365500
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 4.1664836035928915e-05,
+      "loss": 2.3553,
+      "step": 3366000
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 4.166359744950283e-05,
+      "loss": 2.3641,
+      "step": 3366500
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 4.166235886307675e-05,
+      "loss": 2.3704,
+      "step": 3367000
+    },
+    {
+      "epoch": 16.68,
+      "learning_rate": 4.1661120276650666e-05,
+      "loss": 2.3547,
+      "step": 3367500
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 4.165988169022458e-05,
+      "loss": 2.3821,
+      "step": 3368000
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 4.16586431037985e-05,
+      "loss": 2.3792,
+      "step": 3368500
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 4.165740699454527e-05,
+      "loss": 2.3716,
+      "step": 3369000
+    },
+    {
+      "epoch": 16.69,
+      "learning_rate": 4.1656168408119185e-05,
+      "loss": 2.3796,
+      "step": 3369500
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 4.16549298216931e-05,
+      "loss": 2.3907,
+      "step": 3370000
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 4.165369123526702e-05,
+      "loss": 2.3926,
+      "step": 3370500
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 4.165245512601378e-05,
+      "loss": 2.3875,
+      "step": 3371000
+    },
+    {
+      "epoch": 16.7,
+      "learning_rate": 4.16512165395877e-05,
+      "loss": 2.3632,
+      "step": 3371500
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 4.1649977953161615e-05,
+      "loss": 2.3904,
+      "step": 3372000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 4.1648741843908384e-05,
+      "loss": 2.3739,
+      "step": 3372500
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 4.16475032574823e-05,
+      "loss": 2.3697,
+      "step": 3373000
+    },
+    {
+      "epoch": 16.71,
+      "learning_rate": 4.164626467105622e-05,
+      "loss": 2.3728,
+      "step": 3373500
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 4.164502856180299e-05,
+      "loss": 2.3865,
+      "step": 3374000
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 4.16437899753769e-05,
+      "loss": 2.3485,
+      "step": 3374500
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 4.164255138895082e-05,
+      "loss": 2.3721,
+      "step": 3375000
+    },
+    {
+      "epoch": 16.72,
+      "learning_rate": 4.164131280252474e-05,
+      "loss": 2.3634,
+      "step": 3375500
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 4.1640076693271506e-05,
+      "loss": 2.3643,
+      "step": 3376000
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 4.163883810684542e-05,
+      "loss": 2.3706,
+      "step": 3376500
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 4.163759952041934e-05,
+      "loss": 2.3708,
+      "step": 3377000
+    },
+    {
+      "epoch": 16.73,
+      "learning_rate": 4.163636093399325e-05,
+      "loss": 2.3637,
+      "step": 3377500
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 4.163512234756717e-05,
+      "loss": 2.378,
+      "step": 3378000
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 4.1633883761141084e-05,
+      "loss": 2.3886,
+      "step": 3378500
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 4.1632645174715e-05,
+      "loss": 2.3935,
+      "step": 3379000
+    },
+    {
+      "epoch": 16.74,
+      "learning_rate": 4.163140658828892e-05,
+      "loss": 2.3653,
+      "step": 3379500
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 4.1630168001862835e-05,
+      "loss": 2.3607,
+      "step": 3380000
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 4.162892941543675e-05,
+      "loss": 2.3704,
+      "step": 3380500
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 4.162769082901067e-05,
+      "loss": 2.3722,
+      "step": 3381000
+    },
+    {
+      "epoch": 16.75,
+      "learning_rate": 4.1626452242584585e-05,
+      "loss": 2.3535,
+      "step": 3381500
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 4.16252136561585e-05,
+      "loss": 2.3634,
+      "step": 3382000
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 4.162397754690527e-05,
+      "loss": 2.3759,
+      "step": 3382500
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 4.162273896047919e-05,
+      "loss": 2.3764,
+      "step": 3383000
+    },
+    {
+      "epoch": 16.76,
+      "learning_rate": 4.1621500374053105e-05,
+      "loss": 2.3739,
+      "step": 3383500
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 4.1620261787627015e-05,
+      "loss": 2.3762,
+      "step": 3384000
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 4.161902320120093e-05,
+      "loss": 2.3656,
+      "step": 3384500
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 4.16177870919477e-05,
+      "loss": 2.3756,
+      "step": 3385000
+    },
+    {
+      "epoch": 16.77,
+      "learning_rate": 4.161654850552162e-05,
+      "loss": 2.3657,
+      "step": 3385500
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 4.1615309919095535e-05,
+      "loss": 2.3836,
+      "step": 3386000
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 4.161407133266945e-05,
+      "loss": 2.3893,
+      "step": 3386500
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 4.161283274624337e-05,
+      "loss": 2.3778,
+      "step": 3387000
+    },
+    {
+      "epoch": 16.78,
+      "learning_rate": 4.1611594159817286e-05,
+      "loss": 2.3606,
+      "step": 3387500
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 4.16103555733912e-05,
+      "loss": 2.3618,
+      "step": 3388000
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 4.160911698696512e-05,
+      "loss": 2.3565,
+      "step": 3388500
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 4.160788087771189e-05,
+      "loss": 2.3636,
+      "step": 3389000
+    },
+    {
+      "epoch": 16.79,
+      "learning_rate": 4.1606642291285805e-05,
+      "loss": 2.3672,
+      "step": 3389500
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 4.160540370485972e-05,
+      "loss": 2.3587,
+      "step": 3390000
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 4.160416511843364e-05,
+      "loss": 2.3782,
+      "step": 3390500
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 4.160292653200755e-05,
+      "loss": 2.3666,
+      "step": 3391000
+    },
+    {
+      "epoch": 16.8,
+      "learning_rate": 4.160169042275432e-05,
+      "loss": 2.3514,
+      "step": 3391500
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 4.1600451836328235e-05,
+      "loss": 2.3646,
+      "step": 3392000
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 4.159921324990215e-05,
+      "loss": 2.3601,
+      "step": 3392500
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 4.159797466347607e-05,
+      "loss": 2.3792,
+      "step": 3393000
+    },
+    {
+      "epoch": 16.81,
+      "learning_rate": 4.1596736077049986e-05,
+      "loss": 2.355,
+      "step": 3393500
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 4.1595499967796754e-05,
+      "loss": 2.3727,
+      "step": 3394000
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 4.159426138137067e-05,
+      "loss": 2.3774,
+      "step": 3394500
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 4.159302279494459e-05,
+      "loss": 2.3607,
+      "step": 3395000
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 4.1591784208518505e-05,
+      "loss": 2.3917,
+      "step": 3395500
+    },
+    {
+      "epoch": 16.82,
+      "learning_rate": 4.159054562209242e-05,
+      "loss": 2.3518,
+      "step": 3396000
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 4.158930703566634e-05,
+      "loss": 2.3856,
+      "step": 3396500
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 4.1588068449240256e-05,
+      "loss": 2.387,
+      "step": 3397000
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 4.1586829862814166e-05,
+      "loss": 2.3707,
+      "step": 3397500
+    },
+    {
+      "epoch": 16.83,
+      "learning_rate": 4.158559127638808e-05,
+      "loss": 2.3603,
+      "step": 3398000
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 4.158435764430771e-05,
+      "loss": 2.3707,
+      "step": 3398500
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 4.158311905788163e-05,
+      "loss": 2.3861,
+      "step": 3399000
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 4.158188047145554e-05,
+      "loss": 2.3695,
+      "step": 3399500
+    },
+    {
+      "epoch": 16.84,
+      "learning_rate": 4.1580641885029455e-05,
+      "loss": 2.3408,
+      "step": 3400000
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 4.157940329860337e-05,
+      "loss": 2.3676,
+      "step": 3400500
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 4.157816718935014e-05,
+      "loss": 2.3638,
+      "step": 3401000
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 4.157692860292406e-05,
+      "loss": 2.3828,
+      "step": 3401500
+    },
+    {
+      "epoch": 16.85,
+      "learning_rate": 4.1575690016497974e-05,
+      "loss": 2.3853,
+      "step": 3402000
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 4.1574451430071884e-05,
+      "loss": 2.3722,
+      "step": 3402500
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 4.15732128436458e-05,
+      "loss": 2.3818,
+      "step": 3403000
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 4.157197425721972e-05,
+      "loss": 2.3493,
+      "step": 3403500
+    },
+    {
+      "epoch": 16.86,
+      "learning_rate": 4.1570735670793635e-05,
+      "loss": 2.3795,
+      "step": 3404000
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 4.156949708436755e-05,
+      "loss": 2.3624,
+      "step": 3404500
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 4.156826097511433e-05,
+      "loss": 2.3783,
+      "step": 3405000
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 4.1567022388688245e-05,
+      "loss": 2.379,
+      "step": 3405500
+    },
+    {
+      "epoch": 16.87,
+      "learning_rate": 4.1565783802262155e-05,
+      "loss": 2.3758,
+      "step": 3406000
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 4.156454521583607e-05,
+      "loss": 2.3549,
+      "step": 3406500
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 4.156330662940999e-05,
+      "loss": 2.3986,
+      "step": 3407000
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 4.1562068042983905e-05,
+      "loss": 2.3615,
+      "step": 3407500
+    },
+    {
+      "epoch": 16.88,
+      "learning_rate": 4.156082945655782e-05,
+      "loss": 2.3583,
+      "step": 3408000
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 4.155959334730459e-05,
+      "loss": 2.3854,
+      "step": 3408500
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 4.15583547608785e-05,
+      "loss": 2.3673,
+      "step": 3409000
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 4.155711617445242e-05,
+      "loss": 2.3642,
+      "step": 3409500
+    },
+    {
+      "epoch": 16.89,
+      "learning_rate": 4.1555877588026335e-05,
+      "loss": 2.3299,
+      "step": 3410000
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 4.155463900160025e-05,
+      "loss": 2.3807,
+      "step": 3410500
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 4.155340041517417e-05,
+      "loss": 2.3767,
+      "step": 3411000
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 4.1552161828748086e-05,
+      "loss": 2.3789,
+      "step": 3411500
+    },
+    {
+      "epoch": 16.9,
+      "learning_rate": 4.1550923242322e-05,
+      "loss": 2.3876,
+      "step": 3412000
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 4.154968713306877e-05,
+      "loss": 2.3517,
+      "step": 3412500
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 4.154844854664269e-05,
+      "loss": 2.4007,
+      "step": 3413000
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 4.1547209960216606e-05,
+      "loss": 2.3944,
+      "step": 3413500
+    },
+    {
+      "epoch": 16.91,
+      "learning_rate": 4.154597137379052e-05,
+      "loss": 2.3647,
+      "step": 3414000
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 4.154473278736444e-05,
+      "loss": 2.3501,
+      "step": 3414500
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 4.1543494200938356e-05,
+      "loss": 2.3941,
+      "step": 3415000
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 4.154225561451227e-05,
+      "loss": 2.3809,
+      "step": 3415500
+    },
+    {
+      "epoch": 16.92,
+      "learning_rate": 4.154101702808619e-05,
+      "loss": 2.3663,
+      "step": 3416000
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 4.153978091883295e-05,
+      "loss": 2.3529,
+      "step": 3416500
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 4.153854233240687e-05,
+      "loss": 2.3712,
+      "step": 3417000
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 4.1537306223153645e-05,
+      "loss": 2.3917,
+      "step": 3417500
+    },
+    {
+      "epoch": 16.93,
+      "learning_rate": 4.153606763672756e-05,
+      "loss": 2.3671,
+      "step": 3418000
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 4.153482905030147e-05,
+      "loss": 2.3691,
+      "step": 3418500
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 4.153359046387539e-05,
+      "loss": 2.3647,
+      "step": 3419000
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 4.1532351877449306e-05,
+      "loss": 2.3684,
+      "step": 3419500
+    },
+    {
+      "epoch": 16.94,
+      "learning_rate": 4.153111329102322e-05,
+      "loss": 2.3566,
+      "step": 3420000
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 4.152987718176999e-05,
+      "loss": 2.3672,
+      "step": 3420500
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 4.152863859534391e-05,
+      "loss": 2.383,
+      "step": 3421000
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 4.152740248609068e-05,
+      "loss": 2.3671,
+      "step": 3421500
+    },
+    {
+      "epoch": 16.95,
+      "learning_rate": 4.1526163899664594e-05,
+      "loss": 2.3578,
+      "step": 3422000
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 4.152492531323851e-05,
+      "loss": 2.3837,
+      "step": 3422500
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 4.152368672681243e-05,
+      "loss": 2.3824,
+      "step": 3423000
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 4.1522448140386345e-05,
+      "loss": 2.3624,
+      "step": 3423500
+    },
+    {
+      "epoch": 16.96,
+      "learning_rate": 4.152120955396026e-05,
+      "loss": 2.3848,
+      "step": 3424000
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 4.151997096753417e-05,
+      "loss": 2.379,
+      "step": 3424500
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 4.151873238110809e-05,
+      "loss": 2.3548,
+      "step": 3425000
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 4.1517493794682006e-05,
+      "loss": 2.363,
+      "step": 3425500
+    },
+    {
+      "epoch": 16.97,
+      "learning_rate": 4.151625520825592e-05,
+      "loss": 2.3791,
+      "step": 3426000
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 4.151501662182984e-05,
+      "loss": 2.3776,
+      "step": 3426500
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 4.151378051257661e-05,
+      "loss": 2.3708,
+      "step": 3427000
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 4.1512541926150525e-05,
+      "loss": 2.3777,
+      "step": 3427500
+    },
+    {
+      "epoch": 16.98,
+      "learning_rate": 4.1511303339724436e-05,
+      "loss": 2.3577,
+      "step": 3428000
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 4.151006475329835e-05,
+      "loss": 2.3726,
+      "step": 3428500
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 4.150882616687227e-05,
+      "loss": 2.393,
+      "step": 3429000
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 4.1507590057619045e-05,
+      "loss": 2.356,
+      "step": 3429500
+    },
+    {
+      "epoch": 16.99,
+      "learning_rate": 4.150635147119296e-05,
+      "loss": 2.3563,
+      "step": 3430000
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.150511288476688e-05,
+      "loss": 2.3842,
+      "step": 3430500
+    },
+    {
+      "epoch": 17.0,
+      "learning_rate": 4.150387429834079e-05,
+      "loss": 2.3678,
+      "step": 3431000
+    },
+    {
+      "epoch": 17.0,
+      "eval_accuracy": 0.6498075948776978,
+      "eval_accuracy_mlm": 0.6043656048690601,
+      "eval_accuracy_nsp": 0.8643428943477186,
+      "eval_loss": 2.376680374145508,
+      "eval_runtime": 145.8417,
+      "eval_samples_per_second": 1748.189,
+      "eval_steps_per_second": 72.846,
+      "step": 3431331
     }
   ],
   "max_steps": 20184300,
   "num_train_epochs": 100,
-  "total_flos": 2.6129529363964503e+18,
+  "total_flos": 4.441905477808464e+18,
   "trial_name": null,
   "trial_params": null
 }