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": 34.0,
-  "global_step": 6862662,
+  "epoch": 40.0,
+  "global_step": 8073720,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
@@ -82730,11 +82730,14609 @@
       "eval_samples_per_second": 1750.384,
       "eval_steps_per_second": 72.938,
       "step": 6862662
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.300526399231086e-05,
+      "loss": 2.234,
+      "step": 6863000
+    },
+    {
+      "epoch": 34.0,
+      "learning_rate": 3.300402540588478e-05,
+      "loss": 2.1825,
+      "step": 6863500
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.3002786819458695e-05,
+      "loss": 2.1727,
+      "step": 6864000
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.300154823303261e-05,
+      "loss": 2.1697,
+      "step": 6864500
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.300030964660652e-05,
+      "loss": 2.1944,
+      "step": 6865000
+    },
+    {
+      "epoch": 34.01,
+      "learning_rate": 3.299907106018044e-05,
+      "loss": 2.1777,
+      "step": 6865500
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3.2997832473754355e-05,
+      "loss": 2.1701,
+      "step": 6866000
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3.299659388732827e-05,
+      "loss": 2.2099,
+      "step": 6866500
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3.299536025524789e-05,
+      "loss": 2.1635,
+      "step": 6867000
+    },
+    {
+      "epoch": 34.02,
+      "learning_rate": 3.299412166882181e-05,
+      "loss": 2.173,
+      "step": 6867500
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 3.299288308239573e-05,
+      "loss": 2.1623,
+      "step": 6868000
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 3.2991644495969644e-05,
+      "loss": 2.1848,
+      "step": 6868500
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 3.299040590954356e-05,
+      "loss": 2.1915,
+      "step": 6869000
+    },
+    {
+      "epoch": 34.03,
+      "learning_rate": 3.298916732311748e-05,
+      "loss": 2.193,
+      "step": 6869500
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3.2987928736691395e-05,
+      "loss": 2.2049,
+      "step": 6870000
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3.298669015026531e-05,
+      "loss": 2.1921,
+      "step": 6870500
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3.298545156383923e-05,
+      "loss": 2.1759,
+      "step": 6871000
+    },
+    {
+      "epoch": 34.04,
+      "learning_rate": 3.298421297741314e-05,
+      "loss": 2.1883,
+      "step": 6871500
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 3.298297686815991e-05,
+      "loss": 2.1694,
+      "step": 6872000
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 3.2981738281733824e-05,
+      "loss": 2.1732,
+      "step": 6872500
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 3.298049969530774e-05,
+      "loss": 2.1889,
+      "step": 6873000
+    },
+    {
+      "epoch": 34.05,
+      "learning_rate": 3.297926110888166e-05,
+      "loss": 2.1705,
+      "step": 6873500
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 3.2978022522455575e-05,
+      "loss": 2.1701,
+      "step": 6874000
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 3.2976786413202344e-05,
+      "loss": 2.1949,
+      "step": 6874500
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 3.297554782677626e-05,
+      "loss": 2.1956,
+      "step": 6875000
+    },
+    {
+      "epoch": 34.06,
+      "learning_rate": 3.297430924035018e-05,
+      "loss": 2.1968,
+      "step": 6875500
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3.2973070653924095e-05,
+      "loss": 2.1826,
+      "step": 6876000
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3.297183206749801e-05,
+      "loss": 2.169,
+      "step": 6876500
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3.2970595958244774e-05,
+      "loss": 2.1826,
+      "step": 6877000
+    },
+    {
+      "epoch": 34.07,
+      "learning_rate": 3.296935737181869e-05,
+      "loss": 2.1644,
+      "step": 6877500
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3.296811878539261e-05,
+      "loss": 2.1973,
+      "step": 6878000
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3.2966880198966525e-05,
+      "loss": 2.1765,
+      "step": 6878500
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3.296564161254044e-05,
+      "loss": 2.1869,
+      "step": 6879000
+    },
+    {
+      "epoch": 34.08,
+      "learning_rate": 3.296440302611436e-05,
+      "loss": 2.1799,
+      "step": 6879500
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 3.2963164439688275e-05,
+      "loss": 2.1789,
+      "step": 6880000
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 3.2961925853262185e-05,
+      "loss": 2.1891,
+      "step": 6880500
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 3.29606872668361e-05,
+      "loss": 2.2122,
+      "step": 6881000
+    },
+    {
+      "epoch": 34.09,
+      "learning_rate": 3.295944868041002e-05,
+      "loss": 2.163,
+      "step": 6881500
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 3.2958212571156795e-05,
+      "loss": 2.1841,
+      "step": 6882000
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 3.2956976461903564e-05,
+      "loss": 2.188,
+      "step": 6882500
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 3.2955740352650326e-05,
+      "loss": 2.1696,
+      "step": 6883000
+    },
+    {
+      "epoch": 34.1,
+      "learning_rate": 3.295450176622424e-05,
+      "loss": 2.1982,
+      "step": 6883500
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3.295326317979816e-05,
+      "loss": 2.165,
+      "step": 6884000
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3.2952024593372077e-05,
+      "loss": 2.1517,
+      "step": 6884500
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3.2950786006945993e-05,
+      "loss": 2.1447,
+      "step": 6885000
+    },
+    {
+      "epoch": 34.11,
+      "learning_rate": 3.294954742051991e-05,
+      "loss": 2.1803,
+      "step": 6885500
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3.294830883409383e-05,
+      "loss": 2.1727,
+      "step": 6886000
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3.2947070247667744e-05,
+      "loss": 2.1894,
+      "step": 6886500
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3.2945836615587365e-05,
+      "loss": 2.1625,
+      "step": 6887000
+    },
+    {
+      "epoch": 34.12,
+      "learning_rate": 3.2944598029161275e-05,
+      "loss": 2.1874,
+      "step": 6887500
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 3.294335944273519e-05,
+      "loss": 2.1846,
+      "step": 6888000
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 3.294212333348197e-05,
+      "loss": 2.1903,
+      "step": 6888500
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 3.2940884747055885e-05,
+      "loss": 2.1673,
+      "step": 6889000
+    },
+    {
+      "epoch": 34.13,
+      "learning_rate": 3.29396461606298e-05,
+      "loss": 2.1791,
+      "step": 6889500
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 3.293840757420372e-05,
+      "loss": 2.1863,
+      "step": 6890000
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 3.2937168987777635e-05,
+      "loss": 2.1819,
+      "step": 6890500
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 3.2935930401351545e-05,
+      "loss": 2.2075,
+      "step": 6891000
+    },
+    {
+      "epoch": 34.14,
+      "learning_rate": 3.293469181492546e-05,
+      "loss": 2.2059,
+      "step": 6891500
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 3.293345322849938e-05,
+      "loss": 2.1733,
+      "step": 6892000
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 3.2932214642073296e-05,
+      "loss": 2.1764,
+      "step": 6892500
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 3.293097605564721e-05,
+      "loss": 2.2026,
+      "step": 6893000
+    },
+    {
+      "epoch": 34.15,
+      "learning_rate": 3.292973746922113e-05,
+      "loss": 2.1988,
+      "step": 6893500
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3.292849888279505e-05,
+      "loss": 2.1872,
+      "step": 6894000
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3.2927260296368964e-05,
+      "loss": 2.1832,
+      "step": 6894500
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3.292602170994288e-05,
+      "loss": 2.1525,
+      "step": 6895000
+    },
+    {
+      "epoch": 34.16,
+      "learning_rate": 3.292478560068964e-05,
+      "loss": 2.1841,
+      "step": 6895500
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.292354701426356e-05,
+      "loss": 2.1954,
+      "step": 6896000
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.292230842783748e-05,
+      "loss": 2.2155,
+      "step": 6896500
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.2921069841411394e-05,
+      "loss": 2.1929,
+      "step": 6897000
+    },
+    {
+      "epoch": 34.17,
+      "learning_rate": 3.291983125498531e-05,
+      "loss": 2.1928,
+      "step": 6897500
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3.291859266855923e-05,
+      "loss": 2.2045,
+      "step": 6898000
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3.2917356559305996e-05,
+      "loss": 2.1672,
+      "step": 6898500
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3.291611797287991e-05,
+      "loss": 2.1694,
+      "step": 6899000
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3.291487938645383e-05,
+      "loss": 2.19,
+      "step": 6899500
+    },
+    {
+      "epoch": 34.18,
+      "learning_rate": 3.291364080002775e-05,
+      "loss": 2.1734,
+      "step": 6900000
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 3.291240469077451e-05,
+      "loss": 2.1893,
+      "step": 6900500
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 3.2911166104348426e-05,
+      "loss": 2.16,
+      "step": 6901000
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 3.290992751792234e-05,
+      "loss": 2.1769,
+      "step": 6901500
+    },
+    {
+      "epoch": 34.19,
+      "learning_rate": 3.290868893149626e-05,
+      "loss": 2.1874,
+      "step": 6902000
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3.290745034507018e-05,
+      "loss": 2.1739,
+      "step": 6902500
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3.2906211758644094e-05,
+      "loss": 2.1859,
+      "step": 6903000
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3.290497317221801e-05,
+      "loss": 2.2051,
+      "step": 6903500
+    },
+    {
+      "epoch": 34.2,
+      "learning_rate": 3.290373458579193e-05,
+      "loss": 2.1717,
+      "step": 6904000
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.2902495999365845e-05,
+      "loss": 2.1797,
+      "step": 6904500
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.2901262367285465e-05,
+      "loss": 2.1786,
+      "step": 6905000
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.290002378085938e-05,
+      "loss": 2.1727,
+      "step": 6905500
+    },
+    {
+      "epoch": 34.21,
+      "learning_rate": 3.28987851944333e-05,
+      "loss": 2.1891,
+      "step": 6906000
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 3.289754660800721e-05,
+      "loss": 2.1985,
+      "step": 6906500
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 3.2896308021581126e-05,
+      "loss": 2.1938,
+      "step": 6907000
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 3.28950719123279e-05,
+      "loss": 2.1897,
+      "step": 6907500
+    },
+    {
+      "epoch": 34.22,
+      "learning_rate": 3.289383332590182e-05,
+      "loss": 2.166,
+      "step": 6908000
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3.2892594739475736e-05,
+      "loss": 2.1857,
+      "step": 6908500
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3.2891358630222504e-05,
+      "loss": 2.1884,
+      "step": 6909000
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3.289012004379642e-05,
+      "loss": 2.1808,
+      "step": 6909500
+    },
+    {
+      "epoch": 34.23,
+      "learning_rate": 3.288888145737034e-05,
+      "loss": 2.1817,
+      "step": 6910000
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 3.2887642870944255e-05,
+      "loss": 2.209,
+      "step": 6910500
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 3.2886404284518165e-05,
+      "loss": 2.1976,
+      "step": 6911000
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 3.288516569809208e-05,
+      "loss": 2.2096,
+      "step": 6911500
+    },
+    {
+      "epoch": 34.24,
+      "learning_rate": 3.2883927111666e-05,
+      "loss": 2.1899,
+      "step": 6912000
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.2882688525239916e-05,
+      "loss": 2.1939,
+      "step": 6912500
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.2881449938813826e-05,
+      "loss": 2.1881,
+      "step": 6913000
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.288021135238774e-05,
+      "loss": 2.2121,
+      "step": 6913500
+    },
+    {
+      "epoch": 34.25,
+      "learning_rate": 3.287897276596166e-05,
+      "loss": 2.1759,
+      "step": 6914000
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 3.287773417953558e-05,
+      "loss": 2.1824,
+      "step": 6914500
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 3.287649807028235e-05,
+      "loss": 2.1901,
+      "step": 6915000
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 3.287525948385627e-05,
+      "loss": 2.1768,
+      "step": 6915500
+    },
+    {
+      "epoch": 34.26,
+      "learning_rate": 3.287402089743018e-05,
+      "loss": 2.1994,
+      "step": 6916000
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3.28727823110041e-05,
+      "loss": 2.1775,
+      "step": 6916500
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3.287154867892372e-05,
+      "loss": 2.2061,
+      "step": 6917000
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3.2870310092497634e-05,
+      "loss": 2.1988,
+      "step": 6917500
+    },
+    {
+      "epoch": 34.27,
+      "learning_rate": 3.286907150607155e-05,
+      "loss": 2.174,
+      "step": 6918000
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 3.286783291964547e-05,
+      "loss": 2.1943,
+      "step": 6918500
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 3.2866594333219385e-05,
+      "loss": 2.2035,
+      "step": 6919000
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 3.2865358223966154e-05,
+      "loss": 2.1593,
+      "step": 6919500
+    },
+    {
+      "epoch": 34.28,
+      "learning_rate": 3.286411963754007e-05,
+      "loss": 2.1835,
+      "step": 6920000
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.286288105111399e-05,
+      "loss": 2.2108,
+      "step": 6920500
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.2861642464687905e-05,
+      "loss": 2.2148,
+      "step": 6921000
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.286040635543467e-05,
+      "loss": 2.193,
+      "step": 6921500
+    },
+    {
+      "epoch": 34.29,
+      "learning_rate": 3.2859167769008584e-05,
+      "loss": 2.2132,
+      "step": 6922000
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3.28579291825825e-05,
+      "loss": 2.1787,
+      "step": 6922500
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3.285669555050213e-05,
+      "loss": 2.1977,
+      "step": 6923000
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3.2855456964076045e-05,
+      "loss": 2.1799,
+      "step": 6923500
+    },
+    {
+      "epoch": 34.3,
+      "learning_rate": 3.285421837764996e-05,
+      "loss": 2.186,
+      "step": 6924000
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 3.285297979122388e-05,
+      "loss": 2.1851,
+      "step": 6924500
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 3.285174120479779e-05,
+      "loss": 2.1996,
+      "step": 6925000
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 3.285050509554456e-05,
+      "loss": 2.1971,
+      "step": 6925500
+    },
+    {
+      "epoch": 34.31,
+      "learning_rate": 3.2849266509118475e-05,
+      "loss": 2.1959,
+      "step": 6926000
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3.284802792269239e-05,
+      "loss": 2.1869,
+      "step": 6926500
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3.284678933626631e-05,
+      "loss": 2.196,
+      "step": 6927000
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3.2845550749840226e-05,
+      "loss": 2.1712,
+      "step": 6927500
+    },
+    {
+      "epoch": 34.32,
+      "learning_rate": 3.284431216341414e-05,
+      "loss": 2.1767,
+      "step": 6928000
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.284307357698806e-05,
+      "loss": 2.1987,
+      "step": 6928500
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.284183746773483e-05,
+      "loss": 2.2139,
+      "step": 6929000
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.2840598881308745e-05,
+      "loss": 2.1773,
+      "step": 6929500
+    },
+    {
+      "epoch": 34.33,
+      "learning_rate": 3.283936029488266e-05,
+      "loss": 2.1861,
+      "step": 6930000
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3.283812170845658e-05,
+      "loss": 2.1925,
+      "step": 6930500
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3.2836883122030496e-05,
+      "loss": 2.1843,
+      "step": 6931000
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3.283564453560441e-05,
+      "loss": 2.1896,
+      "step": 6931500
+    },
+    {
+      "epoch": 34.34,
+      "learning_rate": 3.283440594917832e-05,
+      "loss": 2.2057,
+      "step": 6932000
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 3.283316983992509e-05,
+      "loss": 2.1789,
+      "step": 6932500
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 3.283193125349901e-05,
+      "loss": 2.1865,
+      "step": 6933000
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 3.2830692667072926e-05,
+      "loss": 2.18,
+      "step": 6933500
+    },
+    {
+      "epoch": 34.35,
+      "learning_rate": 3.282945408064684e-05,
+      "loss": 2.195,
+      "step": 6934000
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3.282821549422076e-05,
+      "loss": 2.2251,
+      "step": 6934500
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3.282697690779467e-05,
+      "loss": 2.176,
+      "step": 6935000
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3.2825738321368587e-05,
+      "loss": 2.1886,
+      "step": 6935500
+    },
+    {
+      "epoch": 34.36,
+      "learning_rate": 3.2824499734942504e-05,
+      "loss": 2.1885,
+      "step": 6936000
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 3.282326114851642e-05,
+      "loss": 2.2061,
+      "step": 6936500
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 3.282202256209034e-05,
+      "loss": 2.2056,
+      "step": 6937000
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 3.282078645283711e-05,
+      "loss": 2.1923,
+      "step": 6937500
+    },
+    {
+      "epoch": 34.37,
+      "learning_rate": 3.2819550343583875e-05,
+      "loss": 2.1916,
+      "step": 6938000
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.281831175715779e-05,
+      "loss": 2.1716,
+      "step": 6938500
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.281707317073171e-05,
+      "loss": 2.1798,
+      "step": 6939000
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.2815834584305626e-05,
+      "loss": 2.1916,
+      "step": 6939500
+    },
+    {
+      "epoch": 34.38,
+      "learning_rate": 3.281459599787954e-05,
+      "loss": 2.1689,
+      "step": 6940000
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3.281335741145346e-05,
+      "loss": 2.227,
+      "step": 6940500
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3.281212377937308e-05,
+      "loss": 2.1959,
+      "step": 6941000
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3.2810885192947e-05,
+      "loss": 2.2158,
+      "step": 6941500
+    },
+    {
+      "epoch": 34.39,
+      "learning_rate": 3.280964660652091e-05,
+      "loss": 2.2076,
+      "step": 6942000
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3.2808408020094824e-05,
+      "loss": 2.2063,
+      "step": 6942500
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3.280716943366874e-05,
+      "loss": 2.2234,
+      "step": 6943000
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3.280593084724266e-05,
+      "loss": 2.2024,
+      "step": 6943500
+    },
+    {
+      "epoch": 34.4,
+      "learning_rate": 3.2804692260816575e-05,
+      "loss": 2.1942,
+      "step": 6944000
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 3.280345367439049e-05,
+      "loss": 2.1817,
+      "step": 6944500
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 3.280221508796441e-05,
+      "loss": 2.2017,
+      "step": 6945000
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 3.2800976501538326e-05,
+      "loss": 2.198,
+      "step": 6945500
+    },
+    {
+      "epoch": 34.41,
+      "learning_rate": 3.279973791511224e-05,
+      "loss": 2.1933,
+      "step": 6946000
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.279849932868616e-05,
+      "loss": 2.2008,
+      "step": 6946500
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.279726074226008e-05,
+      "loss": 2.2,
+      "step": 6947000
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.279602215583399e-05,
+      "loss": 2.174,
+      "step": 6947500
+    },
+    {
+      "epoch": 34.42,
+      "learning_rate": 3.2794783569407904e-05,
+      "loss": 2.1931,
+      "step": 6948000
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3.279354746015468e-05,
+      "loss": 2.1865,
+      "step": 6948500
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3.2792308873728596e-05,
+      "loss": 2.2144,
+      "step": 6949000
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3.279107028730251e-05,
+      "loss": 2.2073,
+      "step": 6949500
+    },
+    {
+      "epoch": 34.43,
+      "learning_rate": 3.278983170087643e-05,
+      "loss": 2.2077,
+      "step": 6950000
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3.278859311445035e-05,
+      "loss": 2.201,
+      "step": 6950500
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3.278735700519711e-05,
+      "loss": 2.1901,
+      "step": 6951000
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3.2786118418771026e-05,
+      "loss": 2.202,
+      "step": 6951500
+    },
+    {
+      "epoch": 34.44,
+      "learning_rate": 3.278487983234494e-05,
+      "loss": 2.184,
+      "step": 6952000
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3.278364124591886e-05,
+      "loss": 2.191,
+      "step": 6952500
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3.278240513666563e-05,
+      "loss": 2.2036,
+      "step": 6953000
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3.2781166550239546e-05,
+      "loss": 2.1638,
+      "step": 6953500
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3.277992796381346e-05,
+      "loss": 2.1686,
+      "step": 6954000
+    },
+    {
+      "epoch": 34.45,
+      "learning_rate": 3.277868937738738e-05,
+      "loss": 2.2152,
+      "step": 6954500
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.2777450790961296e-05,
+      "loss": 2.196,
+      "step": 6955000
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.277621220453521e-05,
+      "loss": 2.2021,
+      "step": 6955500
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.277497361810913e-05,
+      "loss": 2.1876,
+      "step": 6956000
+    },
+    {
+      "epoch": 34.46,
+      "learning_rate": 3.277373503168305e-05,
+      "loss": 2.2034,
+      "step": 6956500
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 3.277249892242981e-05,
+      "loss": 2.1923,
+      "step": 6957000
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 3.2771260336003726e-05,
+      "loss": 2.2079,
+      "step": 6957500
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 3.277002174957764e-05,
+      "loss": 2.1761,
+      "step": 6958000
+    },
+    {
+      "epoch": 34.47,
+      "learning_rate": 3.276878316315156e-05,
+      "loss": 2.1964,
+      "step": 6958500
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3.276754457672548e-05,
+      "loss": 2.1802,
+      "step": 6959000
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3.2766305990299394e-05,
+      "loss": 2.196,
+      "step": 6959500
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3.2765067403873304e-05,
+      "loss": 2.2058,
+      "step": 6960000
+    },
+    {
+      "epoch": 34.48,
+      "learning_rate": 3.276383129462008e-05,
+      "loss": 2.1698,
+      "step": 6960500
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3.276259518536685e-05,
+      "loss": 2.1963,
+      "step": 6961000
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3.2761356598940765e-05,
+      "loss": 2.1675,
+      "step": 6961500
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3.276011801251468e-05,
+      "loss": 2.1949,
+      "step": 6962000
+    },
+    {
+      "epoch": 34.49,
+      "learning_rate": 3.275887942608859e-05,
+      "loss": 2.2113,
+      "step": 6962500
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.275764083966251e-05,
+      "loss": 2.2099,
+      "step": 6963000
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.2756402253236426e-05,
+      "loss": 2.2009,
+      "step": 6963500
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.275516366681034e-05,
+      "loss": 2.1868,
+      "step": 6964000
+    },
+    {
+      "epoch": 34.5,
+      "learning_rate": 3.275392508038426e-05,
+      "loss": 2.2024,
+      "step": 6964500
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3.275268649395818e-05,
+      "loss": 2.1997,
+      "step": 6965000
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3.2751447907532094e-05,
+      "loss": 2.1985,
+      "step": 6965500
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3.275021179827886e-05,
+      "loss": 2.1987,
+      "step": 6966000
+    },
+    {
+      "epoch": 34.51,
+      "learning_rate": 3.274897321185278e-05,
+      "loss": 2.2142,
+      "step": 6966500
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3.27477346254267e-05,
+      "loss": 2.2064,
+      "step": 6967000
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3.2746496039000614e-05,
+      "loss": 2.1865,
+      "step": 6967500
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3.274525745257453e-05,
+      "loss": 2.1711,
+      "step": 6968000
+    },
+    {
+      "epoch": 34.52,
+      "learning_rate": 3.274401886614845e-05,
+      "loss": 2.2059,
+      "step": 6968500
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3.2742780279722364e-05,
+      "loss": 2.199,
+      "step": 6969000
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3.2741544170469126e-05,
+      "loss": 2.2009,
+      "step": 6969500
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3.274030558404304e-05,
+      "loss": 2.1918,
+      "step": 6970000
+    },
+    {
+      "epoch": 34.53,
+      "learning_rate": 3.273906699761696e-05,
+      "loss": 2.1988,
+      "step": 6970500
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3.273782841119088e-05,
+      "loss": 2.2398,
+      "step": 6971000
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3.2736592301937646e-05,
+      "loss": 2.2103,
+      "step": 6971500
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3.273535371551156e-05,
+      "loss": 2.1703,
+      "step": 6972000
+    },
+    {
+      "epoch": 34.54,
+      "learning_rate": 3.273411512908548e-05,
+      "loss": 2.189,
+      "step": 6972500
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3.27328765426594e-05,
+      "loss": 2.21,
+      "step": 6973000
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3.2731637956233314e-05,
+      "loss": 2.201,
+      "step": 6973500
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3.273040184698008e-05,
+      "loss": 2.2171,
+      "step": 6974000
+    },
+    {
+      "epoch": 34.55,
+      "learning_rate": 3.2729163260554e-05,
+      "loss": 2.2048,
+      "step": 6974500
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 3.2727924674127916e-05,
+      "loss": 2.2046,
+      "step": 6975000
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 3.272668608770183e-05,
+      "loss": 2.1846,
+      "step": 6975500
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 3.2725447501275743e-05,
+      "loss": 2.1931,
+      "step": 6976000
+    },
+    {
+      "epoch": 34.56,
+      "learning_rate": 3.2724213869195364e-05,
+      "loss": 2.1719,
+      "step": 6976500
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3.272297528276928e-05,
+      "loss": 2.181,
+      "step": 6977000
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3.27217366963432e-05,
+      "loss": 2.18,
+      "step": 6977500
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3.2720498109917115e-05,
+      "loss": 2.2069,
+      "step": 6978000
+    },
+    {
+      "epoch": 34.57,
+      "learning_rate": 3.271925952349103e-05,
+      "loss": 2.2135,
+      "step": 6978500
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3.271802093706495e-05,
+      "loss": 2.2009,
+      "step": 6979000
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3.2716782350638866e-05,
+      "loss": 2.1917,
+      "step": 6979500
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3.271554376421278e-05,
+      "loss": 2.2079,
+      "step": 6980000
+    },
+    {
+      "epoch": 34.58,
+      "learning_rate": 3.27143051777867e-05,
+      "loss": 2.182,
+      "step": 6980500
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3.2713066591360617e-05,
+      "loss": 2.2022,
+      "step": 6981000
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3.271183048210738e-05,
+      "loss": 2.199,
+      "step": 6981500
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3.2710591895681295e-05,
+      "loss": 2.1813,
+      "step": 6982000
+    },
+    {
+      "epoch": 34.59,
+      "learning_rate": 3.270935330925521e-05,
+      "loss": 2.1995,
+      "step": 6982500
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 3.270811472282913e-05,
+      "loss": 2.2181,
+      "step": 6983000
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 3.2706876136403046e-05,
+      "loss": 2.2099,
+      "step": 6983500
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 3.2705640027149815e-05,
+      "loss": 2.1953,
+      "step": 6984000
+    },
+    {
+      "epoch": 34.6,
+      "learning_rate": 3.270440144072373e-05,
+      "loss": 2.2185,
+      "step": 6984500
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3.270316285429765e-05,
+      "loss": 2.1995,
+      "step": 6985000
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3.2701924267871566e-05,
+      "loss": 2.1897,
+      "step": 6985500
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3.270068568144548e-05,
+      "loss": 2.2206,
+      "step": 6986000
+    },
+    {
+      "epoch": 34.61,
+      "learning_rate": 3.2699449572192245e-05,
+      "loss": 2.1854,
+      "step": 6986500
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3.269821098576616e-05,
+      "loss": 2.2143,
+      "step": 6987000
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3.269697239934008e-05,
+      "loss": 2.196,
+      "step": 6987500
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3.2695733812913996e-05,
+      "loss": 2.2031,
+      "step": 6988000
+    },
+    {
+      "epoch": 34.62,
+      "learning_rate": 3.269449522648791e-05,
+      "loss": 2.1955,
+      "step": 6988500
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3.269325911723468e-05,
+      "loss": 2.1923,
+      "step": 6989000
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3.26920205308086e-05,
+      "loss": 2.2108,
+      "step": 6989500
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3.2690781944382515e-05,
+      "loss": 2.2116,
+      "step": 6990000
+    },
+    {
+      "epoch": 34.63,
+      "learning_rate": 3.268954335795643e-05,
+      "loss": 2.2157,
+      "step": 6990500
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 3.268830477153035e-05,
+      "loss": 2.1915,
+      "step": 6991000
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 3.2687066185104266e-05,
+      "loss": 2.2114,
+      "step": 6991500
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 3.268582759867818e-05,
+      "loss": 2.1786,
+      "step": 6992000
+    },
+    {
+      "epoch": 34.64,
+      "learning_rate": 3.26845890122521e-05,
+      "loss": 2.1929,
+      "step": 6992500
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 3.268335042582602e-05,
+      "loss": 2.1735,
+      "step": 6993000
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 3.2682111839399934e-05,
+      "loss": 2.205,
+      "step": 6993500
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 3.268087325297385e-05,
+      "loss": 2.1987,
+      "step": 6994000
+    },
+    {
+      "epoch": 34.65,
+      "learning_rate": 3.267963466654777e-05,
+      "loss": 2.1879,
+      "step": 6994500
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.267839855729453e-05,
+      "loss": 2.2025,
+      "step": 6995000
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.2677159970868447e-05,
+      "loss": 2.1804,
+      "step": 6995500
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.2675926338788074e-05,
+      "loss": 2.2173,
+      "step": 6996000
+    },
+    {
+      "epoch": 34.66,
+      "learning_rate": 3.267468775236199e-05,
+      "loss": 2.1693,
+      "step": 6996500
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.26734491659359e-05,
+      "loss": 2.1852,
+      "step": 6997000
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.267221057950982e-05,
+      "loss": 2.2025,
+      "step": 6997500
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.2670971993083735e-05,
+      "loss": 2.1977,
+      "step": 6998000
+    },
+    {
+      "epoch": 34.67,
+      "learning_rate": 3.266973340665765e-05,
+      "loss": 2.1816,
+      "step": 6998500
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3.266849482023157e-05,
+      "loss": 2.2068,
+      "step": 6999000
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3.266725623380548e-05,
+      "loss": 2.1962,
+      "step": 6999500
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3.2666017647379396e-05,
+      "loss": 2.2068,
+      "step": 7000000
+    },
+    {
+      "epoch": 34.68,
+      "learning_rate": 3.266478153812617e-05,
+      "loss": 2.2036,
+      "step": 7000500
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 3.266354295170009e-05,
+      "loss": 2.2308,
+      "step": 7001000
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 3.2662304365274e-05,
+      "loss": 2.1966,
+      "step": 7001500
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 3.2661065778847915e-05,
+      "loss": 2.207,
+      "step": 7002000
+    },
+    {
+      "epoch": 34.69,
+      "learning_rate": 3.265982719242183e-05,
+      "loss": 2.1952,
+      "step": 7002500
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3.265859108316861e-05,
+      "loss": 2.1811,
+      "step": 7003000
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3.265735497391537e-05,
+      "loss": 2.1887,
+      "step": 7003500
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3.265611638748929e-05,
+      "loss": 2.2001,
+      "step": 7004000
+    },
+    {
+      "epoch": 34.7,
+      "learning_rate": 3.2654877801063204e-05,
+      "loss": 2.1961,
+      "step": 7004500
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3.265363921463712e-05,
+      "loss": 2.21,
+      "step": 7005000
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3.265240062821104e-05,
+      "loss": 2.1943,
+      "step": 7005500
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3.2651162041784955e-05,
+      "loss": 2.2111,
+      "step": 7006000
+    },
+    {
+      "epoch": 34.71,
+      "learning_rate": 3.264992345535887e-05,
+      "loss": 2.1912,
+      "step": 7006500
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3.264868486893279e-05,
+      "loss": 2.1894,
+      "step": 7007000
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3.26474462825067e-05,
+      "loss": 2.1854,
+      "step": 7007500
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3.2646207696080616e-05,
+      "loss": 2.2093,
+      "step": 7008000
+    },
+    {
+      "epoch": 34.72,
+      "learning_rate": 3.264497158682739e-05,
+      "loss": 2.2025,
+      "step": 7008500
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3.264373300040131e-05,
+      "loss": 2.1801,
+      "step": 7009000
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3.2642494413975225e-05,
+      "loss": 2.1916,
+      "step": 7009500
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3.264125582754914e-05,
+      "loss": 2.2205,
+      "step": 7010000
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3.264001724112305e-05,
+      "loss": 2.1982,
+      "step": 7010500
+    },
+    {
+      "epoch": 34.73,
+      "learning_rate": 3.263877865469697e-05,
+      "loss": 2.1991,
+      "step": 7011000
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3.2637540068270886e-05,
+      "loss": 2.2063,
+      "step": 7011500
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3.26363014818448e-05,
+      "loss": 2.1884,
+      "step": 7012000
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3.263506289541872e-05,
+      "loss": 2.1918,
+      "step": 7012500
+    },
+    {
+      "epoch": 34.74,
+      "learning_rate": 3.263382430899263e-05,
+      "loss": 2.2083,
+      "step": 7013000
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3.2632588199739406e-05,
+      "loss": 2.1801,
+      "step": 7013500
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3.2631349613313316e-05,
+      "loss": 2.1982,
+      "step": 7014000
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3.263011102688723e-05,
+      "loss": 2.1761,
+      "step": 7014500
+    },
+    {
+      "epoch": 34.75,
+      "learning_rate": 3.262887491763401e-05,
+      "loss": 2.2228,
+      "step": 7015000
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3.2627636331207925e-05,
+      "loss": 2.1739,
+      "step": 7015500
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3.262639774478184e-05,
+      "loss": 2.2176,
+      "step": 7016000
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3.2625161635528604e-05,
+      "loss": 2.1896,
+      "step": 7016500
+    },
+    {
+      "epoch": 34.76,
+      "learning_rate": 3.262392304910252e-05,
+      "loss": 2.1842,
+      "step": 7017000
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 3.262268446267644e-05,
+      "loss": 2.2216,
+      "step": 7017500
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 3.2621445876250355e-05,
+      "loss": 2.1878,
+      "step": 7018000
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 3.262020728982427e-05,
+      "loss": 2.2043,
+      "step": 7018500
+    },
+    {
+      "epoch": 34.77,
+      "learning_rate": 3.261896870339819e-05,
+      "loss": 2.192,
+      "step": 7019000
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3.2617730116972106e-05,
+      "loss": 2.2035,
+      "step": 7019500
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3.2616491530546016e-05,
+      "loss": 2.2266,
+      "step": 7020000
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3.261525294411993e-05,
+      "loss": 2.1957,
+      "step": 7020500
+    },
+    {
+      "epoch": 34.78,
+      "learning_rate": 3.261401435769385e-05,
+      "loss": 2.2057,
+      "step": 7021000
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3.2612778248440625e-05,
+      "loss": 2.1826,
+      "step": 7021500
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3.261153966201454e-05,
+      "loss": 2.1922,
+      "step": 7022000
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3.261030107558846e-05,
+      "loss": 2.1976,
+      "step": 7022500
+    },
+    {
+      "epoch": 34.79,
+      "learning_rate": 3.2609062489162376e-05,
+      "loss": 2.2111,
+      "step": 7023000
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3.2607823902736286e-05,
+      "loss": 2.1923,
+      "step": 7023500
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3.2606587793483055e-05,
+      "loss": 2.2014,
+      "step": 7024000
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3.260534920705697e-05,
+      "loss": 2.1941,
+      "step": 7024500
+    },
+    {
+      "epoch": 34.8,
+      "learning_rate": 3.260411062063089e-05,
+      "loss": 2.1953,
+      "step": 7025000
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3.2602872034204806e-05,
+      "loss": 2.213,
+      "step": 7025500
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3.2601635924951575e-05,
+      "loss": 2.1574,
+      "step": 7026000
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3.260039733852549e-05,
+      "loss": 2.2092,
+      "step": 7026500
+    },
+    {
+      "epoch": 34.81,
+      "learning_rate": 3.259915875209941e-05,
+      "loss": 2.2085,
+      "step": 7027000
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 3.259792264284617e-05,
+      "loss": 2.1949,
+      "step": 7027500
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 3.259668405642009e-05,
+      "loss": 2.2119,
+      "step": 7028000
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 3.2595445469994004e-05,
+      "loss": 2.1811,
+      "step": 7028500
+    },
+    {
+      "epoch": 34.82,
+      "learning_rate": 3.259420688356792e-05,
+      "loss": 2.1964,
+      "step": 7029000
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.259297077431469e-05,
+      "loss": 2.2017,
+      "step": 7029500
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.259173218788861e-05,
+      "loss": 2.1988,
+      "step": 7030000
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.2590493601462524e-05,
+      "loss": 2.2063,
+      "step": 7030500
+    },
+    {
+      "epoch": 34.83,
+      "learning_rate": 3.258925501503644e-05,
+      "loss": 2.1986,
+      "step": 7031000
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3.258801642861036e-05,
+      "loss": 2.1811,
+      "step": 7031500
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3.2586777842184275e-05,
+      "loss": 2.1918,
+      "step": 7032000
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3.258553925575819e-05,
+      "loss": 2.2125,
+      "step": 7032500
+    },
+    {
+      "epoch": 34.84,
+      "learning_rate": 3.258430066933211e-05,
+      "loss": 2.209,
+      "step": 7033000
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3.2583062082906026e-05,
+      "loss": 2.2314,
+      "step": 7033500
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3.258182349647994e-05,
+      "loss": 2.1891,
+      "step": 7034000
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3.258058491005386e-05,
+      "loss": 2.2034,
+      "step": 7034500
+    },
+    {
+      "epoch": 34.85,
+      "learning_rate": 3.2579346323627776e-05,
+      "loss": 2.1971,
+      "step": 7035000
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 3.257810773720169e-05,
+      "loss": 2.1942,
+      "step": 7035500
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 3.25768691507756e-05,
+      "loss": 2.1755,
+      "step": 7036000
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 3.257563304152237e-05,
+      "loss": 2.1964,
+      "step": 7036500
+    },
+    {
+      "epoch": 34.86,
+      "learning_rate": 3.257439940944199e-05,
+      "loss": 2.1967,
+      "step": 7037000
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3.257316082301591e-05,
+      "loss": 2.2141,
+      "step": 7037500
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3.257192223658983e-05,
+      "loss": 2.2195,
+      "step": 7038000
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3.2570683650163744e-05,
+      "loss": 2.2072,
+      "step": 7038500
+    },
+    {
+      "epoch": 34.87,
+      "learning_rate": 3.256944506373766e-05,
+      "loss": 2.2148,
+      "step": 7039000
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3.256820647731158e-05,
+      "loss": 2.213,
+      "step": 7039500
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3.2566967890885494e-05,
+      "loss": 2.1913,
+      "step": 7040000
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3.2565731781632256e-05,
+      "loss": 2.2049,
+      "step": 7040500
+    },
+    {
+      "epoch": 34.88,
+      "learning_rate": 3.2564493195206173e-05,
+      "loss": 2.2205,
+      "step": 7041000
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3.256325460878009e-05,
+      "loss": 2.1822,
+      "step": 7041500
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3.256201602235401e-05,
+      "loss": 2.1775,
+      "step": 7042000
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3.2560777435927924e-05,
+      "loss": 2.2201,
+      "step": 7042500
+    },
+    {
+      "epoch": 34.89,
+      "learning_rate": 3.255953884950184e-05,
+      "loss": 2.1917,
+      "step": 7043000
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 3.255830026307576e-05,
+      "loss": 2.2215,
+      "step": 7043500
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 3.2557061676649675e-05,
+      "loss": 2.212,
+      "step": 7044000
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 3.2555825567396444e-05,
+      "loss": 2.1995,
+      "step": 7044500
+    },
+    {
+      "epoch": 34.9,
+      "learning_rate": 3.255458698097036e-05,
+      "loss": 2.2085,
+      "step": 7045000
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3.255334839454428e-05,
+      "loss": 2.1986,
+      "step": 7045500
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3.2552109808118195e-05,
+      "loss": 2.1863,
+      "step": 7046000
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3.255087122169211e-05,
+      "loss": 2.2211,
+      "step": 7046500
+    },
+    {
+      "epoch": 34.91,
+      "learning_rate": 3.2549635112438874e-05,
+      "loss": 2.1993,
+      "step": 7047000
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3.254839652601279e-05,
+      "loss": 2.1953,
+      "step": 7047500
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3.254715793958671e-05,
+      "loss": 2.2076,
+      "step": 7048000
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3.2545919353160624e-05,
+      "loss": 2.1892,
+      "step": 7048500
+    },
+    {
+      "epoch": 34.92,
+      "learning_rate": 3.254468324390739e-05,
+      "loss": 2.2138,
+      "step": 7049000
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3.254344465748131e-05,
+      "loss": 2.1946,
+      "step": 7049500
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3.254220607105523e-05,
+      "loss": 2.204,
+      "step": 7050000
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3.2540967484629144e-05,
+      "loss": 2.212,
+      "step": 7050500
+    },
+    {
+      "epoch": 34.93,
+      "learning_rate": 3.253972889820306e-05,
+      "loss": 2.1801,
+      "step": 7051000
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 3.253849031177698e-05,
+      "loss": 2.1985,
+      "step": 7051500
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 3.2537251725350895e-05,
+      "loss": 2.1711,
+      "step": 7052000
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 3.253601313892481e-05,
+      "loss": 2.2165,
+      "step": 7052500
+    },
+    {
+      "epoch": 34.94,
+      "learning_rate": 3.2534777029671574e-05,
+      "loss": 2.2252,
+      "step": 7053000
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3.253353844324549e-05,
+      "loss": 2.1821,
+      "step": 7053500
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3.2532302333992266e-05,
+      "loss": 2.1778,
+      "step": 7054000
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3.253106374756618e-05,
+      "loss": 2.2102,
+      "step": 7054500
+    },
+    {
+      "epoch": 34.95,
+      "learning_rate": 3.252982516114009e-05,
+      "loss": 2.2164,
+      "step": 7055000
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3.252858657471401e-05,
+      "loss": 2.2235,
+      "step": 7055500
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3.252734798828793e-05,
+      "loss": 2.2477,
+      "step": 7056000
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3.2526111879034696e-05,
+      "loss": 2.1983,
+      "step": 7056500
+    },
+    {
+      "epoch": 34.96,
+      "learning_rate": 3.252487329260861e-05,
+      "loss": 2.2055,
+      "step": 7057000
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 3.252363718335538e-05,
+      "loss": 2.2165,
+      "step": 7057500
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 3.25223985969293e-05,
+      "loss": 2.2136,
+      "step": 7058000
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 3.2521160010503216e-05,
+      "loss": 2.2037,
+      "step": 7058500
+    },
+    {
+      "epoch": 34.97,
+      "learning_rate": 3.251992142407713e-05,
+      "loss": 2.1987,
+      "step": 7059000
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3.251868283765105e-05,
+      "loss": 2.2126,
+      "step": 7059500
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3.2517444251224966e-05,
+      "loss": 2.2202,
+      "step": 7060000
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3.251620566479888e-05,
+      "loss": 2.2023,
+      "step": 7060500
+    },
+    {
+      "epoch": 34.98,
+      "learning_rate": 3.25149670783728e-05,
+      "loss": 2.2168,
+      "step": 7061000
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 3.251372849194671e-05,
+      "loss": 2.1998,
+      "step": 7061500
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 3.251249238269348e-05,
+      "loss": 2.1941,
+      "step": 7062000
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 3.2511253796267396e-05,
+      "loss": 2.1872,
+      "step": 7062500
+    },
+    {
+      "epoch": 34.99,
+      "learning_rate": 3.2510017687014165e-05,
+      "loss": 2.2097,
+      "step": 7063000
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.250877910058808e-05,
+      "loss": 2.2149,
+      "step": 7063500
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.2507540514162e-05,
+      "loss": 2.2237,
+      "step": 7064000
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.2506301927735916e-05,
+      "loss": 2.201,
+      "step": 7064500
+    },
+    {
+      "epoch": 35.0,
+      "eval_accuracy": 0.662072274008178,
+      "eval_accuracy_mlm": 0.6186920132733609,
+      "eval_accuracy_nsp": 0.8666060033181806,
+      "eval_loss": 2.2997031211853027,
+      "eval_runtime": 145.6764,
+      "eval_samples_per_second": 1750.173,
+      "eval_steps_per_second": 72.929,
+      "step": 7064505
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.250506334130983e-05,
+      "loss": 2.1687,
+      "step": 7065000
+    },
+    {
+      "epoch": 35.0,
+      "learning_rate": 3.25038272320566e-05,
+      "loss": 2.1827,
+      "step": 7065500
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3.250258864563052e-05,
+      "loss": 2.1597,
+      "step": 7066000
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3.2501350059204435e-05,
+      "loss": 2.1611,
+      "step": 7066500
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3.250011147277835e-05,
+      "loss": 2.1661,
+      "step": 7067000
+    },
+    {
+      "epoch": 35.01,
+      "learning_rate": 3.249887288635227e-05,
+      "loss": 2.158,
+      "step": 7067500
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 3.249763677709903e-05,
+      "loss": 2.1792,
+      "step": 7068000
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 3.249639819067295e-05,
+      "loss": 2.1561,
+      "step": 7068500
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 3.2495159604246865e-05,
+      "loss": 2.1664,
+      "step": 7069000
+    },
+    {
+      "epoch": 35.02,
+      "learning_rate": 3.2493923494993634e-05,
+      "loss": 2.1907,
+      "step": 7069500
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 3.249268490856755e-05,
+      "loss": 2.1791,
+      "step": 7070000
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 3.249144632214147e-05,
+      "loss": 2.1948,
+      "step": 7070500
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 3.2490207735715385e-05,
+      "loss": 2.1503,
+      "step": 7071000
+    },
+    {
+      "epoch": 35.03,
+      "learning_rate": 3.24889691492893e-05,
+      "loss": 2.1801,
+      "step": 7071500
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 3.248773056286322e-05,
+      "loss": 2.1924,
+      "step": 7072000
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 3.2486491976437135e-05,
+      "loss": 2.1651,
+      "step": 7072500
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 3.248525339001105e-05,
+      "loss": 2.1392,
+      "step": 7073000
+    },
+    {
+      "epoch": 35.04,
+      "learning_rate": 3.248401480358497e-05,
+      "loss": 2.1702,
+      "step": 7073500
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 3.2482776217158886e-05,
+      "loss": 2.1746,
+      "step": 7074000
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 3.24815376307328e-05,
+      "loss": 2.1503,
+      "step": 7074500
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 3.248029904430672e-05,
+      "loss": 2.1645,
+      "step": 7075000
+    },
+    {
+      "epoch": 35.05,
+      "learning_rate": 3.247906045788063e-05,
+      "loss": 2.1635,
+      "step": 7075500
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 3.247782187145455e-05,
+      "loss": 2.1599,
+      "step": 7076000
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 3.2476583285028464e-05,
+      "loss": 2.164,
+      "step": 7076500
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 3.247534469860238e-05,
+      "loss": 2.1623,
+      "step": 7077000
+    },
+    {
+      "epoch": 35.06,
+      "learning_rate": 3.24741061121763e-05,
+      "loss": 2.1668,
+      "step": 7077500
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 3.247287000292307e-05,
+      "loss": 2.1809,
+      "step": 7078000
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 3.2471633893669835e-05,
+      "loss": 2.1799,
+      "step": 7078500
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 3.247039530724375e-05,
+      "loss": 2.1867,
+      "step": 7079000
+    },
+    {
+      "epoch": 35.07,
+      "learning_rate": 3.246915672081767e-05,
+      "loss": 2.1799,
+      "step": 7079500
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 3.2467918134391586e-05,
+      "loss": 2.1792,
+      "step": 7080000
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 3.24666795479655e-05,
+      "loss": 2.1811,
+      "step": 7080500
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 3.246544096153942e-05,
+      "loss": 2.1582,
+      "step": 7081000
+    },
+    {
+      "epoch": 35.08,
+      "learning_rate": 3.246420237511334e-05,
+      "loss": 2.1756,
+      "step": 7081500
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 3.2462963788687254e-05,
+      "loss": 2.2023,
+      "step": 7082000
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 3.2461725202261164e-05,
+      "loss": 2.1908,
+      "step": 7082500
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 3.246048909300793e-05,
+      "loss": 2.1654,
+      "step": 7083000
+    },
+    {
+      "epoch": 35.09,
+      "learning_rate": 3.245925050658185e-05,
+      "loss": 2.1965,
+      "step": 7083500
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 3.245801192015577e-05,
+      "loss": 2.1755,
+      "step": 7084000
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 3.2456773333729684e-05,
+      "loss": 2.1879,
+      "step": 7084500
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 3.24555347473036e-05,
+      "loss": 2.1588,
+      "step": 7085000
+    },
+    {
+      "epoch": 35.1,
+      "learning_rate": 3.245429863805037e-05,
+      "loss": 2.1715,
+      "step": 7085500
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 3.2453060051624286e-05,
+      "loss": 2.1767,
+      "step": 7086000
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 3.24518214651982e-05,
+      "loss": 2.1845,
+      "step": 7086500
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 3.245058287877212e-05,
+      "loss": 2.1522,
+      "step": 7087000
+    },
+    {
+      "epoch": 35.11,
+      "learning_rate": 3.244934429234604e-05,
+      "loss": 2.1601,
+      "step": 7087500
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 3.244811066026565e-05,
+      "loss": 2.1818,
+      "step": 7088000
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 3.244687455101243e-05,
+      "loss": 2.198,
+      "step": 7088500
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 3.2445635964586344e-05,
+      "loss": 2.1568,
+      "step": 7089000
+    },
+    {
+      "epoch": 35.12,
+      "learning_rate": 3.244439737816026e-05,
+      "loss": 2.1736,
+      "step": 7089500
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 3.244315879173417e-05,
+      "loss": 2.1725,
+      "step": 7090000
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 3.244192020530809e-05,
+      "loss": 2.1716,
+      "step": 7090500
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 3.2440681618882005e-05,
+      "loss": 2.1593,
+      "step": 7091000
+    },
+    {
+      "epoch": 35.13,
+      "learning_rate": 3.243944303245592e-05,
+      "loss": 2.1417,
+      "step": 7091500
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 3.243820444602984e-05,
+      "loss": 2.1594,
+      "step": 7092000
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 3.243696585960375e-05,
+      "loss": 2.1771,
+      "step": 7092500
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 3.2435727273177665e-05,
+      "loss": 2.1621,
+      "step": 7093000
+    },
+    {
+      "epoch": 35.14,
+      "learning_rate": 3.243448868675158e-05,
+      "loss": 2.1602,
+      "step": 7093500
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 3.24332501003255e-05,
+      "loss": 2.1644,
+      "step": 7094000
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 3.2432011513899416e-05,
+      "loss": 2.1716,
+      "step": 7094500
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 3.243077292747333e-05,
+      "loss": 2.1893,
+      "step": 7095000
+    },
+    {
+      "epoch": 35.15,
+      "learning_rate": 3.242953434104725e-05,
+      "loss": 2.1837,
+      "step": 7095500
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 3.242829575462117e-05,
+      "loss": 2.1803,
+      "step": 7096000
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 3.2427059645367936e-05,
+      "loss": 2.1727,
+      "step": 7096500
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 3.242582105894185e-05,
+      "loss": 2.1676,
+      "step": 7097000
+    },
+    {
+      "epoch": 35.16,
+      "learning_rate": 3.242458247251577e-05,
+      "loss": 2.1982,
+      "step": 7097500
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 3.242334388608969e-05,
+      "loss": 2.203,
+      "step": 7098000
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 3.2422105299663604e-05,
+      "loss": 2.1756,
+      "step": 7098500
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 3.2420869190410366e-05,
+      "loss": 2.1352,
+      "step": 7099000
+    },
+    {
+      "epoch": 35.17,
+      "learning_rate": 3.241963060398428e-05,
+      "loss": 2.1828,
+      "step": 7099500
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 3.24183920175582e-05,
+      "loss": 2.1801,
+      "step": 7100000
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 3.2417153431132116e-05,
+      "loss": 2.1608,
+      "step": 7100500
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 3.241591484470603e-05,
+      "loss": 2.1727,
+      "step": 7101000
+    },
+    {
+      "epoch": 35.18,
+      "learning_rate": 3.241467625827995e-05,
+      "loss": 2.1829,
+      "step": 7101500
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 3.241343767185387e-05,
+      "loss": 2.1832,
+      "step": 7102000
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 3.2412199085427784e-05,
+      "loss": 2.1647,
+      "step": 7102500
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 3.241096297617455e-05,
+      "loss": 2.153,
+      "step": 7103000
+    },
+    {
+      "epoch": 35.19,
+      "learning_rate": 3.240972438974847e-05,
+      "loss": 2.164,
+      "step": 7103500
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 3.240848580332239e-05,
+      "loss": 2.1728,
+      "step": 7104000
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 3.2407247216896304e-05,
+      "loss": 2.1788,
+      "step": 7104500
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 3.240600863047022e-05,
+      "loss": 2.1901,
+      "step": 7105000
+    },
+    {
+      "epoch": 35.2,
+      "learning_rate": 3.240477004404414e-05,
+      "loss": 2.1619,
+      "step": 7105500
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 3.2403531457618054e-05,
+      "loss": 2.1746,
+      "step": 7106000
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 3.240229287119197e-05,
+      "loss": 2.19,
+      "step": 7106500
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 3.240105428476589e-05,
+      "loss": 2.1823,
+      "step": 7107000
+    },
+    {
+      "epoch": 35.21,
+      "learning_rate": 3.239981817551265e-05,
+      "loss": 2.1837,
+      "step": 7107500
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 3.239857958908657e-05,
+      "loss": 2.1808,
+      "step": 7108000
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 3.2397341002660484e-05,
+      "loss": 2.1844,
+      "step": 7108500
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 3.23961024162344e-05,
+      "loss": 2.1701,
+      "step": 7109000
+    },
+    {
+      "epoch": 35.22,
+      "learning_rate": 3.239486382980832e-05,
+      "loss": 2.1671,
+      "step": 7109500
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 3.2393625243382235e-05,
+      "loss": 2.2084,
+      "step": 7110000
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 3.2392389134129004e-05,
+      "loss": 2.1831,
+      "step": 7110500
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 3.239115054770292e-05,
+      "loss": 2.1894,
+      "step": 7111000
+    },
+    {
+      "epoch": 35.23,
+      "learning_rate": 3.238991196127684e-05,
+      "loss": 2.1976,
+      "step": 7111500
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 3.2388673374850755e-05,
+      "loss": 2.1777,
+      "step": 7112000
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 3.238743726559752e-05,
+      "loss": 2.1676,
+      "step": 7112500
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 3.2386198679171434e-05,
+      "loss": 2.1834,
+      "step": 7113000
+    },
+    {
+      "epoch": 35.24,
+      "learning_rate": 3.238496504709106e-05,
+      "loss": 2.1865,
+      "step": 7113500
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 3.238372646066498e-05,
+      "loss": 2.1847,
+      "step": 7114000
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 3.238249035141174e-05,
+      "loss": 2.1596,
+      "step": 7114500
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 3.238125176498566e-05,
+      "loss": 2.1785,
+      "step": 7115000
+    },
+    {
+      "epoch": 35.25,
+      "learning_rate": 3.2380013178559574e-05,
+      "loss": 2.1876,
+      "step": 7115500
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 3.237877459213349e-05,
+      "loss": 2.1801,
+      "step": 7116000
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 3.237753600570741e-05,
+      "loss": 2.1576,
+      "step": 7116500
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 3.2376297419281325e-05,
+      "loss": 2.1708,
+      "step": 7117000
+    },
+    {
+      "epoch": 35.26,
+      "learning_rate": 3.237505883285524e-05,
+      "loss": 2.188,
+      "step": 7117500
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 3.237382024642916e-05,
+      "loss": 2.1749,
+      "step": 7118000
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 3.237258166000307e-05,
+      "loss": 2.1658,
+      "step": 7118500
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 3.2371345550749844e-05,
+      "loss": 2.1957,
+      "step": 7119000
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 3.237010696432376e-05,
+      "loss": 2.1925,
+      "step": 7119500
+    },
+    {
+      "epoch": 35.27,
+      "learning_rate": 3.236886837789768e-05,
+      "loss": 2.1954,
+      "step": 7120000
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 3.2367629791471595e-05,
+      "loss": 2.171,
+      "step": 7120500
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 3.236639120504551e-05,
+      "loss": 2.1574,
+      "step": 7121000
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 3.236515261861942e-05,
+      "loss": 2.1797,
+      "step": 7121500
+    },
+    {
+      "epoch": 35.28,
+      "learning_rate": 3.236391403219334e-05,
+      "loss": 2.1708,
+      "step": 7122000
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 3.2362675445767256e-05,
+      "loss": 2.1776,
+      "step": 7122500
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 3.236143685934117e-05,
+      "loss": 2.1824,
+      "step": 7123000
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 3.236019827291509e-05,
+      "loss": 2.1585,
+      "step": 7123500
+    },
+    {
+      "epoch": 35.29,
+      "learning_rate": 3.235895968648901e-05,
+      "loss": 2.1665,
+      "step": 7124000
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 3.2357721100062924e-05,
+      "loss": 2.162,
+      "step": 7124500
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 3.235648251363684e-05,
+      "loss": 2.1863,
+      "step": 7125000
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 3.235524392721076e-05,
+      "loss": 2.1742,
+      "step": 7125500
+    },
+    {
+      "epoch": 35.3,
+      "learning_rate": 3.2354005340784674e-05,
+      "loss": 2.1806,
+      "step": 7126000
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 3.2352766754358585e-05,
+      "loss": 2.1616,
+      "step": 7126500
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 3.23515281679325e-05,
+      "loss": 2.1634,
+      "step": 7127000
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 3.235029453585213e-05,
+      "loss": 2.1948,
+      "step": 7127500
+    },
+    {
+      "epoch": 35.31,
+      "learning_rate": 3.234905594942604e-05,
+      "loss": 2.1992,
+      "step": 7128000
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 3.2347817362999956e-05,
+      "loss": 2.1518,
+      "step": 7128500
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 3.234657877657387e-05,
+      "loss": 2.1799,
+      "step": 7129000
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 3.234534266732064e-05,
+      "loss": 2.1858,
+      "step": 7129500
+    },
+    {
+      "epoch": 35.32,
+      "learning_rate": 3.234410408089456e-05,
+      "loss": 2.1683,
+      "step": 7130000
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 3.2342865494468476e-05,
+      "loss": 2.1536,
+      "step": 7130500
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 3.2341626908042386e-05,
+      "loss": 2.1983,
+      "step": 7131000
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 3.23403883216163e-05,
+      "loss": 2.1816,
+      "step": 7131500
+    },
+    {
+      "epoch": 35.33,
+      "learning_rate": 3.233914973519022e-05,
+      "loss": 2.1769,
+      "step": 7132000
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3.2337911148764137e-05,
+      "loss": 2.1931,
+      "step": 7132500
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3.2336672562338054e-05,
+      "loss": 2.1675,
+      "step": 7133000
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3.233543397591197e-05,
+      "loss": 2.1926,
+      "step": 7133500
+    },
+    {
+      "epoch": 35.34,
+      "learning_rate": 3.233419538948589e-05,
+      "loss": 2.1825,
+      "step": 7134000
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 3.2332959280232656e-05,
+      "loss": 2.196,
+      "step": 7134500
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 3.2331723170979425e-05,
+      "loss": 2.173,
+      "step": 7135000
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 3.2330487061726194e-05,
+      "loss": 2.1661,
+      "step": 7135500
+    },
+    {
+      "epoch": 35.35,
+      "learning_rate": 3.232924847530011e-05,
+      "loss": 2.1848,
+      "step": 7136000
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 3.232800988887403e-05,
+      "loss": 2.1861,
+      "step": 7136500
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 3.2326771302447945e-05,
+      "loss": 2.1877,
+      "step": 7137000
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 3.232553271602186e-05,
+      "loss": 2.2125,
+      "step": 7137500
+    },
+    {
+      "epoch": 35.36,
+      "learning_rate": 3.232429412959578e-05,
+      "loss": 2.1783,
+      "step": 7138000
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 3.232305802034255e-05,
+      "loss": 2.1706,
+      "step": 7138500
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 3.2321819433916464e-05,
+      "loss": 2.1612,
+      "step": 7139000
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 3.232058084749038e-05,
+      "loss": 2.1701,
+      "step": 7139500
+    },
+    {
+      "epoch": 35.37,
+      "learning_rate": 3.23193422610643e-05,
+      "loss": 2.1693,
+      "step": 7140000
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 3.231810615181106e-05,
+      "loss": 2.1877,
+      "step": 7140500
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 3.231686756538498e-05,
+      "loss": 2.1918,
+      "step": 7141000
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 3.2315628978958894e-05,
+      "loss": 2.1736,
+      "step": 7141500
+    },
+    {
+      "epoch": 35.38,
+      "learning_rate": 3.231439039253281e-05,
+      "loss": 2.1831,
+      "step": 7142000
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 3.231315180610673e-05,
+      "loss": 2.1866,
+      "step": 7142500
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 3.2311913219680645e-05,
+      "loss": 2.2025,
+      "step": 7143000
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 3.231067463325456e-05,
+      "loss": 2.1617,
+      "step": 7143500
+    },
+    {
+      "epoch": 35.39,
+      "learning_rate": 3.230943604682848e-05,
+      "loss": 2.2045,
+      "step": 7144000
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 3.2308197460402395e-05,
+      "loss": 2.1891,
+      "step": 7144500
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 3.230695887397631e-05,
+      "loss": 2.1938,
+      "step": 7145000
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 3.230572028755023e-05,
+      "loss": 2.1791,
+      "step": 7145500
+    },
+    {
+      "epoch": 35.4,
+      "learning_rate": 3.2304484178297e-05,
+      "loss": 2.1913,
+      "step": 7146000
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 3.2303245591870915e-05,
+      "loss": 2.1784,
+      "step": 7146500
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 3.230200700544483e-05,
+      "loss": 2.1682,
+      "step": 7147000
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 3.230076841901874e-05,
+      "loss": 2.1989,
+      "step": 7147500
+    },
+    {
+      "epoch": 35.41,
+      "learning_rate": 3.229952983259266e-05,
+      "loss": 2.173,
+      "step": 7148000
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 3.229829372333943e-05,
+      "loss": 2.1915,
+      "step": 7148500
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 3.2297055136913345e-05,
+      "loss": 2.1816,
+      "step": 7149000
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 3.229581655048726e-05,
+      "loss": 2.1887,
+      "step": 7149500
+    },
+    {
+      "epoch": 35.42,
+      "learning_rate": 3.229458044123403e-05,
+      "loss": 2.1812,
+      "step": 7150000
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 3.229334185480795e-05,
+      "loss": 2.1866,
+      "step": 7150500
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 3.2292103268381864e-05,
+      "loss": 2.1714,
+      "step": 7151000
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 3.229086468195578e-05,
+      "loss": 2.1648,
+      "step": 7151500
+    },
+    {
+      "epoch": 35.43,
+      "learning_rate": 3.22896260955297e-05,
+      "loss": 2.1689,
+      "step": 7152000
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 3.2288387509103615e-05,
+      "loss": 2.1751,
+      "step": 7152500
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 3.228715139985038e-05,
+      "loss": 2.1721,
+      "step": 7153000
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 3.2285912813424294e-05,
+      "loss": 2.1756,
+      "step": 7153500
+    },
+    {
+      "epoch": 35.44,
+      "learning_rate": 3.228467422699821e-05,
+      "loss": 2.1678,
+      "step": 7154000
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 3.228343564057213e-05,
+      "loss": 2.2027,
+      "step": 7154500
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 3.22821995313189e-05,
+      "loss": 2.2017,
+      "step": 7155000
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 3.2280960944892814e-05,
+      "loss": 2.1792,
+      "step": 7155500
+    },
+    {
+      "epoch": 35.45,
+      "learning_rate": 3.227972483563958e-05,
+      "loss": 2.203,
+      "step": 7156000
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 3.22784862492135e-05,
+      "loss": 2.174,
+      "step": 7156500
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 3.2277247662787416e-05,
+      "loss": 2.1771,
+      "step": 7157000
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 3.2276009076361327e-05,
+      "loss": 2.2021,
+      "step": 7157500
+    },
+    {
+      "epoch": 35.46,
+      "learning_rate": 3.2274770489935244e-05,
+      "loss": 2.1874,
+      "step": 7158000
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 3.227353190350916e-05,
+      "loss": 2.1839,
+      "step": 7158500
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 3.227229331708308e-05,
+      "loss": 2.1788,
+      "step": 7159000
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 3.2271054730656994e-05,
+      "loss": 2.1972,
+      "step": 7159500
+    },
+    {
+      "epoch": 35.47,
+      "learning_rate": 3.226981614423091e-05,
+      "loss": 2.1936,
+      "step": 7160000
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 3.226857755780483e-05,
+      "loss": 2.1646,
+      "step": 7160500
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 3.2267338971378745e-05,
+      "loss": 2.1965,
+      "step": 7161000
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 3.226610038495266e-05,
+      "loss": 2.1918,
+      "step": 7161500
+    },
+    {
+      "epoch": 35.48,
+      "learning_rate": 3.226486179852658e-05,
+      "loss": 2.2044,
+      "step": 7162000
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 3.2263623212100496e-05,
+      "loss": 2.1982,
+      "step": 7162500
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 3.2262387102847265e-05,
+      "loss": 2.1941,
+      "step": 7163000
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 3.226114851642118e-05,
+      "loss": 2.1976,
+      "step": 7163500
+    },
+    {
+      "epoch": 35.49,
+      "learning_rate": 3.22599099299951e-05,
+      "loss": 2.1736,
+      "step": 7164000
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 3.2258671343569015e-05,
+      "loss": 2.1829,
+      "step": 7164500
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 3.225743275714293e-05,
+      "loss": 2.1834,
+      "step": 7165000
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 3.225619417071685e-05,
+      "loss": 2.1704,
+      "step": 7165500
+    },
+    {
+      "epoch": 35.5,
+      "learning_rate": 3.2254955584290766e-05,
+      "loss": 2.1748,
+      "step": 7166000
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 3.225371699786468e-05,
+      "loss": 2.1674,
+      "step": 7166500
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 3.22524784114386e-05,
+      "loss": 2.2049,
+      "step": 7167000
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 3.225123982501251e-05,
+      "loss": 2.1529,
+      "step": 7167500
+    },
+    {
+      "epoch": 35.51,
+      "learning_rate": 3.225000619293213e-05,
+      "loss": 2.1924,
+      "step": 7168000
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 3.224876760650605e-05,
+      "loss": 2.1786,
+      "step": 7168500
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 3.2247529020079965e-05,
+      "loss": 2.1862,
+      "step": 7169000
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 3.224629043365388e-05,
+      "loss": 2.1665,
+      "step": 7169500
+    },
+    {
+      "epoch": 35.52,
+      "learning_rate": 3.22450518472278e-05,
+      "loss": 2.1811,
+      "step": 7170000
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 3.224381573797457e-05,
+      "loss": 2.1747,
+      "step": 7170500
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 3.224257715154848e-05,
+      "loss": 2.2019,
+      "step": 7171000
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 3.2241338565122395e-05,
+      "loss": 2.2123,
+      "step": 7171500
+    },
+    {
+      "epoch": 35.53,
+      "learning_rate": 3.224009997869631e-05,
+      "loss": 2.1879,
+      "step": 7172000
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 3.223886386944308e-05,
+      "loss": 2.2036,
+      "step": 7172500
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 3.2237625283017e-05,
+      "loss": 2.1967,
+      "step": 7173000
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 3.2236386696590914e-05,
+      "loss": 2.191,
+      "step": 7173500
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 3.223514811016483e-05,
+      "loss": 2.1963,
+      "step": 7174000
+    },
+    {
+      "epoch": 35.54,
+      "learning_rate": 3.223390952373875e-05,
+      "loss": 2.1806,
+      "step": 7174500
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 3.2232670937312665e-05,
+      "loss": 2.202,
+      "step": 7175000
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 3.2231434828059434e-05,
+      "loss": 2.1755,
+      "step": 7175500
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 3.223019624163335e-05,
+      "loss": 2.1722,
+      "step": 7176000
+    },
+    {
+      "epoch": 35.55,
+      "learning_rate": 3.222895765520727e-05,
+      "loss": 2.1985,
+      "step": 7176500
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 3.2227719068781185e-05,
+      "loss": 2.1716,
+      "step": 7177000
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 3.22264804823551e-05,
+      "loss": 2.1865,
+      "step": 7177500
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 3.222524189592901e-05,
+      "loss": 2.1747,
+      "step": 7178000
+    },
+    {
+      "epoch": 35.56,
+      "learning_rate": 3.222400578667578e-05,
+      "loss": 2.1841,
+      "step": 7178500
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 3.22227672002497e-05,
+      "loss": 2.2028,
+      "step": 7179000
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 3.2221528613823614e-05,
+      "loss": 2.1779,
+      "step": 7179500
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 3.222029002739753e-05,
+      "loss": 2.2021,
+      "step": 7180000
+    },
+    {
+      "epoch": 35.57,
+      "learning_rate": 3.221905391814431e-05,
+      "loss": 2.1877,
+      "step": 7180500
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 3.2217815331718224e-05,
+      "loss": 2.192,
+      "step": 7181000
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 3.2216576745292134e-05,
+      "loss": 2.1805,
+      "step": 7181500
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 3.221533815886605e-05,
+      "loss": 2.2088,
+      "step": 7182000
+    },
+    {
+      "epoch": 35.58,
+      "learning_rate": 3.221409957243997e-05,
+      "loss": 2.1966,
+      "step": 7182500
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 3.2212860986013885e-05,
+      "loss": 2.1925,
+      "step": 7183000
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 3.22116223995878e-05,
+      "loss": 2.1957,
+      "step": 7183500
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 3.221038629033457e-05,
+      "loss": 2.1982,
+      "step": 7184000
+    },
+    {
+      "epoch": 35.59,
+      "learning_rate": 3.220914770390848e-05,
+      "loss": 2.2168,
+      "step": 7184500
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 3.22079091174824e-05,
+      "loss": 2.1911,
+      "step": 7185000
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 3.2206670531056314e-05,
+      "loss": 2.1777,
+      "step": 7185500
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 3.220543194463023e-05,
+      "loss": 2.1778,
+      "step": 7186000
+    },
+    {
+      "epoch": 35.6,
+      "learning_rate": 3.220419335820415e-05,
+      "loss": 2.1783,
+      "step": 7186500
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 3.2202954771778065e-05,
+      "loss": 2.2068,
+      "step": 7187000
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 3.220171618535198e-05,
+      "loss": 2.2157,
+      "step": 7187500
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 3.22004775989259e-05,
+      "loss": 2.1735,
+      "step": 7188000
+    },
+    {
+      "epoch": 35.61,
+      "learning_rate": 3.219924148967267e-05,
+      "loss": 2.166,
+      "step": 7188500
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 3.2198002903246585e-05,
+      "loss": 2.195,
+      "step": 7189000
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 3.2196766793993354e-05,
+      "loss": 2.2019,
+      "step": 7189500
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 3.219552820756727e-05,
+      "loss": 2.1698,
+      "step": 7190000
+    },
+    {
+      "epoch": 35.62,
+      "learning_rate": 3.219428962114119e-05,
+      "loss": 2.1708,
+      "step": 7190500
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 3.21930510347151e-05,
+      "loss": 2.198,
+      "step": 7191000
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 3.2191812448289015e-05,
+      "loss": 2.1899,
+      "step": 7191500
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 3.219057633903579e-05,
+      "loss": 2.188,
+      "step": 7192000
+    },
+    {
+      "epoch": 35.63,
+      "learning_rate": 3.218934022978255e-05,
+      "loss": 2.2009,
+      "step": 7192500
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 3.218810164335647e-05,
+      "loss": 2.189,
+      "step": 7193000
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 3.2186863056930386e-05,
+      "loss": 2.1848,
+      "step": 7193500
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 3.21856244705043e-05,
+      "loss": 2.2159,
+      "step": 7194000
+    },
+    {
+      "epoch": 35.64,
+      "learning_rate": 3.218438588407822e-05,
+      "loss": 2.1881,
+      "step": 7194500
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 3.218314729765214e-05,
+      "loss": 2.1793,
+      "step": 7195000
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 3.2181908711226054e-05,
+      "loss": 2.1856,
+      "step": 7195500
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 3.218067012479997e-05,
+      "loss": 2.1974,
+      "step": 7196000
+    },
+    {
+      "epoch": 35.65,
+      "learning_rate": 3.217943153837389e-05,
+      "loss": 2.1806,
+      "step": 7196500
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 3.21781929519478e-05,
+      "loss": 2.1729,
+      "step": 7197000
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 3.2176954365521715e-05,
+      "loss": 2.1895,
+      "step": 7197500
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 3.217571577909563e-05,
+      "loss": 2.2215,
+      "step": 7198000
+    },
+    {
+      "epoch": 35.66,
+      "learning_rate": 3.217447719266955e-05,
+      "loss": 2.1851,
+      "step": 7198500
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 3.2173238606243465e-05,
+      "loss": 2.1652,
+      "step": 7199000
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 3.217200001981738e-05,
+      "loss": 2.192,
+      "step": 7199500
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 3.21707614333913e-05,
+      "loss": 2.2075,
+      "step": 7200000
+    },
+    {
+      "epoch": 35.67,
+      "learning_rate": 3.2169522846965216e-05,
+      "loss": 2.1748,
+      "step": 7200500
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 3.2168286737711985e-05,
+      "loss": 2.1801,
+      "step": 7201000
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 3.21670481512859e-05,
+      "loss": 2.1767,
+      "step": 7201500
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 3.216580956485982e-05,
+      "loss": 2.1897,
+      "step": 7202000
+    },
+    {
+      "epoch": 35.68,
+      "learning_rate": 3.2164570978433736e-05,
+      "loss": 2.2161,
+      "step": 7202500
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 3.216333239200765e-05,
+      "loss": 2.2114,
+      "step": 7203000
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 3.2162096282754415e-05,
+      "loss": 2.2018,
+      "step": 7203500
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 3.216086017350119e-05,
+      "loss": 2.1898,
+      "step": 7204000
+    },
+    {
+      "epoch": 35.69,
+      "learning_rate": 3.215962158707511e-05,
+      "loss": 2.2153,
+      "step": 7204500
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 3.2158385477821876e-05,
+      "loss": 2.1962,
+      "step": 7205000
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 3.2157146891395786e-05,
+      "loss": 2.2066,
+      "step": 7205500
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 3.21559083049697e-05,
+      "loss": 2.1808,
+      "step": 7206000
+    },
+    {
+      "epoch": 35.7,
+      "learning_rate": 3.215466971854362e-05,
+      "loss": 2.1784,
+      "step": 7206500
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 3.215343113211754e-05,
+      "loss": 2.1879,
+      "step": 7207000
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 3.2152192545691454e-05,
+      "loss": 2.2027,
+      "step": 7207500
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 3.215095395926537e-05,
+      "loss": 2.149,
+      "step": 7208000
+    },
+    {
+      "epoch": 35.71,
+      "learning_rate": 3.214971537283929e-05,
+      "loss": 2.1831,
+      "step": 7208500
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 3.2148476786413205e-05,
+      "loss": 2.216,
+      "step": 7209000
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 3.2147238199987115e-05,
+      "loss": 2.216,
+      "step": 7209500
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 3.214600209073389e-05,
+      "loss": 2.1782,
+      "step": 7210000
+    },
+    {
+      "epoch": 35.72,
+      "learning_rate": 3.214476350430781e-05,
+      "loss": 2.1767,
+      "step": 7210500
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 3.2143524917881724e-05,
+      "loss": 2.1726,
+      "step": 7211000
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 3.214228633145564e-05,
+      "loss": 2.1896,
+      "step": 7211500
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 3.214104774502956e-05,
+      "loss": 2.1825,
+      "step": 7212000
+    },
+    {
+      "epoch": 35.73,
+      "learning_rate": 3.2139809158603475e-05,
+      "loss": 2.1887,
+      "step": 7212500
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 3.2138570572177385e-05,
+      "loss": 2.1715,
+      "step": 7213000
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 3.21373319857513e-05,
+      "loss": 2.187,
+      "step": 7213500
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 3.213609339932522e-05,
+      "loss": 2.1838,
+      "step": 7214000
+    },
+    {
+      "epoch": 35.74,
+      "learning_rate": 3.213485729007199e-05,
+      "loss": 2.194,
+      "step": 7214500
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 3.2133618703645905e-05,
+      "loss": 2.1984,
+      "step": 7215000
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 3.213238011721982e-05,
+      "loss": 2.2033,
+      "step": 7215500
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 3.213114153079373e-05,
+      "loss": 2.1934,
+      "step": 7216000
+    },
+    {
+      "epoch": 35.75,
+      "learning_rate": 3.212990294436765e-05,
+      "loss": 2.2022,
+      "step": 7216500
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 3.2128664357941566e-05,
+      "loss": 2.185,
+      "step": 7217000
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 3.212742577151548e-05,
+      "loss": 2.1806,
+      "step": 7217500
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 3.212618966226226e-05,
+      "loss": 2.1936,
+      "step": 7218000
+    },
+    {
+      "epoch": 35.76,
+      "learning_rate": 3.2124951075836175e-05,
+      "loss": 2.163,
+      "step": 7218500
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 3.2123712489410085e-05,
+      "loss": 2.1739,
+      "step": 7219000
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 3.2122473902984e-05,
+      "loss": 2.1609,
+      "step": 7219500
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 3.212123531655792e-05,
+      "loss": 2.1824,
+      "step": 7220000
+    },
+    {
+      "epoch": 35.77,
+      "learning_rate": 3.2119996730131836e-05,
+      "loss": 2.1913,
+      "step": 7220500
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 3.211875814370575e-05,
+      "loss": 2.2127,
+      "step": 7221000
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 3.211751955727967e-05,
+      "loss": 2.1843,
+      "step": 7221500
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 3.211628097085359e-05,
+      "loss": 2.1812,
+      "step": 7222000
+    },
+    {
+      "epoch": 35.78,
+      "learning_rate": 3.211504486160035e-05,
+      "loss": 2.1801,
+      "step": 7222500
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 3.2113806275174266e-05,
+      "loss": 2.1778,
+      "step": 7223000
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 3.211256768874818e-05,
+      "loss": 2.1982,
+      "step": 7223500
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 3.21113291023221e-05,
+      "loss": 2.2053,
+      "step": 7224000
+    },
+    {
+      "epoch": 35.79,
+      "learning_rate": 3.211009051589602e-05,
+      "loss": 2.1966,
+      "step": 7224500
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 3.2108851929469934e-05,
+      "loss": 2.1682,
+      "step": 7225000
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 3.210761334304385e-05,
+      "loss": 2.2173,
+      "step": 7225500
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 3.210637475661777e-05,
+      "loss": 2.1867,
+      "step": 7226000
+    },
+    {
+      "epoch": 35.8,
+      "learning_rate": 3.2105136170191684e-05,
+      "loss": 2.1976,
+      "step": 7226500
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 3.210390006093845e-05,
+      "loss": 2.185,
+      "step": 7227000
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 3.210266147451237e-05,
+      "loss": 2.1922,
+      "step": 7227500
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 3.210142536525914e-05,
+      "loss": 2.1783,
+      "step": 7228000
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 3.210018677883305e-05,
+      "loss": 2.1788,
+      "step": 7228500
+    },
+    {
+      "epoch": 35.81,
+      "learning_rate": 3.2098950669579825e-05,
+      "loss": 2.2044,
+      "step": 7229000
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 3.209771208315374e-05,
+      "loss": 2.1966,
+      "step": 7229500
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 3.209647349672766e-05,
+      "loss": 2.2111,
+      "step": 7230000
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 3.2095234910301575e-05,
+      "loss": 2.1748,
+      "step": 7230500
+    },
+    {
+      "epoch": 35.82,
+      "learning_rate": 3.209400127822119e-05,
+      "loss": 2.1646,
+      "step": 7231000
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 3.2092765168967965e-05,
+      "loss": 2.1985,
+      "step": 7231500
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 3.2091526582541875e-05,
+      "loss": 2.1675,
+      "step": 7232000
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 3.209028799611579e-05,
+      "loss": 2.1727,
+      "step": 7232500
+    },
+    {
+      "epoch": 35.83,
+      "learning_rate": 3.208905188686257e-05,
+      "loss": 2.1669,
+      "step": 7233000
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 3.208781330043648e-05,
+      "loss": 2.19,
+      "step": 7233500
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 3.2086574714010395e-05,
+      "loss": 2.1846,
+      "step": 7234000
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 3.208533612758431e-05,
+      "loss": 2.2084,
+      "step": 7234500
+    },
+    {
+      "epoch": 35.84,
+      "learning_rate": 3.208409754115823e-05,
+      "loss": 2.1889,
+      "step": 7235000
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 3.2082858954732146e-05,
+      "loss": 2.1709,
+      "step": 7235500
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 3.2081620368306056e-05,
+      "loss": 2.192,
+      "step": 7236000
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 3.208038425905283e-05,
+      "loss": 2.1989,
+      "step": 7236500
+    },
+    {
+      "epoch": 35.85,
+      "learning_rate": 3.207914567262675e-05,
+      "loss": 2.1778,
+      "step": 7237000
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 3.2077907086200665e-05,
+      "loss": 2.2039,
+      "step": 7237500
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 3.207666849977458e-05,
+      "loss": 2.1986,
+      "step": 7238000
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 3.207542991334849e-05,
+      "loss": 2.1762,
+      "step": 7238500
+    },
+    {
+      "epoch": 35.86,
+      "learning_rate": 3.207419132692241e-05,
+      "loss": 2.177,
+      "step": 7239000
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 3.2072952740496326e-05,
+      "loss": 2.1804,
+      "step": 7239500
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 3.207171415407024e-05,
+      "loss": 2.194,
+      "step": 7240000
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 3.207047556764416e-05,
+      "loss": 2.1995,
+      "step": 7240500
+    },
+    {
+      "epoch": 35.87,
+      "learning_rate": 3.206923698121808e-05,
+      "loss": 2.227,
+      "step": 7241000
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 3.2067998394791994e-05,
+      "loss": 2.1887,
+      "step": 7241500
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 3.206675980836591e-05,
+      "loss": 2.1928,
+      "step": 7242000
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 3.206552122193983e-05,
+      "loss": 2.1956,
+      "step": 7242500
+    },
+    {
+      "epoch": 35.88,
+      "learning_rate": 3.206428511268659e-05,
+      "loss": 2.217,
+      "step": 7243000
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 3.2063046526260507e-05,
+      "loss": 2.212,
+      "step": 7243500
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 3.206181041700728e-05,
+      "loss": 2.1715,
+      "step": 7244000
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 3.206057183058119e-05,
+      "loss": 2.1658,
+      "step": 7244500
+    },
+    {
+      "epoch": 35.89,
+      "learning_rate": 3.205933324415511e-05,
+      "loss": 2.1871,
+      "step": 7245000
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 3.2058097134901885e-05,
+      "loss": 2.183,
+      "step": 7245500
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 3.20568585484758e-05,
+      "loss": 2.1936,
+      "step": 7246000
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 3.205561996204972e-05,
+      "loss": 2.1959,
+      "step": 7246500
+    },
+    {
+      "epoch": 35.9,
+      "learning_rate": 3.205438137562363e-05,
+      "loss": 2.2226,
+      "step": 7247000
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 3.2053142789197546e-05,
+      "loss": 2.1854,
+      "step": 7247500
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 3.205190420277146e-05,
+      "loss": 2.2058,
+      "step": 7248000
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 3.205066561634538e-05,
+      "loss": 2.1896,
+      "step": 7248500
+    },
+    {
+      "epoch": 35.91,
+      "learning_rate": 3.2049427029919297e-05,
+      "loss": 2.1738,
+      "step": 7249000
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 3.204818844349321e-05,
+      "loss": 2.1808,
+      "step": 7249500
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 3.2046949857067124e-05,
+      "loss": 2.2039,
+      "step": 7250000
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 3.204571127064104e-05,
+      "loss": 2.1845,
+      "step": 7250500
+    },
+    {
+      "epoch": 35.92,
+      "learning_rate": 3.204447268421496e-05,
+      "loss": 2.203,
+      "step": 7251000
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 3.2043234097788874e-05,
+      "loss": 2.1995,
+      "step": 7251500
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 3.204199551136279e-05,
+      "loss": 2.2299,
+      "step": 7252000
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 3.204075692493671e-05,
+      "loss": 2.2146,
+      "step": 7252500
+    },
+    {
+      "epoch": 35.93,
+      "learning_rate": 3.2039518338510625e-05,
+      "loss": 2.2082,
+      "step": 7253000
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 3.2038282229257394e-05,
+      "loss": 2.1813,
+      "step": 7253500
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 3.203704364283131e-05,
+      "loss": 2.1665,
+      "step": 7254000
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 3.203580505640523e-05,
+      "loss": 2.1758,
+      "step": 7254500
+    },
+    {
+      "epoch": 35.94,
+      "learning_rate": 3.2034566469979145e-05,
+      "loss": 2.1813,
+      "step": 7255000
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 3.2033330360725914e-05,
+      "loss": 2.1977,
+      "step": 7255500
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 3.203209177429983e-05,
+      "loss": 2.1928,
+      "step": 7256000
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 3.203085318787374e-05,
+      "loss": 2.2045,
+      "step": 7256500
+    },
+    {
+      "epoch": 35.95,
+      "learning_rate": 3.202961460144766e-05,
+      "loss": 2.1974,
+      "step": 7257000
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 3.2028378492194426e-05,
+      "loss": 2.1832,
+      "step": 7257500
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 3.202713990576834e-05,
+      "loss": 2.1904,
+      "step": 7258000
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 3.202590131934226e-05,
+      "loss": 2.1817,
+      "step": 7258500
+    },
+    {
+      "epoch": 35.96,
+      "learning_rate": 3.202466273291618e-05,
+      "loss": 2.2003,
+      "step": 7259000
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 3.2023424146490094e-05,
+      "loss": 2.1702,
+      "step": 7259500
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 3.202218556006401e-05,
+      "loss": 2.1826,
+      "step": 7260000
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 3.202094697363793e-05,
+      "loss": 2.1929,
+      "step": 7260500
+    },
+    {
+      "epoch": 35.97,
+      "learning_rate": 3.20197108643847e-05,
+      "loss": 2.1713,
+      "step": 7261000
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 3.2018472277958614e-05,
+      "loss": 2.2067,
+      "step": 7261500
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 3.201723369153253e-05,
+      "loss": 2.1851,
+      "step": 7262000
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 3.201599510510645e-05,
+      "loss": 2.2131,
+      "step": 7262500
+    },
+    {
+      "epoch": 35.98,
+      "learning_rate": 3.2014758995853216e-05,
+      "loss": 2.1984,
+      "step": 7263000
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 3.2013520409427127e-05,
+      "loss": 2.1667,
+      "step": 7263500
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 3.2012281823001043e-05,
+      "loss": 2.1926,
+      "step": 7264000
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 3.201104323657496e-05,
+      "loss": 2.2088,
+      "step": 7264500
+    },
+    {
+      "epoch": 35.99,
+      "learning_rate": 3.200980465014888e-05,
+      "loss": 2.1889,
+      "step": 7265000
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.2008566063722794e-05,
+      "loss": 2.1909,
+      "step": 7265500
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.200732747729671e-05,
+      "loss": 2.2,
+      "step": 7266000
+    },
+    {
+      "epoch": 36.0,
+      "eval_accuracy": 0.6622320326905691,
+      "eval_accuracy_mlm": 0.6186748779181579,
+      "eval_accuracy_nsp": 0.8673472989774826,
+      "eval_loss": 2.315115213394165,
+      "eval_runtime": 146.032,
+      "eval_samples_per_second": 1745.911,
+      "eval_steps_per_second": 72.751,
+      "step": 7266348
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.200608889087063e-05,
+      "loss": 2.1755,
+      "step": 7266500
+    },
+    {
+      "epoch": 36.0,
+      "learning_rate": 3.20048527816174e-05,
+      "loss": 2.1472,
+      "step": 7267000
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3.2003614195191314e-05,
+      "loss": 2.1568,
+      "step": 7267500
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3.200237808593808e-05,
+      "loss": 2.1619,
+      "step": 7268000
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3.2001139499512e-05,
+      "loss": 2.1834,
+      "step": 7268500
+    },
+    {
+      "epoch": 36.01,
+      "learning_rate": 3.1999900913085917e-05,
+      "loss": 2.1446,
+      "step": 7269000
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 3.199866232665983e-05,
+      "loss": 2.1727,
+      "step": 7269500
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 3.1997423740233744e-05,
+      "loss": 2.1695,
+      "step": 7270000
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 3.199618515380766e-05,
+      "loss": 2.1422,
+      "step": 7270500
+    },
+    {
+      "epoch": 36.02,
+      "learning_rate": 3.199494656738158e-05,
+      "loss": 2.1672,
+      "step": 7271000
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 3.1993707980955494e-05,
+      "loss": 2.1533,
+      "step": 7271500
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 3.199247187170227e-05,
+      "loss": 2.1539,
+      "step": 7272000
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 3.199123576244903e-05,
+      "loss": 2.1517,
+      "step": 7272500
+    },
+    {
+      "epoch": 36.03,
+      "learning_rate": 3.198999717602295e-05,
+      "loss": 2.1898,
+      "step": 7273000
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 3.1988758589596866e-05,
+      "loss": 2.1593,
+      "step": 7273500
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 3.198752000317078e-05,
+      "loss": 2.1246,
+      "step": 7274000
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 3.19862814167447e-05,
+      "loss": 2.1623,
+      "step": 7274500
+    },
+    {
+      "epoch": 36.04,
+      "learning_rate": 3.198504283031862e-05,
+      "loss": 2.1748,
+      "step": 7275000
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 3.1983804243892534e-05,
+      "loss": 2.1472,
+      "step": 7275500
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 3.1982565657466444e-05,
+      "loss": 2.1667,
+      "step": 7276000
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 3.198132954821322e-05,
+      "loss": 2.185,
+      "step": 7276500
+    },
+    {
+      "epoch": 36.05,
+      "learning_rate": 3.1980090961787136e-05,
+      "loss": 2.1632,
+      "step": 7277000
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 3.197885237536105e-05,
+      "loss": 2.1667,
+      "step": 7277500
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 3.1977616266107815e-05,
+      "loss": 2.1795,
+      "step": 7278000
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 3.197637767968173e-05,
+      "loss": 2.1736,
+      "step": 7278500
+    },
+    {
+      "epoch": 36.06,
+      "learning_rate": 3.197513909325565e-05,
+      "loss": 2.1804,
+      "step": 7279000
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 3.1973900506829566e-05,
+      "loss": 2.1707,
+      "step": 7279500
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 3.197266192040348e-05,
+      "loss": 2.16,
+      "step": 7280000
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 3.19714233339774e-05,
+      "loss": 2.1459,
+      "step": 7280500
+    },
+    {
+      "epoch": 36.07,
+      "learning_rate": 3.197018474755132e-05,
+      "loss": 2.176,
+      "step": 7281000
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 3.1968946161125234e-05,
+      "loss": 2.188,
+      "step": 7281500
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 3.1967707574699144e-05,
+      "loss": 2.1487,
+      "step": 7282000
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 3.196646898827306e-05,
+      "loss": 2.1741,
+      "step": 7282500
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 3.196523040184698e-05,
+      "loss": 2.1627,
+      "step": 7283000
+    },
+    {
+      "epoch": 36.08,
+      "learning_rate": 3.1963991815420895e-05,
+      "loss": 2.1524,
+      "step": 7283500
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 3.196275322899481e-05,
+      "loss": 2.1659,
+      "step": 7284000
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 3.196151464256873e-05,
+      "loss": 2.1753,
+      "step": 7284500
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 3.1960276056142645e-05,
+      "loss": 2.1853,
+      "step": 7285000
+    },
+    {
+      "epoch": 36.09,
+      "learning_rate": 3.195903746971656e-05,
+      "loss": 2.1606,
+      "step": 7285500
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 3.195779888329048e-05,
+      "loss": 2.1595,
+      "step": 7286000
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 3.195656277403725e-05,
+      "loss": 2.1824,
+      "step": 7286500
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 3.1955324187611165e-05,
+      "loss": 2.1565,
+      "step": 7287000
+    },
+    {
+      "epoch": 36.1,
+      "learning_rate": 3.195408560118508e-05,
+      "loss": 2.1437,
+      "step": 7287500
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 3.195284949193185e-05,
+      "loss": 2.1699,
+      "step": 7288000
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 3.195161090550576e-05,
+      "loss": 2.1642,
+      "step": 7288500
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 3.195037231907968e-05,
+      "loss": 2.1632,
+      "step": 7289000
+    },
+    {
+      "epoch": 36.11,
+      "learning_rate": 3.1949133732653595e-05,
+      "loss": 2.1737,
+      "step": 7289500
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 3.194789514622751e-05,
+      "loss": 2.1658,
+      "step": 7290000
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 3.194665655980143e-05,
+      "loss": 2.1809,
+      "step": 7290500
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 3.1945417973375346e-05,
+      "loss": 2.1634,
+      "step": 7291000
+    },
+    {
+      "epoch": 36.12,
+      "learning_rate": 3.194417938694926e-05,
+      "loss": 2.151,
+      "step": 7291500
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 3.194294080052318e-05,
+      "loss": 2.1942,
+      "step": 7292000
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 3.194170469126995e-05,
+      "loss": 2.1557,
+      "step": 7292500
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 3.1940466104843865e-05,
+      "loss": 2.1469,
+      "step": 7293000
+    },
+    {
+      "epoch": 36.13,
+      "learning_rate": 3.193922751841778e-05,
+      "loss": 2.1525,
+      "step": 7293500
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 3.19379889319917e-05,
+      "loss": 2.1464,
+      "step": 7294000
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 3.1936750345565616e-05,
+      "loss": 2.1576,
+      "step": 7294500
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 3.193551175913953e-05,
+      "loss": 2.1805,
+      "step": 7295000
+    },
+    {
+      "epoch": 36.14,
+      "learning_rate": 3.1934278127059154e-05,
+      "loss": 2.1826,
+      "step": 7295500
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 3.193303954063307e-05,
+      "loss": 2.1589,
+      "step": 7296000
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 3.193180095420699e-05,
+      "loss": 2.1538,
+      "step": 7296500
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 3.1930562367780904e-05,
+      "loss": 2.1704,
+      "step": 7297000
+    },
+    {
+      "epoch": 36.15,
+      "learning_rate": 3.192932378135482e-05,
+      "loss": 2.1759,
+      "step": 7297500
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 3.192808519492873e-05,
+      "loss": 2.1512,
+      "step": 7298000
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 3.192684660850265e-05,
+      "loss": 2.1743,
+      "step": 7298500
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 3.1925608022076565e-05,
+      "loss": 2.1707,
+      "step": 7299000
+    },
+    {
+      "epoch": 36.16,
+      "learning_rate": 3.192436943565048e-05,
+      "loss": 2.1599,
+      "step": 7299500
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3.19231308492244e-05,
+      "loss": 2.1745,
+      "step": 7300000
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3.192189473997117e-05,
+      "loss": 2.1474,
+      "step": 7300500
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3.192065615354508e-05,
+      "loss": 2.1428,
+      "step": 7301000
+    },
+    {
+      "epoch": 36.17,
+      "learning_rate": 3.1919420044291854e-05,
+      "loss": 2.1729,
+      "step": 7301500
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 3.191818145786577e-05,
+      "loss": 2.1742,
+      "step": 7302000
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 3.191694287143969e-05,
+      "loss": 2.1903,
+      "step": 7302500
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 3.1915704285013604e-05,
+      "loss": 2.1836,
+      "step": 7303000
+    },
+    {
+      "epoch": 36.18,
+      "learning_rate": 3.191446817576037e-05,
+      "loss": 2.1727,
+      "step": 7303500
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 3.191322958933429e-05,
+      "loss": 2.1738,
+      "step": 7304000
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 3.19119910029082e-05,
+      "loss": 2.165,
+      "step": 7304500
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 3.191075241648212e-05,
+      "loss": 2.171,
+      "step": 7305000
+    },
+    {
+      "epoch": 36.19,
+      "learning_rate": 3.1909513830056034e-05,
+      "loss": 2.1837,
+      "step": 7305500
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 3.190827524362995e-05,
+      "loss": 2.1613,
+      "step": 7306000
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 3.190703665720387e-05,
+      "loss": 2.1567,
+      "step": 7306500
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 3.190579807077778e-05,
+      "loss": 2.1435,
+      "step": 7307000
+    },
+    {
+      "epoch": 36.2,
+      "learning_rate": 3.1904559484351695e-05,
+      "loss": 2.1601,
+      "step": 7307500
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 3.190332089792561e-05,
+      "loss": 2.1849,
+      "step": 7308000
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 3.190208231149953e-05,
+      "loss": 2.1669,
+      "step": 7308500
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 3.1900843725073446e-05,
+      "loss": 2.1879,
+      "step": 7309000
+    },
+    {
+      "epoch": 36.21,
+      "learning_rate": 3.189960513864736e-05,
+      "loss": 2.1559,
+      "step": 7309500
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 3.189836902939414e-05,
+      "loss": 2.1691,
+      "step": 7310000
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 3.189713292014091e-05,
+      "loss": 2.163,
+      "step": 7310500
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 3.1895894333714824e-05,
+      "loss": 2.1601,
+      "step": 7311000
+    },
+    {
+      "epoch": 36.22,
+      "learning_rate": 3.1894655747288734e-05,
+      "loss": 2.1817,
+      "step": 7311500
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 3.189341716086265e-05,
+      "loss": 2.197,
+      "step": 7312000
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 3.189217857443657e-05,
+      "loss": 2.1816,
+      "step": 7312500
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 3.1890939988010485e-05,
+      "loss": 2.152,
+      "step": 7313000
+    },
+    {
+      "epoch": 36.23,
+      "learning_rate": 3.18897014015844e-05,
+      "loss": 2.1727,
+      "step": 7313500
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 3.188846281515831e-05,
+      "loss": 2.1761,
+      "step": 7314000
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 3.188722670590509e-05,
+      "loss": 2.1642,
+      "step": 7314500
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 3.1885988119479005e-05,
+      "loss": 2.1726,
+      "step": 7315000
+    },
+    {
+      "epoch": 36.24,
+      "learning_rate": 3.1884752010225773e-05,
+      "loss": 2.1537,
+      "step": 7315500
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 3.188351342379969e-05,
+      "loss": 2.1697,
+      "step": 7316000
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 3.188227483737361e-05,
+      "loss": 2.179,
+      "step": 7316500
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 3.1881036250947524e-05,
+      "loss": 2.1536,
+      "step": 7317000
+    },
+    {
+      "epoch": 36.25,
+      "learning_rate": 3.187979766452144e-05,
+      "loss": 2.1655,
+      "step": 7317500
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 3.18785615552682e-05,
+      "loss": 2.1739,
+      "step": 7318000
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 3.187732296884212e-05,
+      "loss": 2.1631,
+      "step": 7318500
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 3.187608685958889e-05,
+      "loss": 2.164,
+      "step": 7319000
+    },
+    {
+      "epoch": 36.26,
+      "learning_rate": 3.1874848273162806e-05,
+      "loss": 2.1755,
+      "step": 7319500
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 3.187360968673672e-05,
+      "loss": 2.1961,
+      "step": 7320000
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 3.187237357748349e-05,
+      "loss": 2.1747,
+      "step": 7320500
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 3.187113499105741e-05,
+      "loss": 2.1745,
+      "step": 7321000
+    },
+    {
+      "epoch": 36.27,
+      "learning_rate": 3.186989640463132e-05,
+      "loss": 2.146,
+      "step": 7321500
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 3.1868657818205236e-05,
+      "loss": 2.1678,
+      "step": 7322000
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 3.186741923177915e-05,
+      "loss": 2.1745,
+      "step": 7322500
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 3.186618312252593e-05,
+      "loss": 2.1547,
+      "step": 7323000
+    },
+    {
+      "epoch": 36.28,
+      "learning_rate": 3.186494453609984e-05,
+      "loss": 2.1704,
+      "step": 7323500
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 3.1863705949673755e-05,
+      "loss": 2.1843,
+      "step": 7324000
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 3.186246736324767e-05,
+      "loss": 2.1645,
+      "step": 7324500
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 3.186122877682159e-05,
+      "loss": 2.1699,
+      "step": 7325000
+    },
+    {
+      "epoch": 36.29,
+      "learning_rate": 3.1859990190395506e-05,
+      "loss": 2.188,
+      "step": 7325500
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3.1858754081142275e-05,
+      "loss": 2.172,
+      "step": 7326000
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3.185751549471619e-05,
+      "loss": 2.1751,
+      "step": 7326500
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3.185627690829011e-05,
+      "loss": 2.1842,
+      "step": 7327000
+    },
+    {
+      "epoch": 36.3,
+      "learning_rate": 3.185504079903688e-05,
+      "loss": 2.1814,
+      "step": 7327500
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 3.1853802212610794e-05,
+      "loss": 2.1958,
+      "step": 7328000
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 3.185256362618471e-05,
+      "loss": 2.1448,
+      "step": 7328500
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 3.185132503975863e-05,
+      "loss": 2.1794,
+      "step": 7329000
+    },
+    {
+      "epoch": 36.31,
+      "learning_rate": 3.1850086453332545e-05,
+      "loss": 2.1839,
+      "step": 7329500
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 3.1848847866906455e-05,
+      "loss": 2.2051,
+      "step": 7330000
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 3.184760928048037e-05,
+      "loss": 2.1578,
+      "step": 7330500
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 3.184637069405429e-05,
+      "loss": 2.1948,
+      "step": 7331000
+    },
+    {
+      "epoch": 36.32,
+      "learning_rate": 3.1845132107628206e-05,
+      "loss": 2.1852,
+      "step": 7331500
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 3.184389599837498e-05,
+      "loss": 2.1675,
+      "step": 7332000
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 3.184265741194889e-05,
+      "loss": 2.1613,
+      "step": 7332500
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 3.184141882552281e-05,
+      "loss": 2.1787,
+      "step": 7333000
+    },
+    {
+      "epoch": 36.33,
+      "learning_rate": 3.1840180239096726e-05,
+      "loss": 2.1896,
+      "step": 7333500
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 3.183894165267064e-05,
+      "loss": 2.1535,
+      "step": 7334000
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 3.183770306624456e-05,
+      "loss": 2.1596,
+      "step": 7334500
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 3.183646447981847e-05,
+      "loss": 2.1774,
+      "step": 7335000
+    },
+    {
+      "epoch": 36.34,
+      "learning_rate": 3.183522589339239e-05,
+      "loss": 2.1598,
+      "step": 7335500
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 3.1833987306966304e-05,
+      "loss": 2.1618,
+      "step": 7336000
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 3.183274872054022e-05,
+      "loss": 2.1732,
+      "step": 7336500
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 3.183151013411414e-05,
+      "loss": 2.2031,
+      "step": 7337000
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 3.1830274024860906e-05,
+      "loss": 2.1618,
+      "step": 7337500
+    },
+    {
+      "epoch": 36.35,
+      "learning_rate": 3.182903543843482e-05,
+      "loss": 2.1865,
+      "step": 7338000
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 3.182779685200874e-05,
+      "loss": 2.1547,
+      "step": 7338500
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 3.182655826558266e-05,
+      "loss": 2.1551,
+      "step": 7339000
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 3.1825319679156574e-05,
+      "loss": 2.169,
+      "step": 7339500
+    },
+    {
+      "epoch": 36.36,
+      "learning_rate": 3.182408356990334e-05,
+      "loss": 2.1587,
+      "step": 7340000
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 3.182284498347726e-05,
+      "loss": 2.1691,
+      "step": 7340500
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 3.182160639705118e-05,
+      "loss": 2.161,
+      "step": 7341000
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 3.1820367810625094e-05,
+      "loss": 2.1649,
+      "step": 7341500
+    },
+    {
+      "epoch": 36.37,
+      "learning_rate": 3.181913170137186e-05,
+      "loss": 2.1787,
+      "step": 7342000
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 3.181789311494577e-05,
+      "loss": 2.1742,
+      "step": 7342500
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 3.181665452851969e-05,
+      "loss": 2.1531,
+      "step": 7343000
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 3.1815415942093606e-05,
+      "loss": 2.1657,
+      "step": 7343500
+    },
+    {
+      "epoch": 36.38,
+      "learning_rate": 3.181417735566752e-05,
+      "loss": 2.1664,
+      "step": 7344000
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 3.181293876924144e-05,
+      "loss": 2.1728,
+      "step": 7344500
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 3.181170018281536e-05,
+      "loss": 2.1725,
+      "step": 7345000
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 3.1810461596389274e-05,
+      "loss": 2.1948,
+      "step": 7345500
+    },
+    {
+      "epoch": 36.39,
+      "learning_rate": 3.180922300996319e-05,
+      "loss": 2.1758,
+      "step": 7346000
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 3.180798442353711e-05,
+      "loss": 2.1829,
+      "step": 7346500
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 3.1806745837111025e-05,
+      "loss": 2.2015,
+      "step": 7347000
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 3.180550725068494e-05,
+      "loss": 2.1804,
+      "step": 7347500
+    },
+    {
+      "epoch": 36.4,
+      "learning_rate": 3.180426866425886e-05,
+      "loss": 2.179,
+      "step": 7348000
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 3.1803030077832776e-05,
+      "loss": 2.1528,
+      "step": 7348500
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 3.180179396857954e-05,
+      "loss": 2.193,
+      "step": 7349000
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 3.1800555382153455e-05,
+      "loss": 2.154,
+      "step": 7349500
+    },
+    {
+      "epoch": 36.41,
+      "learning_rate": 3.179931679572737e-05,
+      "loss": 2.1824,
+      "step": 7350000
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 3.179808068647414e-05,
+      "loss": 2.1668,
+      "step": 7350500
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 3.179684210004806e-05,
+      "loss": 2.1718,
+      "step": 7351000
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 3.1795603513621974e-05,
+      "loss": 2.1842,
+      "step": 7351500
+    },
+    {
+      "epoch": 36.42,
+      "learning_rate": 3.179436492719589e-05,
+      "loss": 2.1563,
+      "step": 7352000
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 3.179312634076981e-05,
+      "loss": 2.1707,
+      "step": 7352500
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 3.1791887754343725e-05,
+      "loss": 2.1689,
+      "step": 7353000
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 3.179064916791764e-05,
+      "loss": 2.1638,
+      "step": 7353500
+    },
+    {
+      "epoch": 36.43,
+      "learning_rate": 3.178941058149156e-05,
+      "loss": 2.1388,
+      "step": 7354000
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 3.1788171995065476e-05,
+      "loss": 2.1856,
+      "step": 7354500
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 3.178693340863939e-05,
+      "loss": 2.1731,
+      "step": 7355000
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 3.1785697299386155e-05,
+      "loss": 2.1694,
+      "step": 7355500
+    },
+    {
+      "epoch": 36.44,
+      "learning_rate": 3.1784461190132924e-05,
+      "loss": 2.1709,
+      "step": 7356000
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 3.178322260370684e-05,
+      "loss": 2.1522,
+      "step": 7356500
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 3.178198401728076e-05,
+      "loss": 2.1841,
+      "step": 7357000
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 3.1780745430854674e-05,
+      "loss": 2.1854,
+      "step": 7357500
+    },
+    {
+      "epoch": 36.45,
+      "learning_rate": 3.177950932160144e-05,
+      "loss": 2.2015,
+      "step": 7358000
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 3.177827073517536e-05,
+      "loss": 2.1727,
+      "step": 7358500
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 3.177703214874928e-05,
+      "loss": 2.1739,
+      "step": 7359000
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 3.17757985166689e-05,
+      "loss": 2.1939,
+      "step": 7359500
+    },
+    {
+      "epoch": 36.46,
+      "learning_rate": 3.1774559930242815e-05,
+      "loss": 2.1778,
+      "step": 7360000
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 3.177332134381673e-05,
+      "loss": 2.1905,
+      "step": 7360500
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 3.177208275739065e-05,
+      "loss": 2.1869,
+      "step": 7361000
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 3.1770844170964565e-05,
+      "loss": 2.1856,
+      "step": 7361500
+    },
+    {
+      "epoch": 36.47,
+      "learning_rate": 3.176960558453848e-05,
+      "loss": 2.1606,
+      "step": 7362000
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 3.17683669981124e-05,
+      "loss": 2.1803,
+      "step": 7362500
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 3.1767128411686316e-05,
+      "loss": 2.1836,
+      "step": 7363000
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 3.176588982526023e-05,
+      "loss": 2.1588,
+      "step": 7363500
+    },
+    {
+      "epoch": 36.48,
+      "learning_rate": 3.176465123883414e-05,
+      "loss": 2.1725,
+      "step": 7364000
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 3.176341512958091e-05,
+      "loss": 2.1961,
+      "step": 7364500
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 3.176217654315483e-05,
+      "loss": 2.1519,
+      "step": 7365000
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 3.1760937956728746e-05,
+      "loss": 2.1619,
+      "step": 7365500
+    },
+    {
+      "epoch": 36.49,
+      "learning_rate": 3.175969937030266e-05,
+      "loss": 2.1723,
+      "step": 7366000
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 3.175846078387658e-05,
+      "loss": 2.1882,
+      "step": 7366500
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 3.17572221974505e-05,
+      "loss": 2.1582,
+      "step": 7367000
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 3.175598361102441e-05,
+      "loss": 2.171,
+      "step": 7367500
+    },
+    {
+      "epoch": 36.5,
+      "learning_rate": 3.1754745024598324e-05,
+      "loss": 2.174,
+      "step": 7368000
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 3.175350643817224e-05,
+      "loss": 2.1664,
+      "step": 7368500
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 3.175226785174616e-05,
+      "loss": 2.1808,
+      "step": 7369000
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 3.1751029265320075e-05,
+      "loss": 2.212,
+      "step": 7369500
+    },
+    {
+      "epoch": 36.51,
+      "learning_rate": 3.174979315606685e-05,
+      "loss": 2.1871,
+      "step": 7370000
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 3.174855456964076e-05,
+      "loss": 2.1889,
+      "step": 7370500
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 3.174731598321468e-05,
+      "loss": 2.1857,
+      "step": 7371000
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 3.1746077396788594e-05,
+      "loss": 2.1746,
+      "step": 7371500
+    },
+    {
+      "epoch": 36.52,
+      "learning_rate": 3.174483881036251e-05,
+      "loss": 2.1546,
+      "step": 7372000
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 3.174360022393643e-05,
+      "loss": 2.1701,
+      "step": 7372500
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 3.1742361637510345e-05,
+      "loss": 2.1688,
+      "step": 7373000
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 3.174112552825711e-05,
+      "loss": 2.1697,
+      "step": 7373500
+    },
+    {
+      "epoch": 36.53,
+      "learning_rate": 3.1739886941831024e-05,
+      "loss": 2.1417,
+      "step": 7374000
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 3.173864835540494e-05,
+      "loss": 2.1972,
+      "step": 7374500
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 3.1737412246151716e-05,
+      "loss": 2.1895,
+      "step": 7375000
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 3.1736173659725633e-05,
+      "loss": 2.1635,
+      "step": 7375500
+    },
+    {
+      "epoch": 36.54,
+      "learning_rate": 3.17349375504724e-05,
+      "loss": 2.1726,
+      "step": 7376000
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 3.173369896404631e-05,
+      "loss": 2.191,
+      "step": 7376500
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 3.173246037762023e-05,
+      "loss": 2.1821,
+      "step": 7377000
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 3.1731221791194146e-05,
+      "loss": 2.1846,
+      "step": 7377500
+    },
+    {
+      "epoch": 36.55,
+      "learning_rate": 3.172998320476806e-05,
+      "loss": 2.1596,
+      "step": 7378000
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 3.172874461834198e-05,
+      "loss": 2.1764,
+      "step": 7378500
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 3.17275060319159e-05,
+      "loss": 2.1806,
+      "step": 7379000
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 3.1726267445489814e-05,
+      "loss": 2.1604,
+      "step": 7379500
+    },
+    {
+      "epoch": 36.56,
+      "learning_rate": 3.1725028859063724e-05,
+      "loss": 2.2086,
+      "step": 7380000
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 3.172379027263764e-05,
+      "loss": 2.1964,
+      "step": 7380500
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 3.172255168621156e-05,
+      "loss": 2.1792,
+      "step": 7381000
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 3.1721313099785475e-05,
+      "loss": 2.17,
+      "step": 7381500
+    },
+    {
+      "epoch": 36.57,
+      "learning_rate": 3.172007451335939e-05,
+      "loss": 2.1855,
+      "step": 7382000
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 3.171883592693331e-05,
+      "loss": 2.2097,
+      "step": 7382500
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 3.1717597340507226e-05,
+      "loss": 2.1862,
+      "step": 7383000
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 3.171635875408114e-05,
+      "loss": 2.184,
+      "step": 7383500
+    },
+    {
+      "epoch": 36.58,
+      "learning_rate": 3.171512016765506e-05,
+      "loss": 2.1804,
+      "step": 7384000
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 3.171388405840183e-05,
+      "loss": 2.1422,
+      "step": 7384500
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 3.17126479491486e-05,
+      "loss": 2.1633,
+      "step": 7385000
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 3.1711411839895366e-05,
+      "loss": 2.1765,
+      "step": 7385500
+    },
+    {
+      "epoch": 36.59,
+      "learning_rate": 3.171017325346928e-05,
+      "loss": 2.1716,
+      "step": 7386000
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 3.17089346670432e-05,
+      "loss": 2.1797,
+      "step": 7386500
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 3.170769608061712e-05,
+      "loss": 2.2091,
+      "step": 7387000
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 3.1706457494191034e-05,
+      "loss": 2.1754,
+      "step": 7387500
+    },
+    {
+      "epoch": 36.6,
+      "learning_rate": 3.170521890776495e-05,
+      "loss": 2.1538,
+      "step": 7388000
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 3.170398032133887e-05,
+      "loss": 2.1559,
+      "step": 7388500
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 3.170274173491278e-05,
+      "loss": 2.1987,
+      "step": 7389000
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 3.170150562565955e-05,
+      "loss": 2.194,
+      "step": 7389500
+    },
+    {
+      "epoch": 36.61,
+      "learning_rate": 3.1700267039233463e-05,
+      "loss": 2.1804,
+      "step": 7390000
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 3.169902845280738e-05,
+      "loss": 2.2139,
+      "step": 7390500
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 3.16977898663813e-05,
+      "loss": 2.1842,
+      "step": 7391000
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 3.1696553757128066e-05,
+      "loss": 2.1761,
+      "step": 7391500
+    },
+    {
+      "epoch": 36.62,
+      "learning_rate": 3.169531517070198e-05,
+      "loss": 2.1856,
+      "step": 7392000
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 3.16940765842759e-05,
+      "loss": 2.1945,
+      "step": 7392500
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 3.169283799784982e-05,
+      "loss": 2.1783,
+      "step": 7393000
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 3.1691599411423734e-05,
+      "loss": 2.182,
+      "step": 7393500
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 3.169036082499765e-05,
+      "loss": 2.1868,
+      "step": 7394000
+    },
+    {
+      "epoch": 36.63,
+      "learning_rate": 3.168912223857157e-05,
+      "loss": 2.1836,
+      "step": 7394500
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 3.1687886129318336e-05,
+      "loss": 2.1808,
+      "step": 7395000
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 3.168664754289225e-05,
+      "loss": 2.1812,
+      "step": 7395500
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 3.168540895646617e-05,
+      "loss": 2.1927,
+      "step": 7396000
+    },
+    {
+      "epoch": 36.64,
+      "learning_rate": 3.168417037004008e-05,
+      "loss": 2.2173,
+      "step": 7396500
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 3.1682931783614e-05,
+      "loss": 2.168,
+      "step": 7397000
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 3.1681693197187914e-05,
+      "loss": 2.1856,
+      "step": 7397500
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 3.168045461076183e-05,
+      "loss": 2.1885,
+      "step": 7398000
+    },
+    {
+      "epoch": 36.65,
+      "learning_rate": 3.167921602433575e-05,
+      "loss": 2.1744,
+      "step": 7398500
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 3.167797743790966e-05,
+      "loss": 2.1875,
+      "step": 7399000
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 3.1676738851483575e-05,
+      "loss": 2.1897,
+      "step": 7399500
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 3.167550026505749e-05,
+      "loss": 2.1933,
+      "step": 7400000
+    },
+    {
+      "epoch": 36.66,
+      "learning_rate": 3.167426415580427e-05,
+      "loss": 2.1792,
+      "step": 7400500
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 3.1673025569378185e-05,
+      "loss": 2.1766,
+      "step": 7401000
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 3.1671786982952095e-05,
+      "loss": 2.1659,
+      "step": 7401500
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 3.167054839652601e-05,
+      "loss": 2.169,
+      "step": 7402000
+    },
+    {
+      "epoch": 36.67,
+      "learning_rate": 3.166930981009993e-05,
+      "loss": 2.1943,
+      "step": 7402500
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 3.1668071223673846e-05,
+      "loss": 2.1795,
+      "step": 7403000
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 3.166683263724776e-05,
+      "loss": 2.1787,
+      "step": 7403500
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 3.166559405082168e-05,
+      "loss": 2.1887,
+      "step": 7404000
+    },
+    {
+      "epoch": 36.68,
+      "learning_rate": 3.1664355464395596e-05,
+      "loss": 2.1796,
+      "step": 7404500
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 3.1663119355142365e-05,
+      "loss": 2.1882,
+      "step": 7405000
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 3.166188076871628e-05,
+      "loss": 2.1677,
+      "step": 7405500
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 3.166064465946305e-05,
+      "loss": 2.183,
+      "step": 7406000
+    },
+    {
+      "epoch": 36.69,
+      "learning_rate": 3.165940607303697e-05,
+      "loss": 2.1747,
+      "step": 7406500
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 3.1658167486610885e-05,
+      "loss": 2.1711,
+      "step": 7407000
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 3.16569289001848e-05,
+      "loss": 2.1921,
+      "step": 7407500
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 3.165569031375871e-05,
+      "loss": 2.2008,
+      "step": 7408000
+    },
+    {
+      "epoch": 36.7,
+      "learning_rate": 3.165445172733263e-05,
+      "loss": 2.1975,
+      "step": 7408500
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 3.1653213140906546e-05,
+      "loss": 2.1755,
+      "step": 7409000
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 3.1651979508826166e-05,
+      "loss": 2.1676,
+      "step": 7409500
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 3.165074092240008e-05,
+      "loss": 2.1594,
+      "step": 7410000
+    },
+    {
+      "epoch": 36.71,
+      "learning_rate": 3.1649502335974e-05,
+      "loss": 2.1739,
+      "step": 7410500
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 3.164826374954792e-05,
+      "loss": 2.1926,
+      "step": 7411000
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 3.1647025163121834e-05,
+      "loss": 2.1763,
+      "step": 7411500
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 3.164578657669575e-05,
+      "loss": 2.192,
+      "step": 7412000
+    },
+    {
+      "epoch": 36.72,
+      "learning_rate": 3.164455046744252e-05,
+      "loss": 2.1917,
+      "step": 7412500
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 3.164331188101644e-05,
+      "loss": 2.1724,
+      "step": 7413000
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 3.16420757717632e-05,
+      "loss": 2.2043,
+      "step": 7413500
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 3.1640837185337116e-05,
+      "loss": 2.1866,
+      "step": 7414000
+    },
+    {
+      "epoch": 36.73,
+      "learning_rate": 3.163959859891103e-05,
+      "loss": 2.1808,
+      "step": 7414500
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 3.163836001248495e-05,
+      "loss": 2.1876,
+      "step": 7415000
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 3.1637121426058867e-05,
+      "loss": 2.1957,
+      "step": 7415500
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 3.1635882839632783e-05,
+      "loss": 2.1583,
+      "step": 7416000
+    },
+    {
+      "epoch": 36.74,
+      "learning_rate": 3.163464673037955e-05,
+      "loss": 2.1689,
+      "step": 7416500
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 3.163340814395347e-05,
+      "loss": 2.1737,
+      "step": 7417000
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 3.1632169557527386e-05,
+      "loss": 2.2015,
+      "step": 7417500
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 3.16309309711013e-05,
+      "loss": 2.1799,
+      "step": 7418000
+    },
+    {
+      "epoch": 36.75,
+      "learning_rate": 3.162969238467522e-05,
+      "loss": 2.1741,
+      "step": 7418500
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 3.162845379824914e-05,
+      "loss": 2.1589,
+      "step": 7419000
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 3.1627215211823054e-05,
+      "loss": 2.1929,
+      "step": 7419500
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 3.162597662539697e-05,
+      "loss": 2.1637,
+      "step": 7420000
+    },
+    {
+      "epoch": 36.76,
+      "learning_rate": 3.162473803897089e-05,
+      "loss": 2.1771,
+      "step": 7420500
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 3.1623499452544805e-05,
+      "loss": 2.193,
+      "step": 7421000
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 3.162226086611872e-05,
+      "loss": 2.197,
+      "step": 7421500
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 3.162102227969264e-05,
+      "loss": 2.179,
+      "step": 7422000
+    },
+    {
+      "epoch": 36.77,
+      "learning_rate": 3.16197861704394e-05,
+      "loss": 2.1923,
+      "step": 7422500
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 3.161854758401332e-05,
+      "loss": 2.2144,
+      "step": 7423000
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 3.1617308997587234e-05,
+      "loss": 2.1794,
+      "step": 7423500
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 3.161607041116115e-05,
+      "loss": 2.1906,
+      "step": 7424000
+    },
+    {
+      "epoch": 36.78,
+      "learning_rate": 3.161483182473507e-05,
+      "loss": 2.177,
+      "step": 7424500
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 3.1613593238308985e-05,
+      "loss": 2.1639,
+      "step": 7425000
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 3.16123546518829e-05,
+      "loss": 2.1855,
+      "step": 7425500
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 3.161111606545682e-05,
+      "loss": 2.1537,
+      "step": 7426000
+    },
+    {
+      "epoch": 36.79,
+      "learning_rate": 3.160987747903073e-05,
+      "loss": 2.1746,
+      "step": 7426500
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 3.1608638892604646e-05,
+      "loss": 2.1806,
+      "step": 7427000
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 3.160740030617856e-05,
+      "loss": 2.1547,
+      "step": 7427500
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 3.160616171975248e-05,
+      "loss": 2.1845,
+      "step": 7428000
+    },
+    {
+      "epoch": 36.8,
+      "learning_rate": 3.1604925610499256e-05,
+      "loss": 2.1727,
+      "step": 7428500
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 3.160368702407317e-05,
+      "loss": 2.1838,
+      "step": 7429000
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 3.160244843764709e-05,
+      "loss": 2.1918,
+      "step": 7429500
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 3.1601209851221e-05,
+      "loss": 2.2049,
+      "step": 7430000
+    },
+    {
+      "epoch": 36.81,
+      "learning_rate": 3.1599971264794916e-05,
+      "loss": 2.1743,
+      "step": 7430500
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 3.1598735155541685e-05,
+      "loss": 2.1785,
+      "step": 7431000
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 3.15974965691156e-05,
+      "loss": 2.1674,
+      "step": 7431500
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 3.159625798268952e-05,
+      "loss": 2.1937,
+      "step": 7432000
+    },
+    {
+      "epoch": 36.82,
+      "learning_rate": 3.1595019396263436e-05,
+      "loss": 2.1767,
+      "step": 7432500
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 3.1593780809837346e-05,
+      "loss": 2.1666,
+      "step": 7433000
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 3.159254470058412e-05,
+      "loss": 2.1748,
+      "step": 7433500
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 3.159130611415804e-05,
+      "loss": 2.2004,
+      "step": 7434000
+    },
+    {
+      "epoch": 36.83,
+      "learning_rate": 3.1590067527731956e-05,
+      "loss": 2.1619,
+      "step": 7434500
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 3.158883141847872e-05,
+      "loss": 2.1845,
+      "step": 7435000
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 3.1587592832052635e-05,
+      "loss": 2.1957,
+      "step": 7435500
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 3.1586356722799403e-05,
+      "loss": 2.1968,
+      "step": 7436000
+    },
+    {
+      "epoch": 36.84,
+      "learning_rate": 3.158511813637332e-05,
+      "loss": 2.1949,
+      "step": 7436500
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 3.158387954994724e-05,
+      "loss": 2.183,
+      "step": 7437000
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 3.1582640963521154e-05,
+      "loss": 2.1959,
+      "step": 7437500
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 3.158140485426792e-05,
+      "loss": 2.1838,
+      "step": 7438000
+    },
+    {
+      "epoch": 36.85,
+      "learning_rate": 3.158016874501469e-05,
+      "loss": 2.1668,
+      "step": 7438500
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 3.157893015858861e-05,
+      "loss": 2.1742,
+      "step": 7439000
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 3.1577691572162526e-05,
+      "loss": 2.2019,
+      "step": 7439500
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 3.1576452985736436e-05,
+      "loss": 2.1639,
+      "step": 7440000
+    },
+    {
+      "epoch": 36.86,
+      "learning_rate": 3.157521687648321e-05,
+      "loss": 2.1704,
+      "step": 7440500
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 3.157397829005713e-05,
+      "loss": 2.1903,
+      "step": 7441000
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 3.1572739703631045e-05,
+      "loss": 2.2107,
+      "step": 7441500
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 3.157150111720496e-05,
+      "loss": 2.1606,
+      "step": 7442000
+    },
+    {
+      "epoch": 36.87,
+      "learning_rate": 3.157026253077888e-05,
+      "loss": 2.1743,
+      "step": 7442500
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 3.156902394435279e-05,
+      "loss": 2.1895,
+      "step": 7443000
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 3.1567785357926706e-05,
+      "loss": 2.2057,
+      "step": 7443500
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 3.156654677150062e-05,
+      "loss": 2.1841,
+      "step": 7444000
+    },
+    {
+      "epoch": 36.88,
+      "learning_rate": 3.156530818507454e-05,
+      "loss": 2.1785,
+      "step": 7444500
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 3.156406959864846e-05,
+      "loss": 2.1737,
+      "step": 7445000
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 3.1562831012222374e-05,
+      "loss": 2.1713,
+      "step": 7445500
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 3.156159242579629e-05,
+      "loss": 2.17,
+      "step": 7446000
+    },
+    {
+      "epoch": 36.89,
+      "learning_rate": 3.156035383937021e-05,
+      "loss": 2.1862,
+      "step": 7446500
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 3.1559115252944125e-05,
+      "loss": 2.1914,
+      "step": 7447000
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 3.1557876666518035e-05,
+      "loss": 2.1914,
+      "step": 7447500
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 3.155663808009195e-05,
+      "loss": 2.1828,
+      "step": 7448000
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 3.155539949366587e-05,
+      "loss": 2.1949,
+      "step": 7448500
+    },
+    {
+      "epoch": 36.9,
+      "learning_rate": 3.1554160907239786e-05,
+      "loss": 2.1743,
+      "step": 7449000
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 3.1552927275159406e-05,
+      "loss": 2.1951,
+      "step": 7449500
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 3.1551691165906175e-05,
+      "loss": 2.2043,
+      "step": 7450000
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 3.155045257948009e-05,
+      "loss": 2.172,
+      "step": 7450500
+    },
+    {
+      "epoch": 36.91,
+      "learning_rate": 3.154921399305401e-05,
+      "loss": 2.1704,
+      "step": 7451000
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 3.1547975406627926e-05,
+      "loss": 2.1903,
+      "step": 7451500
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 3.154673682020184e-05,
+      "loss": 2.1893,
+      "step": 7452000
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 3.154549823377575e-05,
+      "loss": 2.1799,
+      "step": 7452500
+    },
+    {
+      "epoch": 36.92,
+      "learning_rate": 3.154425964734967e-05,
+      "loss": 2.1961,
+      "step": 7453000
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 3.154302106092359e-05,
+      "loss": 2.1879,
+      "step": 7453500
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 3.1541782474497504e-05,
+      "loss": 2.2118,
+      "step": 7454000
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 3.154054388807142e-05,
+      "loss": 2.1753,
+      "step": 7454500
+    },
+    {
+      "epoch": 36.93,
+      "learning_rate": 3.153930530164534e-05,
+      "loss": 2.1622,
+      "step": 7455000
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 3.1538066715219255e-05,
+      "loss": 2.1666,
+      "step": 7455500
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 3.153682812879317e-05,
+      "loss": 2.18,
+      "step": 7456000
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 3.153558954236709e-05,
+      "loss": 2.2071,
+      "step": 7456500
+    },
+    {
+      "epoch": 36.94,
+      "learning_rate": 3.153435343311386e-05,
+      "loss": 2.1606,
+      "step": 7457000
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 3.1533114846687774e-05,
+      "loss": 2.183,
+      "step": 7457500
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 3.153187626026169e-05,
+      "loss": 2.176,
+      "step": 7458000
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 3.153063767383561e-05,
+      "loss": 2.1672,
+      "step": 7458500
+    },
+    {
+      "epoch": 36.95,
+      "learning_rate": 3.1529399087409525e-05,
+      "loss": 2.1663,
+      "step": 7459000
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 3.152816050098344e-05,
+      "loss": 2.2065,
+      "step": 7459500
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 3.152692191455736e-05,
+      "loss": 2.1735,
+      "step": 7460000
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 3.1525683328131276e-05,
+      "loss": 2.1947,
+      "step": 7460500
+    },
+    {
+      "epoch": 36.96,
+      "learning_rate": 3.152444721887804e-05,
+      "loss": 2.206,
+      "step": 7461000
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 3.1523211109624807e-05,
+      "loss": 2.1445,
+      "step": 7461500
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 3.1521975000371575e-05,
+      "loss": 2.1961,
+      "step": 7462000
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 3.152073641394549e-05,
+      "loss": 2.1584,
+      "step": 7462500
+    },
+    {
+      "epoch": 36.97,
+      "learning_rate": 3.151949782751941e-05,
+      "loss": 2.161,
+      "step": 7463000
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 3.151826171826618e-05,
+      "loss": 2.1464,
+      "step": 7463500
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 3.1517023131840095e-05,
+      "loss": 2.1763,
+      "step": 7464000
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 3.151578454541401e-05,
+      "loss": 2.1944,
+      "step": 7464500
+    },
+    {
+      "epoch": 36.98,
+      "learning_rate": 3.151454843616078e-05,
+      "loss": 2.179,
+      "step": 7465000
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 3.15133098497347e-05,
+      "loss": 2.204,
+      "step": 7465500
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 3.1512071263308615e-05,
+      "loss": 2.179,
+      "step": 7466000
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 3.151083515405538e-05,
+      "loss": 2.1788,
+      "step": 7466500
+    },
+    {
+      "epoch": 36.99,
+      "learning_rate": 3.1509596567629294e-05,
+      "loss": 2.2008,
+      "step": 7467000
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.150835798120321e-05,
+      "loss": 2.168,
+      "step": 7467500
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.150711939477713e-05,
+      "loss": 2.1852,
+      "step": 7468000
+    },
+    {
+      "epoch": 37.0,
+      "eval_accuracy": 0.662816636573663,
+      "eval_accuracy_mlm": 0.6197658435400042,
+      "eval_accuracy_nsp": 0.8659980624335677,
+      "eval_loss": 2.305152416229248,
+      "eval_runtime": 145.847,
+      "eval_samples_per_second": 1748.126,
+      "eval_steps_per_second": 72.843,
+      "step": 7468191
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.1505880808351044e-05,
+      "loss": 2.1681,
+      "step": 7468500
+    },
+    {
+      "epoch": 37.0,
+      "learning_rate": 3.150464222192496e-05,
+      "loss": 2.1523,
+      "step": 7469000
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3.150340363549888e-05,
+      "loss": 2.1706,
+      "step": 7469500
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3.1502165049072795e-05,
+      "loss": 2.1522,
+      "step": 7470000
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3.150092646264671e-05,
+      "loss": 2.1346,
+      "step": 7470500
+    },
+    {
+      "epoch": 37.01,
+      "learning_rate": 3.149968787622063e-05,
+      "loss": 2.1607,
+      "step": 7471000
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 3.1498449289794546e-05,
+      "loss": 2.1727,
+      "step": 7471500
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 3.149721070336846e-05,
+      "loss": 2.1578,
+      "step": 7472000
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 3.149597459411523e-05,
+      "loss": 2.1535,
+      "step": 7472500
+    },
+    {
+      "epoch": 37.02,
+      "learning_rate": 3.149473600768915e-05,
+      "loss": 2.1529,
+      "step": 7473000
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 3.1493497421263066e-05,
+      "loss": 2.1526,
+      "step": 7473500
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 3.149225883483698e-05,
+      "loss": 2.1306,
+      "step": 7474000
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 3.1491022725583744e-05,
+      "loss": 2.1446,
+      "step": 7474500
+    },
+    {
+      "epoch": 37.03,
+      "learning_rate": 3.148978413915766e-05,
+      "loss": 2.1611,
+      "step": 7475000
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 3.148854555273158e-05,
+      "loss": 2.1508,
+      "step": 7475500
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 3.1487306966305495e-05,
+      "loss": 2.1516,
+      "step": 7476000
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 3.148606837987941e-05,
+      "loss": 2.1576,
+      "step": 7476500
+    },
+    {
+      "epoch": 37.04,
+      "learning_rate": 3.148482979345333e-05,
+      "loss": 2.1554,
+      "step": 7477000
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 3.1483591207027246e-05,
+      "loss": 2.1549,
+      "step": 7477500
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 3.148235262060116e-05,
+      "loss": 2.137,
+      "step": 7478000
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 3.148111403417508e-05,
+      "loss": 2.1573,
+      "step": 7478500
+    },
+    {
+      "epoch": 37.05,
+      "learning_rate": 3.1479875447749e-05,
+      "loss": 2.1439,
+      "step": 7479000
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 3.1478636861322914e-05,
+      "loss": 2.1432,
+      "step": 7479500
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 3.147739827489683e-05,
+      "loss": 2.1633,
+      "step": 7480000
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 3.14761621656436e-05,
+      "loss": 2.1335,
+      "step": 7480500
+    },
+    {
+      "epoch": 37.06,
+      "learning_rate": 3.1474923579217516e-05,
+      "loss": 2.1549,
+      "step": 7481000
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 3.147368499279143e-05,
+      "loss": 2.1481,
+      "step": 7481500
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 3.1472446406365344e-05,
+      "loss": 2.144,
+      "step": 7482000
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 3.147121029711211e-05,
+      "loss": 2.1578,
+      "step": 7482500
+    },
+    {
+      "epoch": 37.07,
+      "learning_rate": 3.146997418785888e-05,
+      "loss": 2.1529,
+      "step": 7483000
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 3.14687356014328e-05,
+      "loss": 2.1599,
+      "step": 7483500
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 3.1467497015006715e-05,
+      "loss": 2.1403,
+      "step": 7484000
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 3.146625842858063e-05,
+      "loss": 2.1674,
+      "step": 7484500
+    },
+    {
+      "epoch": 37.08,
+      "learning_rate": 3.146501984215455e-05,
+      "loss": 2.1574,
+      "step": 7485000
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 3.1463781255728466e-05,
+      "loss": 2.1732,
+      "step": 7485500
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 3.146254266930238e-05,
+      "loss": 2.1428,
+      "step": 7486000
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 3.14613040828763e-05,
+      "loss": 2.1599,
+      "step": 7486500
+    },
+    {
+      "epoch": 37.09,
+      "learning_rate": 3.1460065496450217e-05,
+      "loss": 2.138,
+      "step": 7487000
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 3.1458826910024133e-05,
+      "loss": 2.1664,
+      "step": 7487500
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 3.145758832359805e-05,
+      "loss": 2.1581,
+      "step": 7488000
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 3.145635221434481e-05,
+      "loss": 2.176,
+      "step": 7488500
+    },
+    {
+      "epoch": 37.1,
+      "learning_rate": 3.145511362791873e-05,
+      "loss": 2.1708,
+      "step": 7489000
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 3.1453875041492646e-05,
+      "loss": 2.1769,
+      "step": 7489500
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 3.145263645506656e-05,
+      "loss": 2.1438,
+      "step": 7490000
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 3.145139786864048e-05,
+      "loss": 2.1697,
+      "step": 7490500
+    },
+    {
+      "epoch": 37.11,
+      "learning_rate": 3.14501592822144e-05,
+      "loss": 2.1765,
+      "step": 7491000
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 3.1448920695788314e-05,
+      "loss": 2.1288,
+      "step": 7491500
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 3.144768210936223e-05,
+      "loss": 2.143,
+      "step": 7492000
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 3.1446446000109e-05,
+      "loss": 2.1439,
+      "step": 7492500
+    },
+    {
+      "epoch": 37.12,
+      "learning_rate": 3.144520741368292e-05,
+      "loss": 2.1486,
+      "step": 7493000
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 3.1443968827256834e-05,
+      "loss": 2.1771,
+      "step": 7493500
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 3.144273024083075e-05,
+      "loss": 2.1703,
+      "step": 7494000
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 3.144149165440467e-05,
+      "loss": 2.1512,
+      "step": 7494500
+    },
+    {
+      "epoch": 37.13,
+      "learning_rate": 3.1440253067978584e-05,
+      "loss": 2.1605,
+      "step": 7495000
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 3.1439014481552495e-05,
+      "loss": 2.1528,
+      "step": 7495500
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 3.143777589512641e-05,
+      "loss": 2.1338,
+      "step": 7496000
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 3.143653978587318e-05,
+      "loss": 2.1437,
+      "step": 7496500
+    },
+    {
+      "epoch": 37.14,
+      "learning_rate": 3.14353011994471e-05,
+      "loss": 2.1349,
+      "step": 7497000
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 3.1434062613021014e-05,
+      "loss": 2.1594,
+      "step": 7497500
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 3.143282650376778e-05,
+      "loss": 2.1769,
+      "step": 7498000
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 3.14315879173417e-05,
+      "loss": 2.1253,
+      "step": 7498500
+    },
+    {
+      "epoch": 37.15,
+      "learning_rate": 3.143034933091562e-05,
+      "loss": 2.1595,
+      "step": 7499000
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 3.1429110744489534e-05,
+      "loss": 2.1514,
+      "step": 7499500
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 3.142787215806345e-05,
+      "loss": 2.1622,
+      "step": 7500000
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 3.142663357163737e-05,
+      "loss": 2.1447,
+      "step": 7500500
+    },
+    {
+      "epoch": 37.16,
+      "learning_rate": 3.1425394985211285e-05,
+      "loss": 2.1597,
+      "step": 7501000
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 3.14241563987852e-05,
+      "loss": 2.1814,
+      "step": 7501500
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 3.142291781235912e-05,
+      "loss": 2.1547,
+      "step": 7502000
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 3.142168170310588e-05,
+      "loss": 2.1666,
+      "step": 7502500
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 3.14204431166798e-05,
+      "loss": 2.1768,
+      "step": 7503000
+    },
+    {
+      "epoch": 37.17,
+      "learning_rate": 3.1419204530253714e-05,
+      "loss": 2.1721,
+      "step": 7503500
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 3.141796842100048e-05,
+      "loss": 2.1622,
+      "step": 7504000
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 3.14167298345744e-05,
+      "loss": 2.1507,
+      "step": 7504500
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 3.141549372532117e-05,
+      "loss": 2.152,
+      "step": 7505000
+    },
+    {
+      "epoch": 37.18,
+      "learning_rate": 3.141425513889508e-05,
+      "loss": 2.1878,
+      "step": 7505500
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 3.1413016552468996e-05,
+      "loss": 2.1722,
+      "step": 7506000
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 3.141177796604291e-05,
+      "loss": 2.1489,
+      "step": 7506500
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 3.141053937961683e-05,
+      "loss": 2.1713,
+      "step": 7507000
+    },
+    {
+      "epoch": 37.19,
+      "learning_rate": 3.140930079319075e-05,
+      "loss": 2.1761,
+      "step": 7507500
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 3.1408062206764664e-05,
+      "loss": 2.1787,
+      "step": 7508000
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 3.140682362033858e-05,
+      "loss": 2.1452,
+      "step": 7508500
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 3.140558751108535e-05,
+      "loss": 2.1495,
+      "step": 7509000
+    },
+    {
+      "epoch": 37.2,
+      "learning_rate": 3.140435140183212e-05,
+      "loss": 2.165,
+      "step": 7509500
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 3.1403112815406035e-05,
+      "loss": 2.1504,
+      "step": 7510000
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 3.140187422897995e-05,
+      "loss": 2.1651,
+      "step": 7510500
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 3.140063564255387e-05,
+      "loss": 2.1685,
+      "step": 7511000
+    },
+    {
+      "epoch": 37.21,
+      "learning_rate": 3.1399397056127786e-05,
+      "loss": 2.199,
+      "step": 7511500
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 3.13981584697017e-05,
+      "loss": 2.1633,
+      "step": 7512000
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 3.139691988327561e-05,
+      "loss": 2.1725,
+      "step": 7512500
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 3.139568129684953e-05,
+      "loss": 2.1644,
+      "step": 7513000
+    },
+    {
+      "epoch": 37.22,
+      "learning_rate": 3.139444271042345e-05,
+      "loss": 2.1411,
+      "step": 7513500
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 3.1393204123997364e-05,
+      "loss": 2.1723,
+      "step": 7514000
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 3.139196553757128e-05,
+      "loss": 2.1631,
+      "step": 7514500
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 3.139072942831805e-05,
+      "loss": 2.1702,
+      "step": 7515000
+    },
+    {
+      "epoch": 37.23,
+      "learning_rate": 3.1389490841891966e-05,
+      "loss": 2.1552,
+      "step": 7515500
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 3.138825225546588e-05,
+      "loss": 2.1477,
+      "step": 7516000
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 3.13870136690398e-05,
+      "loss": 2.142,
+      "step": 7516500
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 3.138577508261372e-05,
+      "loss": 2.1627,
+      "step": 7517000
+    },
+    {
+      "epoch": 37.24,
+      "learning_rate": 3.1384536496187634e-05,
+      "loss": 2.1808,
+      "step": 7517500
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 3.138329790976155e-05,
+      "loss": 2.1587,
+      "step": 7518000
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 3.138205932333547e-05,
+      "loss": 2.163,
+      "step": 7518500
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 3.1380820736909385e-05,
+      "loss": 2.1469,
+      "step": 7519000
+    },
+    {
+      "epoch": 37.25,
+      "learning_rate": 3.13795821504833e-05,
+      "loss": 2.1555,
+      "step": 7519500
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 3.137834356405722e-05,
+      "loss": 2.1745,
+      "step": 7520000
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 3.1377104977631136e-05,
+      "loss": 2.1535,
+      "step": 7520500
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 3.1375866391205046e-05,
+      "loss": 2.177,
+      "step": 7521000
+    },
+    {
+      "epoch": 37.26,
+      "learning_rate": 3.137462780477896e-05,
+      "loss": 2.1823,
+      "step": 7521500
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 3.137338921835288e-05,
+      "loss": 2.1531,
+      "step": 7522000
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 3.1372150631926797e-05,
+      "loss": 2.154,
+      "step": 7522500
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 3.1370912045500714e-05,
+      "loss": 2.1678,
+      "step": 7523000
+    },
+    {
+      "epoch": 37.27,
+      "learning_rate": 3.136967593624748e-05,
+      "loss": 2.1881,
+      "step": 7523500
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 3.136843734982139e-05,
+      "loss": 2.1854,
+      "step": 7524000
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 3.136719876339531e-05,
+      "loss": 2.1702,
+      "step": 7524500
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 3.1365960176969226e-05,
+      "loss": 2.1338,
+      "step": 7525000
+    },
+    {
+      "epoch": 37.28,
+      "learning_rate": 3.1364724067716e-05,
+      "loss": 2.178,
+      "step": 7525500
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 3.1363487958462764e-05,
+      "loss": 2.1832,
+      "step": 7526000
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 3.136224937203668e-05,
+      "loss": 2.1513,
+      "step": 7526500
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 3.13610107856106e-05,
+      "loss": 2.1825,
+      "step": 7527000
+    },
+    {
+      "epoch": 37.29,
+      "learning_rate": 3.1359772199184515e-05,
+      "loss": 2.1724,
+      "step": 7527500
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 3.135853361275843e-05,
+      "loss": 2.1605,
+      "step": 7528000
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 3.135729502633235e-05,
+      "loss": 2.1593,
+      "step": 7528500
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 3.1356056439906266e-05,
+      "loss": 2.1526,
+      "step": 7529000
+    },
+    {
+      "epoch": 37.3,
+      "learning_rate": 3.135481785348018e-05,
+      "loss": 2.1773,
+      "step": 7529500
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 3.13535792670541e-05,
+      "loss": 2.1671,
+      "step": 7530000
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 3.135234563497372e-05,
+      "loss": 2.1669,
+      "step": 7530500
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 3.135110704854764e-05,
+      "loss": 2.1934,
+      "step": 7531000
+    },
+    {
+      "epoch": 37.31,
+      "learning_rate": 3.1349868462121554e-05,
+      "loss": 2.1723,
+      "step": 7531500
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 3.134862987569547e-05,
+      "loss": 2.1526,
+      "step": 7532000
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 3.134739376644223e-05,
+      "loss": 2.1498,
+      "step": 7532500
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 3.134615518001615e-05,
+      "loss": 2.1688,
+      "step": 7533000
+    },
+    {
+      "epoch": 37.32,
+      "learning_rate": 3.134491659359007e-05,
+      "loss": 2.1782,
+      "step": 7533500
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 3.1343678007163984e-05,
+      "loss": 2.1479,
+      "step": 7534000
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 3.13424394207379e-05,
+      "loss": 2.1636,
+      "step": 7534500
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 3.134120083431182e-05,
+      "loss": 2.1481,
+      "step": 7535000
+    },
+    {
+      "epoch": 37.33,
+      "learning_rate": 3.1339962247885734e-05,
+      "loss": 2.1379,
+      "step": 7535500
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3.133872366145965e-05,
+      "loss": 2.1667,
+      "step": 7536000
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3.133748755220642e-05,
+      "loss": 2.1851,
+      "step": 7536500
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3.133624896578034e-05,
+      "loss": 2.1634,
+      "step": 7537000
+    },
+    {
+      "epoch": 37.34,
+      "learning_rate": 3.1335010379354254e-05,
+      "loss": 2.1672,
+      "step": 7537500
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 3.1333774270101016e-05,
+      "loss": 2.1684,
+      "step": 7538000
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 3.133253568367493e-05,
+      "loss": 2.1728,
+      "step": 7538500
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 3.133129709724885e-05,
+      "loss": 2.1735,
+      "step": 7539000
+    },
+    {
+      "epoch": 37.35,
+      "learning_rate": 3.133005851082277e-05,
+      "loss": 2.1688,
+      "step": 7539500
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 3.1328819924396684e-05,
+      "loss": 2.1657,
+      "step": 7540000
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 3.13275813379706e-05,
+      "loss": 2.1685,
+      "step": 7540500
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 3.132634275154452e-05,
+      "loss": 2.168,
+      "step": 7541000
+    },
+    {
+      "epoch": 37.36,
+      "learning_rate": 3.1325104165118435e-05,
+      "loss": 2.1783,
+      "step": 7541500
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 3.132386557869235e-05,
+      "loss": 2.1726,
+      "step": 7542000
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 3.132262946943912e-05,
+      "loss": 2.1833,
+      "step": 7542500
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 3.132139088301304e-05,
+      "loss": 2.1462,
+      "step": 7543000
+    },
+    {
+      "epoch": 37.37,
+      "learning_rate": 3.13201547737598e-05,
+      "loss": 2.1548,
+      "step": 7543500
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 3.1318916187333716e-05,
+      "loss": 2.1871,
+      "step": 7544000
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 3.131767760090763e-05,
+      "loss": 2.151,
+      "step": 7544500
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 3.131643901448155e-05,
+      "loss": 2.1504,
+      "step": 7545000
+    },
+    {
+      "epoch": 37.38,
+      "learning_rate": 3.131520042805547e-05,
+      "loss": 2.1335,
+      "step": 7545500
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 3.1313961841629384e-05,
+      "loss": 2.1584,
+      "step": 7546000
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 3.13127232552033e-05,
+      "loss": 2.1635,
+      "step": 7546500
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 3.131148714595007e-05,
+      "loss": 2.1373,
+      "step": 7547000
+    },
+    {
+      "epoch": 37.39,
+      "learning_rate": 3.1310248559523987e-05,
+      "loss": 2.1691,
+      "step": 7547500
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 3.1309009973097904e-05,
+      "loss": 2.1757,
+      "step": 7548000
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 3.130777386384467e-05,
+      "loss": 2.1534,
+      "step": 7548500
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 3.130653527741859e-05,
+      "loss": 2.1707,
+      "step": 7549000
+    },
+    {
+      "epoch": 37.4,
+      "learning_rate": 3.1305296690992506e-05,
+      "loss": 2.1685,
+      "step": 7549500
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 3.1304058104566416e-05,
+      "loss": 2.1556,
+      "step": 7550000
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 3.130281951814033e-05,
+      "loss": 2.1833,
+      "step": 7550500
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 3.130158093171425e-05,
+      "loss": 2.1858,
+      "step": 7551000
+    },
+    {
+      "epoch": 37.41,
+      "learning_rate": 3.130034234528817e-05,
+      "loss": 2.1862,
+      "step": 7551500
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 3.129910623603494e-05,
+      "loss": 2.1598,
+      "step": 7552000
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 3.129786764960886e-05,
+      "loss": 2.1699,
+      "step": 7552500
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 3.129663154035563e-05,
+      "loss": 2.1736,
+      "step": 7553000
+    },
+    {
+      "epoch": 37.42,
+      "learning_rate": 3.1295392953929545e-05,
+      "loss": 2.162,
+      "step": 7553500
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3.1294154367503456e-05,
+      "loss": 2.1583,
+      "step": 7554000
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3.129291578107737e-05,
+      "loss": 2.1947,
+      "step": 7554500
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3.129167719465129e-05,
+      "loss": 2.1973,
+      "step": 7555000
+    },
+    {
+      "epoch": 37.43,
+      "learning_rate": 3.129044108539806e-05,
+      "loss": 2.1579,
+      "step": 7555500
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 3.1289202498971975e-05,
+      "loss": 2.1586,
+      "step": 7556000
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 3.128796391254589e-05,
+      "loss": 2.1745,
+      "step": 7556500
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 3.128672532611981e-05,
+      "loss": 2.1567,
+      "step": 7557000
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 3.1285486739693726e-05,
+      "loss": 2.1785,
+      "step": 7557500
+    },
+    {
+      "epoch": 37.44,
+      "learning_rate": 3.1284250630440495e-05,
+      "loss": 2.1605,
+      "step": 7558000
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 3.128301204401441e-05,
+      "loss": 2.17,
+      "step": 7558500
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 3.128177345758833e-05,
+      "loss": 2.1761,
+      "step": 7559000
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 3.1280534871162246e-05,
+      "loss": 2.1689,
+      "step": 7559500
+    },
+    {
+      "epoch": 37.45,
+      "learning_rate": 3.127929628473616e-05,
+      "loss": 2.1641,
+      "step": 7560000
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 3.1278060175482924e-05,
+      "loss": 2.1337,
+      "step": 7560500
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 3.127682158905684e-05,
+      "loss": 2.1564,
+      "step": 7561000
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 3.127558300263076e-05,
+      "loss": 2.1722,
+      "step": 7561500
+    },
+    {
+      "epoch": 37.46,
+      "learning_rate": 3.1274344416204675e-05,
+      "loss": 2.1591,
+      "step": 7562000
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 3.127310582977859e-05,
+      "loss": 2.1634,
+      "step": 7562500
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 3.127186724335251e-05,
+      "loss": 2.1877,
+      "step": 7563000
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 3.1270628656926426e-05,
+      "loss": 2.159,
+      "step": 7563500
+    },
+    {
+      "epoch": 37.47,
+      "learning_rate": 3.126939007050034e-05,
+      "loss": 2.1643,
+      "step": 7564000
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 3.126815148407426e-05,
+      "loss": 2.1624,
+      "step": 7564500
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 3.126691289764818e-05,
+      "loss": 2.1848,
+      "step": 7565000
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 3.126567431122209e-05,
+      "loss": 2.1678,
+      "step": 7565500
+    },
+    {
+      "epoch": 37.48,
+      "learning_rate": 3.1264435724796004e-05,
+      "loss": 2.1883,
+      "step": 7566000
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 3.126319713836992e-05,
+      "loss": 2.174,
+      "step": 7566500
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 3.126195855194384e-05,
+      "loss": 2.1423,
+      "step": 7567000
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 3.1260719965517755e-05,
+      "loss": 2.1437,
+      "step": 7567500
+    },
+    {
+      "epoch": 37.49,
+      "learning_rate": 3.1259483856264523e-05,
+      "loss": 2.1632,
+      "step": 7568000
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3.125824526983844e-05,
+      "loss": 2.1706,
+      "step": 7568500
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3.125700668341236e-05,
+      "loss": 2.139,
+      "step": 7569000
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3.1255770574159126e-05,
+      "loss": 2.1831,
+      "step": 7569500
+    },
+    {
+      "epoch": 37.5,
+      "learning_rate": 3.125453198773304e-05,
+      "loss": 2.1732,
+      "step": 7570000
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 3.125329340130696e-05,
+      "loss": 2.1728,
+      "step": 7570500
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 3.125205481488088e-05,
+      "loss": 2.1557,
+      "step": 7571000
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 3.125081622845479e-05,
+      "loss": 2.1746,
+      "step": 7571500
+    },
+    {
+      "epoch": 37.51,
+      "learning_rate": 3.1249577642028704e-05,
+      "loss": 2.1592,
+      "step": 7572000
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3.124833905560262e-05,
+      "loss": 2.1405,
+      "step": 7572500
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3.124710542352224e-05,
+      "loss": 2.161,
+      "step": 7573000
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3.124586683709616e-05,
+      "loss": 2.1546,
+      "step": 7573500
+    },
+    {
+      "epoch": 37.52,
+      "learning_rate": 3.1244628250670076e-05,
+      "loss": 2.161,
+      "step": 7574000
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 3.124338966424399e-05,
+      "loss": 2.1624,
+      "step": 7574500
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 3.124215603216361e-05,
+      "loss": 2.1927,
+      "step": 7575000
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 3.124091744573753e-05,
+      "loss": 2.1766,
+      "step": 7575500
+    },
+    {
+      "epoch": 37.53,
+      "learning_rate": 3.123967885931145e-05,
+      "loss": 2.1885,
+      "step": 7576000
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 3.1238440272885364e-05,
+      "loss": 2.1728,
+      "step": 7576500
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 3.123720168645928e-05,
+      "loss": 2.1817,
+      "step": 7577000
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 3.123596310003319e-05,
+      "loss": 2.2075,
+      "step": 7577500
+    },
+    {
+      "epoch": 37.54,
+      "learning_rate": 3.123472451360711e-05,
+      "loss": 2.1638,
+      "step": 7578000
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 3.1233485927181025e-05,
+      "loss": 2.1478,
+      "step": 7578500
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 3.123224734075494e-05,
+      "loss": 2.1594,
+      "step": 7579000
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 3.123100875432886e-05,
+      "loss": 2.1663,
+      "step": 7579500
+    },
+    {
+      "epoch": 37.55,
+      "learning_rate": 3.1229770167902776e-05,
+      "loss": 2.1883,
+      "step": 7580000
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 3.122853158147669e-05,
+      "loss": 2.1795,
+      "step": 7580500
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 3.122729299505061e-05,
+      "loss": 2.1715,
+      "step": 7581000
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 3.1226054408624526e-05,
+      "loss": 2.1624,
+      "step": 7581500
+    },
+    {
+      "epoch": 37.56,
+      "learning_rate": 3.122481582219844e-05,
+      "loss": 2.1937,
+      "step": 7582000
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 3.122357723577236e-05,
+      "loss": 2.1976,
+      "step": 7582500
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 3.122233864934628e-05,
+      "loss": 2.1896,
+      "step": 7583000
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 3.1221100062920194e-05,
+      "loss": 2.1723,
+      "step": 7583500
+    },
+    {
+      "epoch": 37.57,
+      "learning_rate": 3.1219861476494104e-05,
+      "loss": 2.171,
+      "step": 7584000
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 3.121862289006802e-05,
+      "loss": 2.1689,
+      "step": 7584500
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 3.121738430364194e-05,
+      "loss": 2.1642,
+      "step": 7585000
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 3.1216148194388714e-05,
+      "loss": 2.1587,
+      "step": 7585500
+    },
+    {
+      "epoch": 37.58,
+      "learning_rate": 3.121490960796263e-05,
+      "loss": 2.1621,
+      "step": 7586000
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 3.121367102153655e-05,
+      "loss": 2.2021,
+      "step": 7586500
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 3.1212432435110465e-05,
+      "loss": 2.1851,
+      "step": 7587000
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 3.1211193848684375e-05,
+      "loss": 2.1986,
+      "step": 7587500
+    },
+    {
+      "epoch": 37.59,
+      "learning_rate": 3.120995526225829e-05,
+      "loss": 2.1908,
+      "step": 7588000
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 3.120871667583221e-05,
+      "loss": 2.1592,
+      "step": 7588500
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 3.1207478089406125e-05,
+      "loss": 2.1781,
+      "step": 7589000
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 3.1206241980152894e-05,
+      "loss": 2.1659,
+      "step": 7589500
+    },
+    {
+      "epoch": 37.6,
+      "learning_rate": 3.120500339372681e-05,
+      "loss": 2.1833,
+      "step": 7590000
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 3.120376480730072e-05,
+      "loss": 2.1932,
+      "step": 7590500
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 3.12025286980475e-05,
+      "loss": 2.1681,
+      "step": 7591000
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 3.1201290111621414e-05,
+      "loss": 2.1914,
+      "step": 7591500
+    },
+    {
+      "epoch": 37.61,
+      "learning_rate": 3.120005152519533e-05,
+      "loss": 2.1588,
+      "step": 7592000
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 3.119881541594209e-05,
+      "loss": 2.1643,
+      "step": 7592500
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 3.119757682951601e-05,
+      "loss": 2.1474,
+      "step": 7593000
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 3.119633824308993e-05,
+      "loss": 2.1326,
+      "step": 7593500
+    },
+    {
+      "epoch": 37.62,
+      "learning_rate": 3.1195099656663844e-05,
+      "loss": 2.1581,
+      "step": 7594000
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 3.119386107023776e-05,
+      "loss": 2.1565,
+      "step": 7594500
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 3.119262496098453e-05,
+      "loss": 2.158,
+      "step": 7595000
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 3.1191386374558446e-05,
+      "loss": 2.1375,
+      "step": 7595500
+    },
+    {
+      "epoch": 37.63,
+      "learning_rate": 3.119014778813236e-05,
+      "loss": 2.1917,
+      "step": 7596000
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 3.118890920170628e-05,
+      "loss": 2.1699,
+      "step": 7596500
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 3.11876706152802e-05,
+      "loss": 2.178,
+      "step": 7597000
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 3.1186434506026966e-05,
+      "loss": 2.1686,
+      "step": 7597500
+    },
+    {
+      "epoch": 37.64,
+      "learning_rate": 3.1185195919600876e-05,
+      "loss": 2.1555,
+      "step": 7598000
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 3.118395733317479e-05,
+      "loss": 2.1557,
+      "step": 7598500
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 3.118271874674871e-05,
+      "loss": 2.175,
+      "step": 7599000
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 3.118148016032263e-05,
+      "loss": 2.1777,
+      "step": 7599500
+    },
+    {
+      "epoch": 37.65,
+      "learning_rate": 3.1180244051069396e-05,
+      "loss": 2.1642,
+      "step": 7600000
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 3.117900546464331e-05,
+      "loss": 2.1832,
+      "step": 7600500
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 3.117776687821723e-05,
+      "loss": 2.1524,
+      "step": 7601000
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 3.1176528291791146e-05,
+      "loss": 2.175,
+      "step": 7601500
+    },
+    {
+      "epoch": 37.66,
+      "learning_rate": 3.117528970536506e-05,
+      "loss": 2.158,
+      "step": 7602000
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 3.117405111893898e-05,
+      "loss": 2.1643,
+      "step": 7602500
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 3.117281500968575e-05,
+      "loss": 2.198,
+      "step": 7603000
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 3.1171576423259666e-05,
+      "loss": 2.1969,
+      "step": 7603500
+    },
+    {
+      "epoch": 37.67,
+      "learning_rate": 3.117033783683358e-05,
+      "loss": 2.1576,
+      "step": 7604000
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 3.116909925040749e-05,
+      "loss": 2.1697,
+      "step": 7604500
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 3.116786066398141e-05,
+      "loss": 2.1615,
+      "step": 7605000
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 3.116662207755533e-05,
+      "loss": 2.1891,
+      "step": 7605500
+    },
+    {
+      "epoch": 37.68,
+      "learning_rate": 3.1165383491129244e-05,
+      "loss": 2.1839,
+      "step": 7606000
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 3.116414738187601e-05,
+      "loss": 2.1447,
+      "step": 7606500
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 3.116290879544993e-05,
+      "loss": 2.178,
+      "step": 7607000
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 3.1161670209023847e-05,
+      "loss": 2.1625,
+      "step": 7607500
+    },
+    {
+      "epoch": 37.69,
+      "learning_rate": 3.1160431622597763e-05,
+      "loss": 2.1604,
+      "step": 7608000
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 3.115919303617168e-05,
+      "loss": 2.1825,
+      "step": 7608500
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 3.11579544497456e-05,
+      "loss": 2.1545,
+      "step": 7609000
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 3.1156718340492366e-05,
+      "loss": 2.1914,
+      "step": 7609500
+    },
+    {
+      "epoch": 37.7,
+      "learning_rate": 3.115548223123913e-05,
+      "loss": 2.1585,
+      "step": 7610000
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 3.1154243644813045e-05,
+      "loss": 2.1652,
+      "step": 7610500
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 3.115300505838696e-05,
+      "loss": 2.1715,
+      "step": 7611000
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 3.115176647196088e-05,
+      "loss": 2.1806,
+      "step": 7611500
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 3.1150527885534796e-05,
+      "loss": 2.1622,
+      "step": 7612000
+    },
+    {
+      "epoch": 37.71,
+      "learning_rate": 3.114928929910871e-05,
+      "loss": 2.1724,
+      "step": 7612500
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 3.114805071268263e-05,
+      "loss": 2.1598,
+      "step": 7613000
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 3.114681212625655e-05,
+      "loss": 2.1558,
+      "step": 7613500
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 3.1145573539830464e-05,
+      "loss": 2.1799,
+      "step": 7614000
+    },
+    {
+      "epoch": 37.72,
+      "learning_rate": 3.114433495340438e-05,
+      "loss": 2.1604,
+      "step": 7614500
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 3.11430963669783e-05,
+      "loss": 2.1678,
+      "step": 7615000
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 3.1141857780552214e-05,
+      "loss": 2.1516,
+      "step": 7615500
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 3.114062167129898e-05,
+      "loss": 2.1665,
+      "step": 7616000
+    },
+    {
+      "epoch": 37.73,
+      "learning_rate": 3.11393830848729e-05,
+      "loss": 2.1556,
+      "step": 7616500
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 3.113814449844682e-05,
+      "loss": 2.1651,
+      "step": 7617000
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 3.1136905912020734e-05,
+      "loss": 2.1926,
+      "step": 7617500
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 3.1135669802767496e-05,
+      "loss": 2.1944,
+      "step": 7618000
+    },
+    {
+      "epoch": 37.74,
+      "learning_rate": 3.113443121634141e-05,
+      "loss": 2.1642,
+      "step": 7618500
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 3.113319262991533e-05,
+      "loss": 2.17,
+      "step": 7619000
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 3.113195404348925e-05,
+      "loss": 2.168,
+      "step": 7619500
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 3.1130715457063164e-05,
+      "loss": 2.1791,
+      "step": 7620000
+    },
+    {
+      "epoch": 37.75,
+      "learning_rate": 3.112947687063708e-05,
+      "loss": 2.1555,
+      "step": 7620500
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 3.1128238284211e-05,
+      "loss": 2.1896,
+      "step": 7621000
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 3.1126999697784914e-05,
+      "loss": 2.184,
+      "step": 7621500
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 3.112576111135883e-05,
+      "loss": 2.1875,
+      "step": 7622000
+    },
+    {
+      "epoch": 37.76,
+      "learning_rate": 3.112452252493275e-05,
+      "loss": 2.1642,
+      "step": 7622500
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 3.1123283938506665e-05,
+      "loss": 2.1688,
+      "step": 7623000
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 3.1122047829253434e-05,
+      "loss": 2.1603,
+      "step": 7623500
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 3.112080924282735e-05,
+      "loss": 2.1725,
+      "step": 7624000
+    },
+    {
+      "epoch": 37.77,
+      "learning_rate": 3.111957065640127e-05,
+      "loss": 2.1852,
+      "step": 7624500
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 3.111833206997518e-05,
+      "loss": 2.1877,
+      "step": 7625000
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 3.111709596072195e-05,
+      "loss": 2.1658,
+      "step": 7625500
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 3.1115857374295864e-05,
+      "loss": 2.191,
+      "step": 7626000
+    },
+    {
+      "epoch": 37.78,
+      "learning_rate": 3.111461878786978e-05,
+      "loss": 2.1775,
+      "step": 7626500
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 3.11133802014437e-05,
+      "loss": 2.1898,
+      "step": 7627000
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 3.1112141615017615e-05,
+      "loss": 2.1753,
+      "step": 7627500
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 3.111090302859153e-05,
+      "loss": 2.1766,
+      "step": 7628000
+    },
+    {
+      "epoch": 37.79,
+      "learning_rate": 3.110966444216545e-05,
+      "loss": 2.153,
+      "step": 7628500
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 3.110842833291222e-05,
+      "loss": 2.1655,
+      "step": 7629000
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 3.110719222365898e-05,
+      "loss": 2.1837,
+      "step": 7629500
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 3.1105956114405755e-05,
+      "loss": 2.1567,
+      "step": 7630000
+    },
+    {
+      "epoch": 37.8,
+      "learning_rate": 3.110471752797967e-05,
+      "loss": 2.1809,
+      "step": 7630500
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 3.110347894155359e-05,
+      "loss": 2.1559,
+      "step": 7631000
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 3.1102240355127506e-05,
+      "loss": 2.1647,
+      "step": 7631500
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 3.1101001768701416e-05,
+      "loss": 2.1759,
+      "step": 7632000
+    },
+    {
+      "epoch": 37.81,
+      "learning_rate": 3.109976318227533e-05,
+      "loss": 2.1911,
+      "step": 7632500
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 3.109852459584925e-05,
+      "loss": 2.1902,
+      "step": 7633000
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 3.1097286009423167e-05,
+      "loss": 2.1445,
+      "step": 7633500
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 3.1096047422997084e-05,
+      "loss": 2.1595,
+      "step": 7634000
+    },
+    {
+      "epoch": 37.82,
+      "learning_rate": 3.1094808836571e-05,
+      "loss": 2.1681,
+      "step": 7634500
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 3.109357520449062e-05,
+      "loss": 2.1935,
+      "step": 7635000
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 3.109233661806454e-05,
+      "loss": 2.1732,
+      "step": 7635500
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 3.1091098031638455e-05,
+      "loss": 2.1912,
+      "step": 7636000
+    },
+    {
+      "epoch": 37.83,
+      "learning_rate": 3.108985944521237e-05,
+      "loss": 2.165,
+      "step": 7636500
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 3.108862085878629e-05,
+      "loss": 2.1409,
+      "step": 7637000
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 3.1087382272360206e-05,
+      "loss": 2.1897,
+      "step": 7637500
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 3.1086143685934116e-05,
+      "loss": 2.1517,
+      "step": 7638000
+    },
+    {
+      "epoch": 37.84,
+      "learning_rate": 3.108490509950803e-05,
+      "loss": 2.1736,
+      "step": 7638500
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 3.108366651308195e-05,
+      "loss": 2.1654,
+      "step": 7639000
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 3.108242792665587e-05,
+      "loss": 2.1516,
+      "step": 7639500
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 3.1081189340229784e-05,
+      "loss": 2.1426,
+      "step": 7640000
+    },
+    {
+      "epoch": 37.85,
+      "learning_rate": 3.107995323097655e-05,
+      "loss": 2.2071,
+      "step": 7640500
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 3.107871464455047e-05,
+      "loss": 2.1758,
+      "step": 7641000
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 3.107747605812438e-05,
+      "loss": 2.1858,
+      "step": 7641500
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 3.1076237471698296e-05,
+      "loss": 2.1854,
+      "step": 7642000
+    },
+    {
+      "epoch": 37.86,
+      "learning_rate": 3.1074998885272213e-05,
+      "loss": 2.1974,
+      "step": 7642500
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 3.107376029884613e-05,
+      "loss": 2.1805,
+      "step": 7643000
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 3.107252171242005e-05,
+      "loss": 2.1728,
+      "step": 7643500
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 3.1071283125993964e-05,
+      "loss": 2.1834,
+      "step": 7644000
+    },
+    {
+      "epoch": 37.87,
+      "learning_rate": 3.107004453956788e-05,
+      "loss": 2.1629,
+      "step": 7644500
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 3.106880843031465e-05,
+      "loss": 2.179,
+      "step": 7645000
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 3.106756984388857e-05,
+      "loss": 2.1791,
+      "step": 7645500
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 3.1066331257462484e-05,
+      "loss": 2.1528,
+      "step": 7646000
+    },
+    {
+      "epoch": 37.88,
+      "learning_rate": 3.10650926710364e-05,
+      "loss": 2.1654,
+      "step": 7646500
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 3.106385408461032e-05,
+      "loss": 2.1706,
+      "step": 7647000
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 3.1062615498184235e-05,
+      "loss": 2.1603,
+      "step": 7647500
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 3.106137691175815e-05,
+      "loss": 2.1772,
+      "step": 7648000
+    },
+    {
+      "epoch": 37.89,
+      "learning_rate": 3.106013832533207e-05,
+      "loss": 2.1699,
+      "step": 7648500
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 3.1058899738905985e-05,
+      "loss": 2.187,
+      "step": 7649000
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 3.105766362965275e-05,
+      "loss": 2.1842,
+      "step": 7649500
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 3.105642752039952e-05,
+      "loss": 2.186,
+      "step": 7650000
+    },
+    {
+      "epoch": 37.9,
+      "learning_rate": 3.105518893397343e-05,
+      "loss": 2.2046,
+      "step": 7650500
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 3.105395034754735e-05,
+      "loss": 2.1892,
+      "step": 7651000
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 3.105271176112127e-05,
+      "loss": 2.1576,
+      "step": 7651500
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 3.1051473174695184e-05,
+      "loss": 2.1767,
+      "step": 7652000
+    },
+    {
+      "epoch": 37.91,
+      "learning_rate": 3.10502345882691e-05,
+      "loss": 2.1738,
+      "step": 7652500
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 3.104899600184302e-05,
+      "loss": 2.1886,
+      "step": 7653000
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 3.1047757415416935e-05,
+      "loss": 2.1829,
+      "step": 7653500
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 3.104651882899085e-05,
+      "loss": 2.145,
+      "step": 7654000
+    },
+    {
+      "epoch": 37.92,
+      "learning_rate": 3.104528024256477e-05,
+      "loss": 2.1616,
+      "step": 7654500
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 3.1044041656138685e-05,
+      "loss": 2.2104,
+      "step": 7655000
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 3.10428030697126e-05,
+      "loss": 2.167,
+      "step": 7655500
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 3.1041566960459364e-05,
+      "loss": 2.1776,
+      "step": 7656000
+    },
+    {
+      "epoch": 37.93,
+      "learning_rate": 3.104032837403328e-05,
+      "loss": 2.1884,
+      "step": 7656500
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 3.103909226478005e-05,
+      "loss": 2.1785,
+      "step": 7657000
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 3.103785367835397e-05,
+      "loss": 2.1838,
+      "step": 7657500
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 3.1036615091927884e-05,
+      "loss": 2.1869,
+      "step": 7658000
+    },
+    {
+      "epoch": 37.94,
+      "learning_rate": 3.10353765055018e-05,
+      "loss": 2.159,
+      "step": 7658500
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 3.103413791907572e-05,
+      "loss": 2.1483,
+      "step": 7659000
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 3.1032899332649635e-05,
+      "loss": 2.1951,
+      "step": 7659500
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 3.103166074622355e-05,
+      "loss": 2.1653,
+      "step": 7660000
+    },
+    {
+      "epoch": 37.95,
+      "learning_rate": 3.103042215979747e-05,
+      "loss": 2.163,
+      "step": 7660500
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 3.1029183573371386e-05,
+      "loss": 2.1676,
+      "step": 7661000
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 3.10279449869453e-05,
+      "loss": 2.1827,
+      "step": 7661500
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 3.1026708877692065e-05,
+      "loss": 2.1703,
+      "step": 7662000
+    },
+    {
+      "epoch": 37.96,
+      "learning_rate": 3.102547276843884e-05,
+      "loss": 2.2001,
+      "step": 7662500
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 3.102423418201275e-05,
+      "loss": 2.1585,
+      "step": 7663000
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 3.102299559558667e-05,
+      "loss": 2.1608,
+      "step": 7663500
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 3.1021757009160584e-05,
+      "loss": 2.182,
+      "step": 7664000
+    },
+    {
+      "epoch": 37.97,
+      "learning_rate": 3.10205184227345e-05,
+      "loss": 2.1963,
+      "step": 7664500
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 3.101927983630842e-05,
+      "loss": 2.1576,
+      "step": 7665000
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 3.1018041249882335e-05,
+      "loss": 2.1685,
+      "step": 7665500
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 3.101680266345625e-05,
+      "loss": 2.1808,
+      "step": 7666000
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 3.101556407703017e-05,
+      "loss": 2.1778,
+      "step": 7666500
+    },
+    {
+      "epoch": 37.98,
+      "learning_rate": 3.101432796777694e-05,
+      "loss": 2.1627,
+      "step": 7667000
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 3.1013089381350855e-05,
+      "loss": 2.1722,
+      "step": 7667500
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 3.101185079492477e-05,
+      "loss": 2.1858,
+      "step": 7668000
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 3.101061220849868e-05,
+      "loss": 2.2135,
+      "step": 7668500
+    },
+    {
+      "epoch": 37.99,
+      "learning_rate": 3.100937609924546e-05,
+      "loss": 2.1998,
+      "step": 7669000
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.100813751281937e-05,
+      "loss": 2.1548,
+      "step": 7669500
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.1006898926393284e-05,
+      "loss": 2.1506,
+      "step": 7670000
+    },
+    {
+      "epoch": 38.0,
+      "eval_accuracy": 0.6640129221845454,
+      "eval_accuracy_mlm": 0.6211758980997927,
+      "eval_accuracy_nsp": 0.8663432159680576,
+      "eval_loss": 2.2940046787261963,
+      "eval_runtime": 146.024,
+      "eval_samples_per_second": 1746.007,
+      "eval_steps_per_second": 72.755,
+      "step": 7670034
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.10056603399672e-05,
+      "loss": 2.1341,
+      "step": 7670500
+    },
+    {
+      "epoch": 38.0,
+      "learning_rate": 3.100442175354112e-05,
+      "loss": 2.1179,
+      "step": 7671000
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.1003183167115035e-05,
+      "loss": 2.1475,
+      "step": 7671500
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.100194458068895e-05,
+      "loss": 2.1367,
+      "step": 7672000
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.100070599426287e-05,
+      "loss": 2.152,
+      "step": 7672500
+    },
+    {
+      "epoch": 38.01,
+      "learning_rate": 3.0999467407836786e-05,
+      "loss": 2.1322,
+      "step": 7673000
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 3.09982288214107e-05,
+      "loss": 2.1336,
+      "step": 7673500
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 3.099699023498462e-05,
+      "loss": 2.1332,
+      "step": 7674000
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 3.099575164855854e-05,
+      "loss": 2.1682,
+      "step": 7674500
+    },
+    {
+      "epoch": 38.02,
+      "learning_rate": 3.0994515539305305e-05,
+      "loss": 2.1566,
+      "step": 7675000
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 3.099327943005207e-05,
+      "loss": 2.1353,
+      "step": 7675500
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 3.099204332079884e-05,
+      "loss": 2.1698,
+      "step": 7676000
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 3.099080473437276e-05,
+      "loss": 2.1583,
+      "step": 7676500
+    },
+    {
+      "epoch": 38.03,
+      "learning_rate": 3.098956614794668e-05,
+      "loss": 2.1478,
+      "step": 7677000
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 3.0988327561520594e-05,
+      "loss": 2.1403,
+      "step": 7677500
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 3.098708897509451e-05,
+      "loss": 2.1576,
+      "step": 7678000
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 3.098585038866842e-05,
+      "loss": 2.1442,
+      "step": 7678500
+    },
+    {
+      "epoch": 38.04,
+      "learning_rate": 3.098461180224234e-05,
+      "loss": 2.1502,
+      "step": 7679000
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 3.0983373215816255e-05,
+      "loss": 2.1695,
+      "step": 7679500
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 3.098213462939017e-05,
+      "loss": 2.1417,
+      "step": 7680000
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 3.098089604296409e-05,
+      "loss": 2.1439,
+      "step": 7680500
+    },
+    {
+      "epoch": 38.05,
+      "learning_rate": 3.0979657456538006e-05,
+      "loss": 2.1506,
+      "step": 7681000
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 3.097841887011192e-05,
+      "loss": 2.107,
+      "step": 7681500
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 3.0977182760858685e-05,
+      "loss": 2.1619,
+      "step": 7682000
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 3.097594665160546e-05,
+      "loss": 2.1356,
+      "step": 7682500
+    },
+    {
+      "epoch": 38.06,
+      "learning_rate": 3.097470806517938e-05,
+      "loss": 2.1428,
+      "step": 7683000
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 3.0973469478753294e-05,
+      "loss": 2.1517,
+      "step": 7683500
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 3.097223089232721e-05,
+      "loss": 2.1489,
+      "step": 7684000
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 3.097099478307397e-05,
+      "loss": 2.1292,
+      "step": 7684500
+    },
+    {
+      "epoch": 38.07,
+      "learning_rate": 3.096975619664789e-05,
+      "loss": 2.1504,
+      "step": 7685000
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 3.096851761022181e-05,
+      "loss": 2.1655,
+      "step": 7685500
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 3.0967281500968576e-05,
+      "loss": 2.1519,
+      "step": 7686000
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 3.096604291454249e-05,
+      "loss": 2.1419,
+      "step": 7686500
+    },
+    {
+      "epoch": 38.08,
+      "learning_rate": 3.096480432811641e-05,
+      "loss": 2.1692,
+      "step": 7687000
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 3.0963565741690326e-05,
+      "loss": 2.1659,
+      "step": 7687500
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 3.096232715526424e-05,
+      "loss": 2.1339,
+      "step": 7688000
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 3.096108856883816e-05,
+      "loss": 2.1517,
+      "step": 7688500
+    },
+    {
+      "epoch": 38.09,
+      "learning_rate": 3.095985245958493e-05,
+      "loss": 2.1371,
+      "step": 7689000
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 3.095861635033169e-05,
+      "loss": 2.1456,
+      "step": 7689500
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 3.095737776390561e-05,
+      "loss": 2.1537,
+      "step": 7690000
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 3.0956139177479525e-05,
+      "loss": 2.167,
+      "step": 7690500
+    },
+    {
+      "epoch": 38.1,
+      "learning_rate": 3.095490059105344e-05,
+      "loss": 2.1423,
+      "step": 7691000
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 3.095366200462736e-05,
+      "loss": 2.1428,
+      "step": 7691500
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 3.095242589537413e-05,
+      "loss": 2.1569,
+      "step": 7692000
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 3.0951187308948045e-05,
+      "loss": 2.1359,
+      "step": 7692500
+    },
+    {
+      "epoch": 38.11,
+      "learning_rate": 3.094994872252196e-05,
+      "loss": 2.1448,
+      "step": 7693000
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 3.094871013609588e-05,
+      "loss": 2.1413,
+      "step": 7693500
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 3.0947471549669795e-05,
+      "loss": 2.1571,
+      "step": 7694000
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 3.094623296324371e-05,
+      "loss": 2.1341,
+      "step": 7694500
+    },
+    {
+      "epoch": 38.12,
+      "learning_rate": 3.094499437681763e-05,
+      "loss": 2.1578,
+      "step": 7695000
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 3.0943755790391546e-05,
+      "loss": 2.1347,
+      "step": 7695500
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 3.094251720396546e-05,
+      "loss": 2.1585,
+      "step": 7696000
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 3.094127861753937e-05,
+      "loss": 2.132,
+      "step": 7696500
+    },
+    {
+      "epoch": 38.13,
+      "learning_rate": 3.094004003111329e-05,
+      "loss": 2.1421,
+      "step": 7697000
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 3.093880144468721e-05,
+      "loss": 2.1641,
+      "step": 7697500
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 3.0937562858261124e-05,
+      "loss": 2.1336,
+      "step": 7698000
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 3.093632427183504e-05,
+      "loss": 2.1314,
+      "step": 7698500
+    },
+    {
+      "epoch": 38.14,
+      "learning_rate": 3.093508816258181e-05,
+      "loss": 2.1504,
+      "step": 7699000
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 3.093385205332858e-05,
+      "loss": 2.1328,
+      "step": 7699500
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 3.0932613466902495e-05,
+      "loss": 2.1288,
+      "step": 7700000
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 3.093137488047641e-05,
+      "loss": 2.1542,
+      "step": 7700500
+    },
+    {
+      "epoch": 38.15,
+      "learning_rate": 3.093013629405033e-05,
+      "loss": 2.1708,
+      "step": 7701000
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 3.0928897707624246e-05,
+      "loss": 2.1498,
+      "step": 7701500
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 3.092765912119816e-05,
+      "loss": 2.1544,
+      "step": 7702000
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 3.0926423011944925e-05,
+      "loss": 2.1696,
+      "step": 7702500
+    },
+    {
+      "epoch": 38.16,
+      "learning_rate": 3.09251869026917e-05,
+      "loss": 2.1817,
+      "step": 7703000
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 3.092394831626562e-05,
+      "loss": 2.1451,
+      "step": 7703500
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 3.0922709729839535e-05,
+      "loss": 2.1684,
+      "step": 7704000
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 3.0921471143413445e-05,
+      "loss": 2.1609,
+      "step": 7704500
+    },
+    {
+      "epoch": 38.17,
+      "learning_rate": 3.092023255698736e-05,
+      "loss": 2.17,
+      "step": 7705000
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 3.091899397056128e-05,
+      "loss": 2.1697,
+      "step": 7705500
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 3.09177603384809e-05,
+      "loss": 2.1148,
+      "step": 7706000
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 3.0916521752054816e-05,
+      "loss": 2.1489,
+      "step": 7706500
+    },
+    {
+      "epoch": 38.18,
+      "learning_rate": 3.091528316562873e-05,
+      "loss": 2.1828,
+      "step": 7707000
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 3.091404457920265e-05,
+      "loss": 2.1395,
+      "step": 7707500
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 3.091280599277657e-05,
+      "loss": 2.1432,
+      "step": 7708000
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 3.0911567406350484e-05,
+      "loss": 2.1308,
+      "step": 7708500
+    },
+    {
+      "epoch": 38.19,
+      "learning_rate": 3.09103288199244e-05,
+      "loss": 2.1493,
+      "step": 7709000
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 3.090909271067117e-05,
+      "loss": 2.1679,
+      "step": 7709500
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 3.090785412424509e-05,
+      "loss": 2.1403,
+      "step": 7710000
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 3.0906615537819004e-05,
+      "loss": 2.1312,
+      "step": 7710500
+    },
+    {
+      "epoch": 38.2,
+      "learning_rate": 3.0905376951392914e-05,
+      "loss": 2.1532,
+      "step": 7711000
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 3.090413836496683e-05,
+      "loss": 2.155,
+      "step": 7711500
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 3.090289977854075e-05,
+      "loss": 2.1598,
+      "step": 7712000
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 3.0901661192114664e-05,
+      "loss": 2.1614,
+      "step": 7712500
+    },
+    {
+      "epoch": 38.21,
+      "learning_rate": 3.090042260568858e-05,
+      "loss": 2.1175,
+      "step": 7713000
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 3.089918401926249e-05,
+      "loss": 2.1656,
+      "step": 7713500
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 3.089794543283641e-05,
+      "loss": 2.1467,
+      "step": 7714000
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 3.0896706846410325e-05,
+      "loss": 2.1399,
+      "step": 7714500
+    },
+    {
+      "epoch": 38.22,
+      "learning_rate": 3.089546825998424e-05,
+      "loss": 2.1499,
+      "step": 7715000
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 3.089422967355816e-05,
+      "loss": 2.1806,
+      "step": 7715500
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 3.0892991087132076e-05,
+      "loss": 2.1437,
+      "step": 7716000
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 3.089175250070599e-05,
+      "loss": 2.1507,
+      "step": 7716500
+    },
+    {
+      "epoch": 38.23,
+      "learning_rate": 3.089051639145276e-05,
+      "loss": 2.1541,
+      "step": 7717000
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 3.088928028219954e-05,
+      "loss": 2.1685,
+      "step": 7717500
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 3.088804169577345e-05,
+      "loss": 2.1489,
+      "step": 7718000
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 3.0886803109347365e-05,
+      "loss": 2.1454,
+      "step": 7718500
+    },
+    {
+      "epoch": 38.24,
+      "learning_rate": 3.088556452292128e-05,
+      "loss": 2.158,
+      "step": 7719000
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 3.08843259364952e-05,
+      "loss": 2.1734,
+      "step": 7719500
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 3.0883087350069115e-05,
+      "loss": 2.1772,
+      "step": 7720000
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 3.0881848763643026e-05,
+      "loss": 2.1496,
+      "step": 7720500
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 3.088061017721694e-05,
+      "loss": 2.1361,
+      "step": 7721000
+    },
+    {
+      "epoch": 38.25,
+      "learning_rate": 3.087937159079086e-05,
+      "loss": 2.1307,
+      "step": 7721500
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 3.087813795871049e-05,
+      "loss": 2.178,
+      "step": 7722000
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 3.087690184945725e-05,
+      "loss": 2.1533,
+      "step": 7722500
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 3.0875663263031166e-05,
+      "loss": 2.1338,
+      "step": 7723000
+    },
+    {
+      "epoch": 38.26,
+      "learning_rate": 3.087442467660508e-05,
+      "loss": 2.149,
+      "step": 7723500
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 3.0873186090179e-05,
+      "loss": 2.1582,
+      "step": 7724000
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 3.0871947503752917e-05,
+      "loss": 2.1487,
+      "step": 7724500
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 3.0870708917326834e-05,
+      "loss": 2.1591,
+      "step": 7725000
+    },
+    {
+      "epoch": 38.27,
+      "learning_rate": 3.086947033090075e-05,
+      "loss": 2.1655,
+      "step": 7725500
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 3.086823174447467e-05,
+      "loss": 2.1918,
+      "step": 7726000
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 3.0866993158048584e-05,
+      "loss": 2.1604,
+      "step": 7726500
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 3.08657545716225e-05,
+      "loss": 2.1778,
+      "step": 7727000
+    },
+    {
+      "epoch": 38.28,
+      "learning_rate": 3.086451598519642e-05,
+      "loss": 2.1669,
+      "step": 7727500
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 3.086327987594319e-05,
+      "loss": 2.1821,
+      "step": 7728000
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 3.0862041289517104e-05,
+      "loss": 2.1247,
+      "step": 7728500
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 3.086080270309102e-05,
+      "loss": 2.1724,
+      "step": 7729000
+    },
+    {
+      "epoch": 38.29,
+      "learning_rate": 3.085956411666494e-05,
+      "loss": 2.1415,
+      "step": 7729500
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 3.0858325530238855e-05,
+      "loss": 2.1568,
+      "step": 7730000
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 3.085708694381277e-05,
+      "loss": 2.1605,
+      "step": 7730500
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 3.085584835738669e-05,
+      "loss": 2.1404,
+      "step": 7731000
+    },
+    {
+      "epoch": 38.3,
+      "learning_rate": 3.08546097709606e-05,
+      "loss": 2.166,
+      "step": 7731500
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 3.085337366170737e-05,
+      "loss": 2.1407,
+      "step": 7732000
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 3.0852135075281284e-05,
+      "loss": 2.151,
+      "step": 7732500
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 3.08508964888552e-05,
+      "loss": 2.1633,
+      "step": 7733000
+    },
+    {
+      "epoch": 38.31,
+      "learning_rate": 3.084965790242912e-05,
+      "loss": 2.1448,
+      "step": 7733500
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 3.0848419316003035e-05,
+      "loss": 2.1802,
+      "step": 7734000
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 3.084718072957695e-05,
+      "loss": 2.1672,
+      "step": 7734500
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 3.084594214315087e-05,
+      "loss": 2.1376,
+      "step": 7735000
+    },
+    {
+      "epoch": 38.32,
+      "learning_rate": 3.084470355672478e-05,
+      "loss": 2.1488,
+      "step": 7735500
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 3.0843464970298696e-05,
+      "loss": 2.1596,
+      "step": 7736000
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 3.084222638387261e-05,
+      "loss": 2.1616,
+      "step": 7736500
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 3.084099027461939e-05,
+      "loss": 2.1577,
+      "step": 7737000
+    },
+    {
+      "epoch": 38.33,
+      "learning_rate": 3.0839751688193306e-05,
+      "loss": 2.1608,
+      "step": 7737500
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 3.083851557894007e-05,
+      "loss": 2.1492,
+      "step": 7738000
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 3.0837276992513985e-05,
+      "loss": 2.1875,
+      "step": 7738500
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 3.08360384060879e-05,
+      "loss": 2.1617,
+      "step": 7739000
+    },
+    {
+      "epoch": 38.34,
+      "learning_rate": 3.083479981966182e-05,
+      "loss": 2.1713,
+      "step": 7739500
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 3.0833561233235735e-05,
+      "loss": 2.1445,
+      "step": 7740000
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 3.083232264680965e-05,
+      "loss": 2.1644,
+      "step": 7740500
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 3.083108406038357e-05,
+      "loss": 2.1407,
+      "step": 7741000
+    },
+    {
+      "epoch": 38.35,
+      "learning_rate": 3.082984795113034e-05,
+      "loss": 2.1581,
+      "step": 7741500
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 3.0828609364704255e-05,
+      "loss": 2.1386,
+      "step": 7742000
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 3.082737077827817e-05,
+      "loss": 2.1513,
+      "step": 7742500
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 3.082613219185209e-05,
+      "loss": 2.1501,
+      "step": 7743000
+    },
+    {
+      "epoch": 38.36,
+      "learning_rate": 3.082489608259885e-05,
+      "loss": 2.1552,
+      "step": 7743500
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 3.082365749617277e-05,
+      "loss": 2.1685,
+      "step": 7744000
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 3.0822418909746685e-05,
+      "loss": 2.1543,
+      "step": 7744500
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 3.08211803233206e-05,
+      "loss": 2.1393,
+      "step": 7745000
+    },
+    {
+      "epoch": 38.37,
+      "learning_rate": 3.081994173689452e-05,
+      "loss": 2.1423,
+      "step": 7745500
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 3.0818703150468435e-05,
+      "loss": 2.1453,
+      "step": 7746000
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 3.081746456404235e-05,
+      "loss": 2.1586,
+      "step": 7746500
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 3.081622597761627e-05,
+      "loss": 2.1634,
+      "step": 7747000
+    },
+    {
+      "epoch": 38.38,
+      "learning_rate": 3.0814987391190186e-05,
+      "loss": 2.1584,
+      "step": 7747500
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 3.0813748804764096e-05,
+      "loss": 2.1648,
+      "step": 7748000
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 3.081251021833801e-05,
+      "loss": 2.1442,
+      "step": 7748500
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 3.081127163191193e-05,
+      "loss": 2.1382,
+      "step": 7749000
+    },
+    {
+      "epoch": 38.39,
+      "learning_rate": 3.081003304548585e-05,
+      "loss": 2.139,
+      "step": 7749500
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 3.0808794459059764e-05,
+      "loss": 2.1524,
+      "step": 7750000
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 3.080755587263368e-05,
+      "loss": 2.1801,
+      "step": 7750500
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 3.080631976338045e-05,
+      "loss": 2.1802,
+      "step": 7751000
+    },
+    {
+      "epoch": 38.4,
+      "learning_rate": 3.080508365412722e-05,
+      "loss": 2.1441,
+      "step": 7751500
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 3.0803845067701136e-05,
+      "loss": 2.1448,
+      "step": 7752000
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 3.0802608958447904e-05,
+      "loss": 2.1827,
+      "step": 7752500
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 3.080137037202182e-05,
+      "loss": 2.1571,
+      "step": 7753000
+    },
+    {
+      "epoch": 38.41,
+      "learning_rate": 3.080013178559574e-05,
+      "loss": 2.1433,
+      "step": 7753500
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 3.0798893199169655e-05,
+      "loss": 2.1795,
+      "step": 7754000
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 3.079765461274357e-05,
+      "loss": 2.1581,
+      "step": 7754500
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 3.079641602631749e-05,
+      "loss": 2.1609,
+      "step": 7755000
+    },
+    {
+      "epoch": 38.42,
+      "learning_rate": 3.0795177439891406e-05,
+      "loss": 2.1601,
+      "step": 7755500
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 3.079394133063817e-05,
+      "loss": 2.1734,
+      "step": 7756000
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 3.0792702744212085e-05,
+      "loss": 2.1364,
+      "step": 7756500
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 3.0791464157786e-05,
+      "loss": 2.1786,
+      "step": 7757000
+    },
+    {
+      "epoch": 38.43,
+      "learning_rate": 3.079022557135992e-05,
+      "loss": 2.174,
+      "step": 7757500
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 3.0788986984933836e-05,
+      "loss": 2.138,
+      "step": 7758000
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 3.078774839850775e-05,
+      "loss": 2.1685,
+      "step": 7758500
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 3.078650981208167e-05,
+      "loss": 2.1682,
+      "step": 7759000
+    },
+    {
+      "epoch": 38.44,
+      "learning_rate": 3.0785271225655587e-05,
+      "loss": 2.1384,
+      "step": 7759500
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 3.0784032639229503e-05,
+      "loss": 2.1512,
+      "step": 7760000
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 3.0782794052803414e-05,
+      "loss": 2.1479,
+      "step": 7760500
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 3.078155546637733e-05,
+      "loss": 2.1676,
+      "step": 7761000
+    },
+    {
+      "epoch": 38.45,
+      "learning_rate": 3.078031687995125e-05,
+      "loss": 2.1731,
+      "step": 7761500
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 3.0779078293525164e-05,
+      "loss": 2.1725,
+      "step": 7762000
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 3.077784218427194e-05,
+      "loss": 2.1648,
+      "step": 7762500
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 3.07766060750187e-05,
+      "loss": 2.1484,
+      "step": 7763000
+    },
+    {
+      "epoch": 38.46,
+      "learning_rate": 3.077536748859262e-05,
+      "loss": 2.1783,
+      "step": 7763500
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 3.0774128902166536e-05,
+      "loss": 2.1605,
+      "step": 7764000
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 3.077289031574045e-05,
+      "loss": 2.1725,
+      "step": 7764500
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 3.077165172931437e-05,
+      "loss": 2.1651,
+      "step": 7765000
+    },
+    {
+      "epoch": 38.47,
+      "learning_rate": 3.077041562006114e-05,
+      "loss": 2.1763,
+      "step": 7765500
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 3.0769177033635055e-05,
+      "loss": 2.146,
+      "step": 7766000
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 3.076793844720897e-05,
+      "loss": 2.168,
+      "step": 7766500
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 3.076669986078289e-05,
+      "loss": 2.145,
+      "step": 7767000
+    },
+    {
+      "epoch": 38.48,
+      "learning_rate": 3.0765461274356806e-05,
+      "loss": 2.1671,
+      "step": 7767500
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 3.076422268793072e-05,
+      "loss": 2.1792,
+      "step": 7768000
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 3.076298410150464e-05,
+      "loss": 2.1563,
+      "step": 7768500
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 3.076174551507856e-05,
+      "loss": 2.1745,
+      "step": 7769000
+    },
+    {
+      "epoch": 38.49,
+      "learning_rate": 3.0760506928652474e-05,
+      "loss": 2.1663,
+      "step": 7769500
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 3.0759270819399236e-05,
+      "loss": 2.1363,
+      "step": 7770000
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 3.075803223297315e-05,
+      "loss": 2.1503,
+      "step": 7770500
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 3.075679364654707e-05,
+      "loss": 2.1608,
+      "step": 7771000
+    },
+    {
+      "epoch": 38.5,
+      "learning_rate": 3.075555506012099e-05,
+      "loss": 2.1483,
+      "step": 7771500
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 3.0754316473694904e-05,
+      "loss": 2.1652,
+      "step": 7772000
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 3.075307788726882e-05,
+      "loss": 2.148,
+      "step": 7772500
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 3.075183930084273e-05,
+      "loss": 2.1688,
+      "step": 7773000
+    },
+    {
+      "epoch": 38.51,
+      "learning_rate": 3.0750603191589506e-05,
+      "loss": 2.1357,
+      "step": 7773500
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 3.074936460516342e-05,
+      "loss": 2.1664,
+      "step": 7774000
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 3.074812849591019e-05,
+      "loss": 2.1626,
+      "step": 7774500
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 3.074688990948411e-05,
+      "loss": 2.15,
+      "step": 7775000
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 3.074565132305802e-05,
+      "loss": 2.1828,
+      "step": 7775500
+    },
+    {
+      "epoch": 38.52,
+      "learning_rate": 3.0744412736631936e-05,
+      "loss": 2.1476,
+      "step": 7776000
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 3.0743176627378705e-05,
+      "loss": 2.1792,
+      "step": 7776500
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 3.074193804095262e-05,
+      "loss": 2.179,
+      "step": 7777000
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 3.074069945452654e-05,
+      "loss": 2.1365,
+      "step": 7777500
+    },
+    {
+      "epoch": 38.53,
+      "learning_rate": 3.0739460868100456e-05,
+      "loss": 2.1565,
+      "step": 7778000
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 3.073822228167437e-05,
+      "loss": 2.1655,
+      "step": 7778500
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 3.073698369524829e-05,
+      "loss": 2.1732,
+      "step": 7779000
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 3.0735745108822206e-05,
+      "loss": 2.1556,
+      "step": 7779500
+    },
+    {
+      "epoch": 38.54,
+      "learning_rate": 3.0734506522396123e-05,
+      "loss": 2.1604,
+      "step": 7780000
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 3.073326793597004e-05,
+      "loss": 2.1461,
+      "step": 7780500
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 3.073203182671681e-05,
+      "loss": 2.1542,
+      "step": 7781000
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 3.0730793240290726e-05,
+      "loss": 2.164,
+      "step": 7781500
+    },
+    {
+      "epoch": 38.55,
+      "learning_rate": 3.0729554653864636e-05,
+      "loss": 2.1792,
+      "step": 7782000
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 3.0728321021784264e-05,
+      "loss": 2.1498,
+      "step": 7782500
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 3.0727082435358174e-05,
+      "loss": 2.1584,
+      "step": 7783000
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 3.072584384893209e-05,
+      "loss": 2.1797,
+      "step": 7783500
+    },
+    {
+      "epoch": 38.56,
+      "learning_rate": 3.072460526250601e-05,
+      "loss": 2.1565,
+      "step": 7784000
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 3.0723366676079925e-05,
+      "loss": 2.1586,
+      "step": 7784500
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 3.072212808965384e-05,
+      "loss": 2.1527,
+      "step": 7785000
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 3.072088950322776e-05,
+      "loss": 2.1582,
+      "step": 7785500
+    },
+    {
+      "epoch": 38.57,
+      "learning_rate": 3.0719650916801675e-05,
+      "loss": 2.1819,
+      "step": 7786000
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 3.071841233037559e-05,
+      "loss": 2.1677,
+      "step": 7786500
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 3.071717374394951e-05,
+      "loss": 2.1631,
+      "step": 7787000
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 3.0715935157523426e-05,
+      "loss": 2.153,
+      "step": 7787500
+    },
+    {
+      "epoch": 38.58,
+      "learning_rate": 3.071469657109734e-05,
+      "loss": 2.1566,
+      "step": 7788000
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 3.071345798467126e-05,
+      "loss": 2.1773,
+      "step": 7788500
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 3.071221939824517e-05,
+      "loss": 2.1527,
+      "step": 7789000
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 3.071098081181909e-05,
+      "loss": 2.1608,
+      "step": 7789500
+    },
+    {
+      "epoch": 38.59,
+      "learning_rate": 3.0709744702565856e-05,
+      "loss": 2.1594,
+      "step": 7790000
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 3.070850611613977e-05,
+      "loss": 2.1544,
+      "step": 7790500
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 3.070726752971369e-05,
+      "loss": 2.1647,
+      "step": 7791000
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 3.070603142046046e-05,
+      "loss": 2.146,
+      "step": 7791500
+    },
+    {
+      "epoch": 38.6,
+      "learning_rate": 3.0704792834034376e-05,
+      "loss": 2.1485,
+      "step": 7792000
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 3.070355424760829e-05,
+      "loss": 2.1703,
+      "step": 7792500
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 3.070231566118221e-05,
+      "loss": 2.1722,
+      "step": 7793000
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 3.0701077074756126e-05,
+      "loss": 2.1854,
+      "step": 7793500
+    },
+    {
+      "epoch": 38.61,
+      "learning_rate": 3.069983848833004e-05,
+      "loss": 2.1533,
+      "step": 7794000
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 3.069859990190396e-05,
+      "loss": 2.141,
+      "step": 7794500
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 3.069736131547788e-05,
+      "loss": 2.1703,
+      "step": 7795000
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 3.069612520622464e-05,
+      "loss": 2.1582,
+      "step": 7795500
+    },
+    {
+      "epoch": 38.62,
+      "learning_rate": 3.0694886619798556e-05,
+      "loss": 2.1393,
+      "step": 7796000
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 3.0693650510545325e-05,
+      "loss": 2.1568,
+      "step": 7796500
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 3.069241192411924e-05,
+      "loss": 2.1373,
+      "step": 7797000
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 3.069117333769316e-05,
+      "loss": 2.1609,
+      "step": 7797500
+    },
+    {
+      "epoch": 38.63,
+      "learning_rate": 3.0689934751267076e-05,
+      "loss": 2.1931,
+      "step": 7798000
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 3.068869616484099e-05,
+      "loss": 2.1741,
+      "step": 7798500
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 3.068745757841491e-05,
+      "loss": 2.1588,
+      "step": 7799000
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 3.068622146916167e-05,
+      "loss": 2.1735,
+      "step": 7799500
+    },
+    {
+      "epoch": 38.64,
+      "learning_rate": 3.068498288273559e-05,
+      "loss": 2.1634,
+      "step": 7800000
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 3.0683744296309505e-05,
+      "loss": 2.1464,
+      "step": 7800500
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 3.068250570988342e-05,
+      "loss": 2.1824,
+      "step": 7801000
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 3.068126712345734e-05,
+      "loss": 2.1648,
+      "step": 7801500
+    },
+    {
+      "epoch": 38.65,
+      "learning_rate": 3.0680028537031256e-05,
+      "loss": 2.1464,
+      "step": 7802000
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 3.067878995060517e-05,
+      "loss": 2.1653,
+      "step": 7802500
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 3.067755136417909e-05,
+      "loss": 2.1694,
+      "step": 7803000
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 3.067631277775301e-05,
+      "loss": 2.1317,
+      "step": 7803500
+    },
+    {
+      "epoch": 38.66,
+      "learning_rate": 3.0675076668499776e-05,
+      "loss": 2.166,
+      "step": 7804000
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 3.067383808207369e-05,
+      "loss": 2.1336,
+      "step": 7804500
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 3.067259949564761e-05,
+      "loss": 2.1756,
+      "step": 7805000
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 3.0671360909221527e-05,
+      "loss": 2.1769,
+      "step": 7805500
+    },
+    {
+      "epoch": 38.67,
+      "learning_rate": 3.0670122322795443e-05,
+      "loss": 2.168,
+      "step": 7806000
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 3.066888373636936e-05,
+      "loss": 2.1716,
+      "step": 7806500
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 3.066764762711612e-05,
+      "loss": 2.1611,
+      "step": 7807000
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 3.066640904069004e-05,
+      "loss": 2.1616,
+      "step": 7807500
+    },
+    {
+      "epoch": 38.68,
+      "learning_rate": 3.0665170454263956e-05,
+      "loss": 2.1478,
+      "step": 7808000
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 3.066393186783787e-05,
+      "loss": 2.1736,
+      "step": 7808500
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 3.066269328141179e-05,
+      "loss": 2.1648,
+      "step": 7809000
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 3.066145469498571e-05,
+      "loss": 2.1752,
+      "step": 7809500
+    },
+    {
+      "epoch": 38.69,
+      "learning_rate": 3.0660216108559624e-05,
+      "loss": 2.1323,
+      "step": 7810000
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 3.065897752213354e-05,
+      "loss": 2.1641,
+      "step": 7810500
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 3.065773893570746e-05,
+      "loss": 2.2023,
+      "step": 7811000
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 3.0656500349281375e-05,
+      "loss": 2.1507,
+      "step": 7811500
+    },
+    {
+      "epoch": 38.7,
+      "learning_rate": 3.065526176285529e-05,
+      "loss": 2.1431,
+      "step": 7812000
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 3.065402317642921e-05,
+      "loss": 2.1635,
+      "step": 7812500
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 3.0652784590003126e-05,
+      "loss": 2.1486,
+      "step": 7813000
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 3.0651548480749894e-05,
+      "loss": 2.1406,
+      "step": 7813500
+    },
+    {
+      "epoch": 38.71,
+      "learning_rate": 3.065030989432381e-05,
+      "loss": 2.1673,
+      "step": 7814000
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 3.064907130789773e-05,
+      "loss": 2.1746,
+      "step": 7814500
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 3.0647832721471645e-05,
+      "loss": 2.163,
+      "step": 7815000
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 3.064659661221841e-05,
+      "loss": 2.1701,
+      "step": 7815500
+    },
+    {
+      "epoch": 38.72,
+      "learning_rate": 3.0645358025792324e-05,
+      "loss": 2.1663,
+      "step": 7816000
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 3.064411943936624e-05,
+      "loss": 2.1682,
+      "step": 7816500
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 3.064288085294016e-05,
+      "loss": 2.1865,
+      "step": 7817000
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 3.064164474368693e-05,
+      "loss": 2.1702,
+      "step": 7817500
+    },
+    {
+      "epoch": 38.73,
+      "learning_rate": 3.0640406157260844e-05,
+      "loss": 2.1665,
+      "step": 7818000
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 3.063916757083476e-05,
+      "loss": 2.1762,
+      "step": 7818500
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 3.063792898440868e-05,
+      "loss": 2.146,
+      "step": 7819000
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 3.0636690397982595e-05,
+      "loss": 2.1585,
+      "step": 7819500
+    },
+    {
+      "epoch": 38.74,
+      "learning_rate": 3.063545181155651e-05,
+      "loss": 2.1604,
+      "step": 7820000
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 3.063421322513043e-05,
+      "loss": 2.1483,
+      "step": 7820500
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 3.0632974638704345e-05,
+      "loss": 2.1786,
+      "step": 7821000
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 3.063173605227826e-05,
+      "loss": 2.1559,
+      "step": 7821500
+    },
+    {
+      "epoch": 38.75,
+      "learning_rate": 3.063049746585218e-05,
+      "loss": 2.1845,
+      "step": 7822000
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 3.062925887942609e-05,
+      "loss": 2.1505,
+      "step": 7822500
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 3.0628020293000006e-05,
+      "loss": 2.1552,
+      "step": 7823000
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 3.0626784183746775e-05,
+      "loss": 2.1628,
+      "step": 7823500
+    },
+    {
+      "epoch": 38.76,
+      "learning_rate": 3.062554559732069e-05,
+      "loss": 2.154,
+      "step": 7824000
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 3.062430701089461e-05,
+      "loss": 2.1525,
+      "step": 7824500
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 3.0623068424468526e-05,
+      "loss": 2.1739,
+      "step": 7825000
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 3.062182983804244e-05,
+      "loss": 2.1721,
+      "step": 7825500
+    },
+    {
+      "epoch": 38.77,
+      "learning_rate": 3.062059125161635e-05,
+      "loss": 2.1654,
+      "step": 7826000
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 3.061935266519027e-05,
+      "loss": 2.1693,
+      "step": 7826500
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 3.061811407876419e-05,
+      "loss": 2.1554,
+      "step": 7827000
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 3.061687796951096e-05,
+      "loss": 2.1482,
+      "step": 7827500
+    },
+    {
+      "epoch": 38.78,
+      "learning_rate": 3.061563938308488e-05,
+      "loss": 2.1546,
+      "step": 7828000
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 3.0614400796658796e-05,
+      "loss": 2.1799,
+      "step": 7828500
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 3.061316221023271e-05,
+      "loss": 2.1825,
+      "step": 7829000
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 3.061192857815233e-05,
+      "loss": 2.1486,
+      "step": 7829500
+    },
+    {
+      "epoch": 38.79,
+      "learning_rate": 3.0610692468899096e-05,
+      "loss": 2.1515,
+      "step": 7830000
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 3.060945388247301e-05,
+      "loss": 2.1725,
+      "step": 7830500
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 3.060821529604693e-05,
+      "loss": 2.1588,
+      "step": 7831000
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 3.060697670962085e-05,
+      "loss": 2.1636,
+      "step": 7831500
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 3.0605738123194764e-05,
+      "loss": 2.1585,
+      "step": 7832000
+    },
+    {
+      "epoch": 38.8,
+      "learning_rate": 3.0604499536768674e-05,
+      "loss": 2.1466,
+      "step": 7832500
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 3.060326095034259e-05,
+      "loss": 2.1806,
+      "step": 7833000
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 3.060202236391651e-05,
+      "loss": 2.1412,
+      "step": 7833500
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 3.0600783777490425e-05,
+      "loss": 2.1794,
+      "step": 7834000
+    },
+    {
+      "epoch": 38.81,
+      "learning_rate": 3.059954519106434e-05,
+      "loss": 2.1508,
+      "step": 7834500
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 3.059830660463826e-05,
+      "loss": 2.1664,
+      "step": 7835000
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 3.0597068018212175e-05,
+      "loss": 2.1578,
+      "step": 7835500
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 3.059582943178609e-05,
+      "loss": 2.1817,
+      "step": 7836000
+    },
+    {
+      "epoch": 38.82,
+      "learning_rate": 3.059459084536001e-05,
+      "loss": 2.1615,
+      "step": 7836500
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 3.059335473610678e-05,
+      "loss": 2.1464,
+      "step": 7837000
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 3.0592116149680695e-05,
+      "loss": 2.1565,
+      "step": 7837500
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 3.059087756325461e-05,
+      "loss": 2.1671,
+      "step": 7838000
+    },
+    {
+      "epoch": 38.83,
+      "learning_rate": 3.058964145400138e-05,
+      "loss": 2.1436,
+      "step": 7838500
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 3.05884028675753e-05,
+      "loss": 2.1457,
+      "step": 7839000
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 3.058716428114921e-05,
+      "loss": 2.1704,
+      "step": 7839500
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 3.0585925694723125e-05,
+      "loss": 2.1605,
+      "step": 7840000
+    },
+    {
+      "epoch": 38.84,
+      "learning_rate": 3.058469206264275e-05,
+      "loss": 2.1886,
+      "step": 7840500
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 3.058345347621667e-05,
+      "loss": 2.1732,
+      "step": 7841000
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 3.0582214889790586e-05,
+      "loss": 2.1919,
+      "step": 7841500
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 3.05809763033645e-05,
+      "loss": 2.1547,
+      "step": 7842000
+    },
+    {
+      "epoch": 38.85,
+      "learning_rate": 3.057973771693841e-05,
+      "loss": 2.182,
+      "step": 7842500
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 3.057849913051233e-05,
+      "loss": 2.1522,
+      "step": 7843000
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 3.057726054408625e-05,
+      "loss": 2.1648,
+      "step": 7843500
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 3.0576021957660164e-05,
+      "loss": 2.1581,
+      "step": 7844000
+    },
+    {
+      "epoch": 38.86,
+      "learning_rate": 3.057478337123408e-05,
+      "loss": 2.1647,
+      "step": 7844500
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 3.0573544784808e-05,
+      "loss": 2.1649,
+      "step": 7845000
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 3.0572306198381915e-05,
+      "loss": 2.149,
+      "step": 7845500
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 3.0571067611955825e-05,
+      "loss": 2.1535,
+      "step": 7846000
+    },
+    {
+      "epoch": 38.87,
+      "learning_rate": 3.056982902552974e-05,
+      "loss": 2.1511,
+      "step": 7846500
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 3.056859043910366e-05,
+      "loss": 2.1786,
+      "step": 7847000
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 3.0567351852677576e-05,
+      "loss": 2.1551,
+      "step": 7847500
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 3.056611326625149e-05,
+      "loss": 2.1863,
+      "step": 7848000
+    },
+    {
+      "epoch": 38.88,
+      "learning_rate": 3.056487715699826e-05,
+      "loss": 2.1915,
+      "step": 7848500
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 3.056364104774503e-05,
+      "loss": 2.163,
+      "step": 7849000
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 3.056240246131895e-05,
+      "loss": 2.1685,
+      "step": 7849500
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 3.0561166352065716e-05,
+      "loss": 2.1734,
+      "step": 7850000
+    },
+    {
+      "epoch": 38.89,
+      "learning_rate": 3.055992776563963e-05,
+      "loss": 2.1701,
+      "step": 7850500
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 3.055868917921355e-05,
+      "loss": 2.1444,
+      "step": 7851000
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 3.055745059278747e-05,
+      "loss": 2.1707,
+      "step": 7851500
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 3.055621200636138e-05,
+      "loss": 2.1581,
+      "step": 7852000
+    },
+    {
+      "epoch": 38.9,
+      "learning_rate": 3.0554973419935294e-05,
+      "loss": 2.1871,
+      "step": 7852500
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 3.055373483350921e-05,
+      "loss": 2.19,
+      "step": 7853000
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 3.055249624708313e-05,
+      "loss": 2.1667,
+      "step": 7853500
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 3.0551257660657044e-05,
+      "loss": 2.1786,
+      "step": 7854000
+    },
+    {
+      "epoch": 38.91,
+      "learning_rate": 3.055002155140382e-05,
+      "loss": 2.1762,
+      "step": 7854500
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 3.054878296497773e-05,
+      "loss": 2.1552,
+      "step": 7855000
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 3.054754437855165e-05,
+      "loss": 2.1586,
+      "step": 7855500
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 3.0546305792125564e-05,
+      "loss": 2.1457,
+      "step": 7856000
+    },
+    {
+      "epoch": 38.92,
+      "learning_rate": 3.054506968287233e-05,
+      "loss": 2.1658,
+      "step": 7856500
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 3.054383109644625e-05,
+      "loss": 2.1832,
+      "step": 7857000
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 3.054259251002017e-05,
+      "loss": 2.1612,
+      "step": 7857500
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 3.054135392359408e-05,
+      "loss": 2.1709,
+      "step": 7858000
+    },
+    {
+      "epoch": 38.93,
+      "learning_rate": 3.0540115337167994e-05,
+      "loss": 2.1588,
+      "step": 7858500
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 3.053887922791477e-05,
+      "loss": 2.1646,
+      "step": 7859000
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 3.0537640641488686e-05,
+      "loss": 2.1794,
+      "step": 7859500
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 3.0536404532235455e-05,
+      "loss": 2.178,
+      "step": 7860000
+    },
+    {
+      "epoch": 38.94,
+      "learning_rate": 3.0535165945809365e-05,
+      "loss": 2.1825,
+      "step": 7860500
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 3.053392735938328e-05,
+      "loss": 2.1545,
+      "step": 7861000
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 3.05326887729572e-05,
+      "loss": 2.1719,
+      "step": 7861500
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 3.053145266370397e-05,
+      "loss": 2.1506,
+      "step": 7862000
+    },
+    {
+      "epoch": 38.95,
+      "learning_rate": 3.0530214077277885e-05,
+      "loss": 2.1656,
+      "step": 7862500
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 3.05289754908518e-05,
+      "loss": 2.1701,
+      "step": 7863000
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 3.052773690442572e-05,
+      "loss": 2.1722,
+      "step": 7863500
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 3.052650079517249e-05,
+      "loss": 2.1621,
+      "step": 7864000
+    },
+    {
+      "epoch": 38.96,
+      "learning_rate": 3.0525262208746404e-05,
+      "loss": 2.1586,
+      "step": 7864500
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 3.052402362232032e-05,
+      "loss": 2.1747,
+      "step": 7865000
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 3.052278503589424e-05,
+      "loss": 2.1669,
+      "step": 7865500
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 3.0521546449468155e-05,
+      "loss": 2.17,
+      "step": 7866000
+    },
+    {
+      "epoch": 38.97,
+      "learning_rate": 3.052030786304207e-05,
+      "loss": 2.1544,
+      "step": 7866500
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 3.051906927661599e-05,
+      "loss": 2.161,
+      "step": 7867000
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 3.05178306901899e-05,
+      "loss": 2.1717,
+      "step": 7867500
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 3.0516594580936668e-05,
+      "loss": 2.1653,
+      "step": 7868000
+    },
+    {
+      "epoch": 38.98,
+      "learning_rate": 3.0515355994510585e-05,
+      "loss": 2.1722,
+      "step": 7868500
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 3.0514117408084502e-05,
+      "loss": 2.1728,
+      "step": 7869000
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 3.051287882165842e-05,
+      "loss": 2.1704,
+      "step": 7869500
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 3.0511640235232336e-05,
+      "loss": 2.1638,
+      "step": 7870000
+    },
+    {
+      "epoch": 38.99,
+      "learning_rate": 3.0510401648806253e-05,
+      "loss": 2.1512,
+      "step": 7870500
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.0509165539553018e-05,
+      "loss": 2.1884,
+      "step": 7871000
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.0507926953126935e-05,
+      "loss": 2.1752,
+      "step": 7871500
+    },
+    {
+      "epoch": 39.0,
+      "eval_accuracy": 0.6649884989419301,
+      "eval_accuracy_mlm": 0.6219056881216726,
+      "eval_accuracy_nsp": 0.867931706666562,
+      "eval_loss": 2.296628952026367,
+      "eval_runtime": 146.0071,
+      "eval_samples_per_second": 1746.209,
+      "eval_steps_per_second": 72.764,
+      "step": 7871877
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.0506688366700852e-05,
+      "loss": 2.1657,
+      "step": 7872000
+    },
+    {
+      "epoch": 39.0,
+      "learning_rate": 3.050544978027477e-05,
+      "loss": 2.1253,
+      "step": 7872500
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.0504211193848686e-05,
+      "loss": 2.1353,
+      "step": 7873000
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.0502972607422603e-05,
+      "loss": 2.1536,
+      "step": 7873500
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.0501736498169368e-05,
+      "loss": 2.1249,
+      "step": 7874000
+    },
+    {
+      "epoch": 39.01,
+      "learning_rate": 3.0500497911743285e-05,
+      "loss": 2.1628,
+      "step": 7874500
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 3.0499259325317202e-05,
+      "loss": 2.1184,
+      "step": 7875000
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 3.049802073889112e-05,
+      "loss": 2.1196,
+      "step": 7875500
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 3.0496782152465036e-05,
+      "loss": 2.1508,
+      "step": 7876000
+    },
+    {
+      "epoch": 39.02,
+      "learning_rate": 3.0495546043211805e-05,
+      "loss": 2.1501,
+      "step": 7876500
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 3.049430745678572e-05,
+      "loss": 2.1245,
+      "step": 7877000
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 3.0493068870359635e-05,
+      "loss": 2.1496,
+      "step": 7877500
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 3.0491830283933552e-05,
+      "loss": 2.1452,
+      "step": 7878000
+    },
+    {
+      "epoch": 39.03,
+      "learning_rate": 3.049059169750747e-05,
+      "loss": 2.1498,
+      "step": 7878500
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 3.0489353111081386e-05,
+      "loss": 2.1396,
+      "step": 7879000
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 3.0488114524655303e-05,
+      "loss": 2.1236,
+      "step": 7879500
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 3.048687593822922e-05,
+      "loss": 2.1515,
+      "step": 7880000
+    },
+    {
+      "epoch": 39.04,
+      "learning_rate": 3.048563982897599e-05,
+      "loss": 2.1484,
+      "step": 7880500
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 3.0484401242549902e-05,
+      "loss": 2.133,
+      "step": 7881000
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 3.048316265612382e-05,
+      "loss": 2.1314,
+      "step": 7881500
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 3.0481926546870588e-05,
+      "loss": 2.1429,
+      "step": 7882000
+    },
+    {
+      "epoch": 39.05,
+      "learning_rate": 3.0480687960444505e-05,
+      "loss": 2.1548,
+      "step": 7882500
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 3.0479451851191277e-05,
+      "loss": 2.1394,
+      "step": 7883000
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 3.0478213264765194e-05,
+      "loss": 2.1301,
+      "step": 7883500
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 3.0476974678339104e-05,
+      "loss": 2.1422,
+      "step": 7884000
+    },
+    {
+      "epoch": 39.06,
+      "learning_rate": 3.047573609191302e-05,
+      "loss": 2.137,
+      "step": 7884500
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 3.0474497505486938e-05,
+      "loss": 2.1319,
+      "step": 7885000
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 3.0473258919060855e-05,
+      "loss": 2.1218,
+      "step": 7885500
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 3.0472020332634772e-05,
+      "loss": 2.1418,
+      "step": 7886000
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 3.047078174620869e-05,
+      "loss": 2.1179,
+      "step": 7886500
+    },
+    {
+      "epoch": 39.07,
+      "learning_rate": 3.0469543159782606e-05,
+      "loss": 2.133,
+      "step": 7887000
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 3.046830457335652e-05,
+      "loss": 2.1354,
+      "step": 7887500
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 3.0467065986930436e-05,
+      "loss": 2.1265,
+      "step": 7888000
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 3.0465827400504353e-05,
+      "loss": 2.137,
+      "step": 7888500
+    },
+    {
+      "epoch": 39.08,
+      "learning_rate": 3.046458881407827e-05,
+      "loss": 2.1538,
+      "step": 7889000
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 3.0463350227652187e-05,
+      "loss": 2.1145,
+      "step": 7889500
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 3.0462111641226104e-05,
+      "loss": 2.1411,
+      "step": 7890000
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 3.046087305480002e-05,
+      "loss": 2.1373,
+      "step": 7890500
+    },
+    {
+      "epoch": 39.09,
+      "learning_rate": 3.0459634468373938e-05,
+      "loss": 2.1356,
+      "step": 7891000
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 3.0458398359120703e-05,
+      "loss": 2.1337,
+      "step": 7891500
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 3.045715977269462e-05,
+      "loss": 2.1247,
+      "step": 7892000
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 3.0455921186268537e-05,
+      "loss": 2.1473,
+      "step": 7892500
+    },
+    {
+      "epoch": 39.1,
+      "learning_rate": 3.0454682599842454e-05,
+      "loss": 2.1322,
+      "step": 7893000
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 3.045344401341637e-05,
+      "loss": 2.1385,
+      "step": 7893500
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 3.0452205426990288e-05,
+      "loss": 2.1519,
+      "step": 7894000
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 3.0450966840564205e-05,
+      "loss": 2.1409,
+      "step": 7894500
+    },
+    {
+      "epoch": 39.11,
+      "learning_rate": 3.044973073131097e-05,
+      "loss": 2.1508,
+      "step": 7895000
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 3.0448492144884887e-05,
+      "loss": 2.1595,
+      "step": 7895500
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 3.0447253558458804e-05,
+      "loss": 2.1137,
+      "step": 7896000
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 3.044601497203272e-05,
+      "loss": 2.1519,
+      "step": 7896500
+    },
+    {
+      "epoch": 39.12,
+      "learning_rate": 3.044477886277949e-05,
+      "loss": 2.1222,
+      "step": 7897000
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 3.0443540276353407e-05,
+      "loss": 2.1545,
+      "step": 7897500
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 3.044230168992732e-05,
+      "loss": 2.132,
+      "step": 7898000
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 3.0441063103501237e-05,
+      "loss": 2.1354,
+      "step": 7898500
+    },
+    {
+      "epoch": 39.13,
+      "learning_rate": 3.0439826994248006e-05,
+      "loss": 2.1435,
+      "step": 7899000
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 3.0438588407821923e-05,
+      "loss": 2.1475,
+      "step": 7899500
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 3.043734982139584e-05,
+      "loss": 2.1216,
+      "step": 7900000
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 3.0436113712142605e-05,
+      "loss": 2.1333,
+      "step": 7900500
+    },
+    {
+      "epoch": 39.14,
+      "learning_rate": 3.0434875125716522e-05,
+      "loss": 2.1707,
+      "step": 7901000
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 3.043363653929044e-05,
+      "loss": 2.1407,
+      "step": 7901500
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 3.0432397952864356e-05,
+      "loss": 2.1273,
+      "step": 7902000
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 3.0431159366438273e-05,
+      "loss": 2.1465,
+      "step": 7902500
+    },
+    {
+      "epoch": 39.15,
+      "learning_rate": 3.042992325718504e-05,
+      "loss": 2.1403,
+      "step": 7903000
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 3.0428684670758955e-05,
+      "loss": 2.1304,
+      "step": 7903500
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 3.0427446084332872e-05,
+      "loss": 2.1517,
+      "step": 7904000
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 3.042620749790679e-05,
+      "loss": 2.1458,
+      "step": 7904500
+    },
+    {
+      "epoch": 39.16,
+      "learning_rate": 3.0424968911480706e-05,
+      "loss": 2.1455,
+      "step": 7905000
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 3.0423730325054623e-05,
+      "loss": 2.1511,
+      "step": 7905500
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 3.042249173862854e-05,
+      "loss": 2.1395,
+      "step": 7906000
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 3.0421253152202457e-05,
+      "loss": 2.1483,
+      "step": 7906500
+    },
+    {
+      "epoch": 39.17,
+      "learning_rate": 3.0420014565776374e-05,
+      "loss": 2.1353,
+      "step": 7907000
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 3.041877597935029e-05,
+      "loss": 2.1086,
+      "step": 7907500
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 3.0417537392924204e-05,
+      "loss": 2.1319,
+      "step": 7908000
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 3.041629880649812e-05,
+      "loss": 2.1362,
+      "step": 7908500
+    },
+    {
+      "epoch": 39.18,
+      "learning_rate": 3.041506269724489e-05,
+      "loss": 2.1574,
+      "step": 7909000
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 3.0413826587991655e-05,
+      "loss": 2.149,
+      "step": 7909500
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 3.0412588001565572e-05,
+      "loss": 2.1484,
+      "step": 7910000
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 3.041134941513949e-05,
+      "loss": 2.1333,
+      "step": 7910500
+    },
+    {
+      "epoch": 39.19,
+      "learning_rate": 3.0410110828713406e-05,
+      "loss": 2.1348,
+      "step": 7911000
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 3.0408872242287323e-05,
+      "loss": 2.1438,
+      "step": 7911500
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 3.040763365586124e-05,
+      "loss": 2.1444,
+      "step": 7912000
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 3.0406395069435157e-05,
+      "loss": 2.1385,
+      "step": 7912500
+    },
+    {
+      "epoch": 39.2,
+      "learning_rate": 3.0405156483009074e-05,
+      "loss": 2.1308,
+      "step": 7913000
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 3.040391789658299e-05,
+      "loss": 2.1406,
+      "step": 7913500
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 3.0402679310156908e-05,
+      "loss": 2.1446,
+      "step": 7914000
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 3.040144072373082e-05,
+      "loss": 2.1441,
+      "step": 7914500
+    },
+    {
+      "epoch": 39.21,
+      "learning_rate": 3.040020461447759e-05,
+      "loss": 2.1577,
+      "step": 7915000
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 3.0398966028051507e-05,
+      "loss": 2.1304,
+      "step": 7915500
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 3.0397727441625424e-05,
+      "loss": 2.1456,
+      "step": 7916000
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 3.039648885519934e-05,
+      "loss": 2.1595,
+      "step": 7916500
+    },
+    {
+      "epoch": 39.22,
+      "learning_rate": 3.0395252745946106e-05,
+      "loss": 2.1297,
+      "step": 7917000
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 3.0394014159520023e-05,
+      "loss": 2.1346,
+      "step": 7917500
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 3.039277557309394e-05,
+      "loss": 2.1546,
+      "step": 7918000
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 3.0391536986667857e-05,
+      "loss": 2.1356,
+      "step": 7918500
+    },
+    {
+      "epoch": 39.23,
+      "learning_rate": 3.0390298400241774e-05,
+      "loss": 2.1417,
+      "step": 7919000
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 3.038905981381569e-05,
+      "loss": 2.1567,
+      "step": 7919500
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 3.0387821227389608e-05,
+      "loss": 2.1488,
+      "step": 7920000
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 3.0386582640963525e-05,
+      "loss": 2.1396,
+      "step": 7920500
+    },
+    {
+      "epoch": 39.24,
+      "learning_rate": 3.0385344054537442e-05,
+      "loss": 2.1383,
+      "step": 7921000
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 3.0384105468111355e-05,
+      "loss": 2.099,
+      "step": 7921500
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 3.0382866881685272e-05,
+      "loss": 2.141,
+      "step": 7922000
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 3.038162829525919e-05,
+      "loss": 2.1426,
+      "step": 7922500
+    },
+    {
+      "epoch": 39.25,
+      "learning_rate": 3.0380392186005958e-05,
+      "loss": 2.1428,
+      "step": 7923000
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 3.0379153599579875e-05,
+      "loss": 2.1092,
+      "step": 7923500
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 3.037791749032664e-05,
+      "loss": 2.1557,
+      "step": 7924000
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 3.0376678903900557e-05,
+      "loss": 2.1542,
+      "step": 7924500
+    },
+    {
+      "epoch": 39.26,
+      "learning_rate": 3.0375440317474474e-05,
+      "loss": 2.1493,
+      "step": 7925000
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 3.037420173104839e-05,
+      "loss": 2.1659,
+      "step": 7925500
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 3.0372965621795157e-05,
+      "loss": 2.1458,
+      "step": 7926000
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 3.037172951254193e-05,
+      "loss": 2.1533,
+      "step": 7926500
+    },
+    {
+      "epoch": 39.27,
+      "learning_rate": 3.0370490926115846e-05,
+      "loss": 2.1417,
+      "step": 7927000
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 3.036925233968976e-05,
+      "loss": 2.1492,
+      "step": 7927500
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 3.0368013753263673e-05,
+      "loss": 2.1322,
+      "step": 7928000
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 3.036677516683759e-05,
+      "loss": 2.1395,
+      "step": 7928500
+    },
+    {
+      "epoch": 39.28,
+      "learning_rate": 3.0365536580411507e-05,
+      "loss": 2.1302,
+      "step": 7929000
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 3.0364297993985424e-05,
+      "loss": 2.1257,
+      "step": 7929500
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 3.036305940755934e-05,
+      "loss": 2.1455,
+      "step": 7930000
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 3.0361820821133257e-05,
+      "loss": 2.1264,
+      "step": 7930500
+    },
+    {
+      "epoch": 39.29,
+      "learning_rate": 3.0360582234707174e-05,
+      "loss": 2.1652,
+      "step": 7931000
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 3.035934364828109e-05,
+      "loss": 2.1703,
+      "step": 7931500
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 3.0358105061855008e-05,
+      "loss": 2.1462,
+      "step": 7932000
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 3.0356866475428925e-05,
+      "loss": 2.1323,
+      "step": 7932500
+    },
+    {
+      "epoch": 39.3,
+      "learning_rate": 3.035563036617569e-05,
+      "loss": 2.1556,
+      "step": 7933000
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 3.0354391779749607e-05,
+      "loss": 2.1391,
+      "step": 7933500
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 3.0353155670496376e-05,
+      "loss": 2.1458,
+      "step": 7934000
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 3.0351917084070293e-05,
+      "loss": 2.1295,
+      "step": 7934500
+    },
+    {
+      "epoch": 39.31,
+      "learning_rate": 3.0350678497644207e-05,
+      "loss": 2.1523,
+      "step": 7935000
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 3.0349439911218124e-05,
+      "loss": 2.1252,
+      "step": 7935500
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 3.0348203801964896e-05,
+      "loss": 2.1702,
+      "step": 7936000
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 3.0346965215538813e-05,
+      "loss": 2.1224,
+      "step": 7936500
+    },
+    {
+      "epoch": 39.32,
+      "learning_rate": 3.034572910628558e-05,
+      "loss": 2.1537,
+      "step": 7937000
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 3.03444905198595e-05,
+      "loss": 2.1324,
+      "step": 7937500
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 3.0343251933433415e-05,
+      "loss": 2.1567,
+      "step": 7938000
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 3.0342013347007332e-05,
+      "loss": 2.1463,
+      "step": 7938500
+    },
+    {
+      "epoch": 39.33,
+      "learning_rate": 3.0340774760581246e-05,
+      "loss": 2.1471,
+      "step": 7939000
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 3.0339538651328015e-05,
+      "loss": 2.1427,
+      "step": 7939500
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 3.033830006490193e-05,
+      "loss": 2.1476,
+      "step": 7940000
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 3.033706147847585e-05,
+      "loss": 2.1357,
+      "step": 7940500
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 3.0335822892049765e-05,
+      "loss": 2.1364,
+      "step": 7941000
+    },
+    {
+      "epoch": 39.34,
+      "learning_rate": 3.0334584305623682e-05,
+      "loss": 2.1433,
+      "step": 7941500
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 3.0333348196370448e-05,
+      "loss": 2.1451,
+      "step": 7942000
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 3.0332109609944365e-05,
+      "loss": 2.1652,
+      "step": 7942500
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 3.0330871023518282e-05,
+      "loss": 2.16,
+      "step": 7943000
+    },
+    {
+      "epoch": 39.35,
+      "learning_rate": 3.03296324370922e-05,
+      "loss": 2.1619,
+      "step": 7943500
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 3.0328393850666116e-05,
+      "loss": 2.1616,
+      "step": 7944000
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 3.032715774141288e-05,
+      "loss": 2.1725,
+      "step": 7944500
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 3.0325919154986798e-05,
+      "loss": 2.1325,
+      "step": 7945000
+    },
+    {
+      "epoch": 39.36,
+      "learning_rate": 3.0324680568560715e-05,
+      "loss": 2.1277,
+      "step": 7945500
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 3.0323441982134632e-05,
+      "loss": 2.1332,
+      "step": 7946000
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 3.032220339570855e-05,
+      "loss": 2.1359,
+      "step": 7946500
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 3.0320967286455314e-05,
+      "loss": 2.1517,
+      "step": 7947000
+    },
+    {
+      "epoch": 39.37,
+      "learning_rate": 3.031972870002923e-05,
+      "loss": 2.152,
+      "step": 7947500
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 3.0318490113603148e-05,
+      "loss": 2.1653,
+      "step": 7948000
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 3.0317251527177065e-05,
+      "loss": 2.1808,
+      "step": 7948500
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 3.0316012940750982e-05,
+      "loss": 2.1778,
+      "step": 7949000
+    },
+    {
+      "epoch": 39.38,
+      "learning_rate": 3.03147743543249e-05,
+      "loss": 2.1583,
+      "step": 7949500
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 3.0313535767898816e-05,
+      "loss": 2.1499,
+      "step": 7950000
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 3.0312297181472733e-05,
+      "loss": 2.1613,
+      "step": 7950500
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 3.031105859504665e-05,
+      "loss": 2.1439,
+      "step": 7951000
+    },
+    {
+      "epoch": 39.39,
+      "learning_rate": 3.0309820008620566e-05,
+      "loss": 2.1565,
+      "step": 7951500
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 3.0308581422194483e-05,
+      "loss": 2.1586,
+      "step": 7952000
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 3.0307342835768394e-05,
+      "loss": 2.1902,
+      "step": 7952500
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 3.030610424934231e-05,
+      "loss": 2.1583,
+      "step": 7953000
+    },
+    {
+      "epoch": 39.4,
+      "learning_rate": 3.0304865662916227e-05,
+      "loss": 2.1433,
+      "step": 7953500
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 3.0303627076490144e-05,
+      "loss": 2.1525,
+      "step": 7954000
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 3.030238849006406e-05,
+      "loss": 2.1442,
+      "step": 7954500
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 3.0301149903637975e-05,
+      "loss": 2.1515,
+      "step": 7955000
+    },
+    {
+      "epoch": 39.41,
+      "learning_rate": 3.0299911317211892e-05,
+      "loss": 2.146,
+      "step": 7955500
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 3.029867520795866e-05,
+      "loss": 2.1483,
+      "step": 7956000
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 3.0297436621532577e-05,
+      "loss": 2.1599,
+      "step": 7956500
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 3.0296198035106494e-05,
+      "loss": 2.1642,
+      "step": 7957000
+    },
+    {
+      "epoch": 39.42,
+      "learning_rate": 3.0294961925853267e-05,
+      "loss": 2.1511,
+      "step": 7957500
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 3.0293723339427184e-05,
+      "loss": 2.1465,
+      "step": 7958000
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 3.02924847530011e-05,
+      "loss": 2.157,
+      "step": 7958500
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 3.0291248643747866e-05,
+      "loss": 2.1562,
+      "step": 7959000
+    },
+    {
+      "epoch": 39.43,
+      "learning_rate": 3.0290010057321783e-05,
+      "loss": 2.1536,
+      "step": 7959500
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 3.02887714708957e-05,
+      "loss": 2.1747,
+      "step": 7960000
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 3.0287532884469617e-05,
+      "loss": 2.1465,
+      "step": 7960500
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 3.0286296775216382e-05,
+      "loss": 2.1475,
+      "step": 7961000
+    },
+    {
+      "epoch": 39.44,
+      "learning_rate": 3.02850581887903e-05,
+      "loss": 2.1554,
+      "step": 7961500
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 3.0283819602364216e-05,
+      "loss": 2.1414,
+      "step": 7962000
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 3.0282581015938133e-05,
+      "loss": 2.1614,
+      "step": 7962500
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 3.028134242951205e-05,
+      "loss": 2.1447,
+      "step": 7963000
+    },
+    {
+      "epoch": 39.45,
+      "learning_rate": 3.0280103843085967e-05,
+      "loss": 2.1288,
+      "step": 7963500
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 3.0278865256659884e-05,
+      "loss": 2.1451,
+      "step": 7964000
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 3.02776266702338e-05,
+      "loss": 2.1575,
+      "step": 7964500
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 3.027638808380771e-05,
+      "loss": 2.1413,
+      "step": 7965000
+    },
+    {
+      "epoch": 39.46,
+      "learning_rate": 3.0275149497381628e-05,
+      "loss": 2.1849,
+      "step": 7965500
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 3.02739133881284e-05,
+      "loss": 2.1532,
+      "step": 7966000
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 3.0272674801702317e-05,
+      "loss": 2.1684,
+      "step": 7966500
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 3.0271436215276234e-05,
+      "loss": 2.1787,
+      "step": 7967000
+    },
+    {
+      "epoch": 39.47,
+      "learning_rate": 3.027019762885015e-05,
+      "loss": 2.1244,
+      "step": 7967500
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 3.0268961519596916e-05,
+      "loss": 2.1448,
+      "step": 7968000
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 3.0267725410343685e-05,
+      "loss": 2.1404,
+      "step": 7968500
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 3.0266486823917602e-05,
+      "loss": 2.1448,
+      "step": 7969000
+    },
+    {
+      "epoch": 39.48,
+      "learning_rate": 3.0265250714664367e-05,
+      "loss": 2.1466,
+      "step": 7969500
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 3.0264012128238284e-05,
+      "loss": 2.1603,
+      "step": 7970000
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 3.02627735418122e-05,
+      "loss": 2.1393,
+      "step": 7970500
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 3.0261534955386118e-05,
+      "loss": 2.1433,
+      "step": 7971000
+    },
+    {
+      "epoch": 39.49,
+      "learning_rate": 3.0260296368960035e-05,
+      "loss": 2.1232,
+      "step": 7971500
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 3.0259057782533952e-05,
+      "loss": 2.1082,
+      "step": 7972000
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 3.025781919610787e-05,
+      "loss": 2.1548,
+      "step": 7972500
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 3.0256580609681782e-05,
+      "loss": 2.1618,
+      "step": 7973000
+    },
+    {
+      "epoch": 39.5,
+      "learning_rate": 3.02553420232557e-05,
+      "loss": 2.1554,
+      "step": 7973500
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 3.0254103436829616e-05,
+      "loss": 2.1582,
+      "step": 7974000
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 3.0252864850403533e-05,
+      "loss": 2.1767,
+      "step": 7974500
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 3.0251628741150302e-05,
+      "loss": 2.1367,
+      "step": 7975000
+    },
+    {
+      "epoch": 39.51,
+      "learning_rate": 3.025039015472422e-05,
+      "loss": 2.151,
+      "step": 7975500
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 3.0249154045470984e-05,
+      "loss": 2.1421,
+      "step": 7976000
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 3.02479154590449e-05,
+      "loss": 2.1534,
+      "step": 7976500
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 3.0246676872618818e-05,
+      "loss": 2.1454,
+      "step": 7977000
+    },
+    {
+      "epoch": 39.52,
+      "learning_rate": 3.0245438286192735e-05,
+      "loss": 2.1565,
+      "step": 7977500
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 3.0244199699766652e-05,
+      "loss": 2.134,
+      "step": 7978000
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 3.024296111334057e-05,
+      "loss": 2.1578,
+      "step": 7978500
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 3.0241722526914486e-05,
+      "loss": 2.1315,
+      "step": 7979000
+    },
+    {
+      "epoch": 39.53,
+      "learning_rate": 3.02404839404884e-05,
+      "loss": 2.1359,
+      "step": 7979500
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 3.0239245354062316e-05,
+      "loss": 2.1432,
+      "step": 7980000
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 3.0238006767636233e-05,
+      "loss": 2.1618,
+      "step": 7980500
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 3.0236770658383002e-05,
+      "loss": 2.1471,
+      "step": 7981000
+    },
+    {
+      "epoch": 39.54,
+      "learning_rate": 3.023553207195692e-05,
+      "loss": 2.1546,
+      "step": 7981500
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 3.0234293485530836e-05,
+      "loss": 2.1428,
+      "step": 7982000
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 3.0233054899104753e-05,
+      "loss": 2.1344,
+      "step": 7982500
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 3.0231816312678666e-05,
+      "loss": 2.1659,
+      "step": 7983000
+    },
+    {
+      "epoch": 39.55,
+      "learning_rate": 3.0230577726252583e-05,
+      "loss": 2.156,
+      "step": 7983500
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 3.02293391398265e-05,
+      "loss": 2.1345,
+      "step": 7984000
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 3.0228100553400417e-05,
+      "loss": 2.1534,
+      "step": 7984500
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 3.0226861966974334e-05,
+      "loss": 2.1419,
+      "step": 7985000
+    },
+    {
+      "epoch": 39.56,
+      "learning_rate": 3.022562338054825e-05,
+      "loss": 2.1344,
+      "step": 7985500
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 3.022438727129502e-05,
+      "loss": 2.1341,
+      "step": 7986000
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 3.0223148684868933e-05,
+      "loss": 2.132,
+      "step": 7986500
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 3.022191009844285e-05,
+      "loss": 2.1422,
+      "step": 7987000
+    },
+    {
+      "epoch": 39.57,
+      "learning_rate": 3.022067398918962e-05,
+      "loss": 2.1957,
+      "step": 7987500
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 3.0219435402763536e-05,
+      "loss": 2.1574,
+      "step": 7988000
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 3.0218196816337453e-05,
+      "loss": 2.1652,
+      "step": 7988500
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 3.021695822991137e-05,
+      "loss": 2.1417,
+      "step": 7989000
+    },
+    {
+      "epoch": 39.58,
+      "learning_rate": 3.0215719643485287e-05,
+      "loss": 2.1492,
+      "step": 7989500
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 3.02144810570592e-05,
+      "loss": 2.1364,
+      "step": 7990000
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 3.0213242470633117e-05,
+      "loss": 2.1672,
+      "step": 7990500
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 3.0212003884207034e-05,
+      "loss": 2.1569,
+      "step": 7991000
+    },
+    {
+      "epoch": 39.59,
+      "learning_rate": 3.021076529778095e-05,
+      "loss": 2.1703,
+      "step": 7991500
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 3.0209526711354868e-05,
+      "loss": 2.1459,
+      "step": 7992000
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 3.0208288124928785e-05,
+      "loss": 2.164,
+      "step": 7992500
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 3.0207052015675554e-05,
+      "loss": 2.1593,
+      "step": 7993000
+    },
+    {
+      "epoch": 39.6,
+      "learning_rate": 3.0205813429249467e-05,
+      "loss": 2.1342,
+      "step": 7993500
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 3.0204574842823384e-05,
+      "loss": 2.1347,
+      "step": 7994000
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 3.02033362563973e-05,
+      "loss": 2.1442,
+      "step": 7994500
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 3.0202097669971218e-05,
+      "loss": 2.1486,
+      "step": 7995000
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 3.0200859083545135e-05,
+      "loss": 2.1455,
+      "step": 7995500
+    },
+    {
+      "epoch": 39.61,
+      "learning_rate": 3.0199620497119052e-05,
+      "loss": 2.166,
+      "step": 7996000
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 3.0198381910692962e-05,
+      "loss": 2.1218,
+      "step": 7996500
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 3.0197145801439734e-05,
+      "loss": 2.1713,
+      "step": 7997000
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 3.019590721501365e-05,
+      "loss": 2.1567,
+      "step": 7997500
+    },
+    {
+      "epoch": 39.62,
+      "learning_rate": 3.0194668628587568e-05,
+      "loss": 2.132,
+      "step": 7998000
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 3.0193430042161485e-05,
+      "loss": 2.1449,
+      "step": 7998500
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 3.0192191455735402e-05,
+      "loss": 2.1452,
+      "step": 7999000
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 3.0190952869309312e-05,
+      "loss": 2.1455,
+      "step": 7999500
+    },
+    {
+      "epoch": 39.63,
+      "learning_rate": 3.018971428288323e-05,
+      "loss": 2.1394,
+      "step": 8000000
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 3.0188475696457146e-05,
+      "loss": 2.1418,
+      "step": 8000500
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 3.0187237110031063e-05,
+      "loss": 2.1492,
+      "step": 8001000
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 3.0186001000777835e-05,
+      "loss": 2.1505,
+      "step": 8001500
+    },
+    {
+      "epoch": 39.64,
+      "learning_rate": 3.0184762414351752e-05,
+      "loss": 2.1721,
+      "step": 8002000
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 3.0183523827925662e-05,
+      "loss": 2.1447,
+      "step": 8002500
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 3.0182290195845286e-05,
+      "loss": 2.1685,
+      "step": 8003000
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 3.0181051609419203e-05,
+      "loss": 2.1416,
+      "step": 8003500
+    },
+    {
+      "epoch": 39.65,
+      "learning_rate": 3.017981302299312e-05,
+      "loss": 2.1407,
+      "step": 8004000
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 3.0178574436567037e-05,
+      "loss": 2.1782,
+      "step": 8004500
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 3.0177335850140954e-05,
+      "loss": 2.143,
+      "step": 8005000
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 3.017609726371487e-05,
+      "loss": 2.1823,
+      "step": 8005500
+    },
+    {
+      "epoch": 39.66,
+      "learning_rate": 3.0174861154461636e-05,
+      "loss": 2.1475,
+      "step": 8006000
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 3.0173622568035553e-05,
+      "loss": 2.1581,
+      "step": 8006500
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 3.017238645878232e-05,
+      "loss": 2.1574,
+      "step": 8007000
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 3.0171147872356236e-05,
+      "loss": 2.1332,
+      "step": 8007500
+    },
+    {
+      "epoch": 39.67,
+      "learning_rate": 3.0169909285930153e-05,
+      "loss": 2.1649,
+      "step": 8008000
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 3.016867069950407e-05,
+      "loss": 2.1523,
+      "step": 8008500
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 3.0167432113077986e-05,
+      "loss": 2.1674,
+      "step": 8009000
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 3.0166193526651903e-05,
+      "loss": 2.1503,
+      "step": 8009500
+    },
+    {
+      "epoch": 39.68,
+      "learning_rate": 3.016495494022582e-05,
+      "loss": 2.1785,
+      "step": 8010000
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 3.0163716353799737e-05,
+      "loss": 2.1555,
+      "step": 8010500
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 3.0162477767373654e-05,
+      "loss": 2.1361,
+      "step": 8011000
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 3.016123918094757e-05,
+      "loss": 2.1458,
+      "step": 8011500
+    },
+    {
+      "epoch": 39.69,
+      "learning_rate": 3.0160000594521488e-05,
+      "loss": 2.1399,
+      "step": 8012000
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 3.0158762008095405e-05,
+      "loss": 2.1611,
+      "step": 8012500
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 3.0157523421669322e-05,
+      "loss": 2.1584,
+      "step": 8013000
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 3.0156287312416087e-05,
+      "loss": 2.1321,
+      "step": 8013500
+    },
+    {
+      "epoch": 39.7,
+      "learning_rate": 3.0155048725990004e-05,
+      "loss": 2.1513,
+      "step": 8014000
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 3.015381013956392e-05,
+      "loss": 2.1694,
+      "step": 8014500
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 3.0152571553137838e-05,
+      "loss": 2.1494,
+      "step": 8015000
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 3.0151332966711755e-05,
+      "loss": 2.1445,
+      "step": 8015500
+    },
+    {
+      "epoch": 39.71,
+      "learning_rate": 3.0150094380285672e-05,
+      "loss": 2.1644,
+      "step": 8016000
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 3.0148858271032437e-05,
+      "loss": 2.1543,
+      "step": 8016500
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 3.0147619684606354e-05,
+      "loss": 2.1643,
+      "step": 8017000
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 3.014638109818027e-05,
+      "loss": 2.1538,
+      "step": 8017500
+    },
+    {
+      "epoch": 39.72,
+      "learning_rate": 3.0145142511754188e-05,
+      "loss": 2.1518,
+      "step": 8018000
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 3.0143903925328105e-05,
+      "loss": 2.1669,
+      "step": 8018500
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 3.0142665338902022e-05,
+      "loss": 2.1774,
+      "step": 8019000
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 3.014142675247594e-05,
+      "loss": 2.1471,
+      "step": 8019500
+    },
+    {
+      "epoch": 39.73,
+      "learning_rate": 3.0140188166049856e-05,
+      "loss": 2.1726,
+      "step": 8020000
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 3.013895205679662e-05,
+      "loss": 2.1384,
+      "step": 8020500
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 3.0137713470370538e-05,
+      "loss": 2.1756,
+      "step": 8021000
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 3.0136474883944455e-05,
+      "loss": 2.1773,
+      "step": 8021500
+    },
+    {
+      "epoch": 39.74,
+      "learning_rate": 3.0135241251864072e-05,
+      "loss": 2.1633,
+      "step": 8022000
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 3.013400266543799e-05,
+      "loss": 2.1297,
+      "step": 8022500
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 3.013276655618476e-05,
+      "loss": 2.1697,
+      "step": 8023000
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 3.013152796975868e-05,
+      "loss": 2.1987,
+      "step": 8023500
+    },
+    {
+      "epoch": 39.75,
+      "learning_rate": 3.0130289383332595e-05,
+      "loss": 2.1599,
+      "step": 8024000
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 3.012905327407936e-05,
+      "loss": 2.1246,
+      "step": 8024500
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 3.0127814687653278e-05,
+      "loss": 2.1334,
+      "step": 8025000
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 3.0126576101227195e-05,
+      "loss": 2.1671,
+      "step": 8025500
+    },
+    {
+      "epoch": 39.76,
+      "learning_rate": 3.012533751480111e-05,
+      "loss": 2.1581,
+      "step": 8026000
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 3.012409892837503e-05,
+      "loss": 2.1355,
+      "step": 8026500
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 3.0122860341948945e-05,
+      "loss": 2.153,
+      "step": 8027000
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 3.0121621755522862e-05,
+      "loss": 2.1714,
+      "step": 8027500
+    },
+    {
+      "epoch": 39.77,
+      "learning_rate": 3.0120383169096773e-05,
+      "loss": 2.1358,
+      "step": 8028000
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 3.011914458267069e-05,
+      "loss": 2.1435,
+      "step": 8028500
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 3.0117905996244606e-05,
+      "loss": 2.1649,
+      "step": 8029000
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 3.0116667409818523e-05,
+      "loss": 2.1489,
+      "step": 8029500
+    },
+    {
+      "epoch": 39.78,
+      "learning_rate": 3.011542882339244e-05,
+      "loss": 2.1574,
+      "step": 8030000
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 3.0114190236966354e-05,
+      "loss": 2.1635,
+      "step": 8030500
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 3.011295412771313e-05,
+      "loss": 2.1656,
+      "step": 8031000
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 3.011171554128704e-05,
+      "loss": 2.1903,
+      "step": 8031500
+    },
+    {
+      "epoch": 39.79,
+      "learning_rate": 3.0110476954860956e-05,
+      "loss": 2.1556,
+      "step": 8032000
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 3.0109238368434873e-05,
+      "loss": 2.145,
+      "step": 8032500
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 3.0108002259181646e-05,
+      "loss": 2.1641,
+      "step": 8033000
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 3.0106763672755563e-05,
+      "loss": 2.1497,
+      "step": 8033500
+    },
+    {
+      "epoch": 39.8,
+      "learning_rate": 3.010552508632948e-05,
+      "loss": 2.1683,
+      "step": 8034000
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 3.010428649990339e-05,
+      "loss": 2.1542,
+      "step": 8034500
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 3.0103047913477307e-05,
+      "loss": 2.1433,
+      "step": 8035000
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 3.0101809327051223e-05,
+      "loss": 2.1602,
+      "step": 8035500
+    },
+    {
+      "epoch": 39.81,
+      "learning_rate": 3.010057074062514e-05,
+      "loss": 2.1568,
+      "step": 8036000
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 3.0099332154199057e-05,
+      "loss": 2.1549,
+      "step": 8036500
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 3.009809356777297e-05,
+      "loss": 2.1468,
+      "step": 8037000
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 3.0096859935692595e-05,
+      "loss": 2.1577,
+      "step": 8037500
+    },
+    {
+      "epoch": 39.82,
+      "learning_rate": 3.009562382643936e-05,
+      "loss": 2.1663,
+      "step": 8038000
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 3.0094385240013277e-05,
+      "loss": 2.1725,
+      "step": 8038500
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 3.0093146653587194e-05,
+      "loss": 2.1502,
+      "step": 8039000
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 3.009190806716111e-05,
+      "loss": 2.1691,
+      "step": 8039500
+    },
+    {
+      "epoch": 39.83,
+      "learning_rate": 3.0090669480735028e-05,
+      "loss": 2.1382,
+      "step": 8040000
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 3.0089430894308945e-05,
+      "loss": 2.1767,
+      "step": 8040500
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 3.0088192307882862e-05,
+      "loss": 2.1753,
+      "step": 8041000
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 3.008695372145678e-05,
+      "loss": 2.161,
+      "step": 8041500
+    },
+    {
+      "epoch": 39.84,
+      "learning_rate": 3.0085715135030696e-05,
+      "loss": 2.1657,
+      "step": 8042000
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 3.0084476548604613e-05,
+      "loss": 2.1477,
+      "step": 8042500
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 3.008323796217853e-05,
+      "loss": 2.1735,
+      "step": 8043000
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 3.0081999375752447e-05,
+      "loss": 2.1239,
+      "step": 8043500
+    },
+    {
+      "epoch": 39.85,
+      "learning_rate": 3.0080760789326357e-05,
+      "loss": 2.1411,
+      "step": 8044000
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 3.0079522202900274e-05,
+      "loss": 2.1439,
+      "step": 8044500
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 3.0078288570819894e-05,
+      "loss": 2.1604,
+      "step": 8045000
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 3.007704998439381e-05,
+      "loss": 2.1593,
+      "step": 8045500
+    },
+    {
+      "epoch": 39.86,
+      "learning_rate": 3.0075811397967728e-05,
+      "loss": 2.1567,
+      "step": 8046000
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 3.0074572811541645e-05,
+      "loss": 2.1343,
+      "step": 8046500
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 3.0073336702288414e-05,
+      "loss": 2.1458,
+      "step": 8047000
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 3.007209811586233e-05,
+      "loss": 2.1592,
+      "step": 8047500
+    },
+    {
+      "epoch": 39.87,
+      "learning_rate": 3.0070859529436244e-05,
+      "loss": 2.1504,
+      "step": 8048000
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 3.006962094301016e-05,
+      "loss": 2.1612,
+      "step": 8048500
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 3.0068382356584078e-05,
+      "loss": 2.1423,
+      "step": 8049000
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 3.0067143770157995e-05,
+      "loss": 2.1497,
+      "step": 8049500
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 3.0065907660904764e-05,
+      "loss": 2.1624,
+      "step": 8050000
+    },
+    {
+      "epoch": 39.88,
+      "learning_rate": 3.006466907447868e-05,
+      "loss": 2.1358,
+      "step": 8050500
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 3.0063430488052598e-05,
+      "loss": 2.161,
+      "step": 8051000
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 3.006219190162651e-05,
+      "loss": 2.1392,
+      "step": 8051500
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 3.006095331520043e-05,
+      "loss": 2.1358,
+      "step": 8052000
+    },
+    {
+      "epoch": 39.89,
+      "learning_rate": 3.0059714728774345e-05,
+      "loss": 2.1436,
+      "step": 8052500
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 3.0058476142348262e-05,
+      "loss": 2.1518,
+      "step": 8053000
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 3.005723755592218e-05,
+      "loss": 2.1306,
+      "step": 8053500
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 3.0055998969496096e-05,
+      "loss": 2.1684,
+      "step": 8054000
+    },
+    {
+      "epoch": 39.9,
+      "learning_rate": 3.0054762860242865e-05,
+      "loss": 2.1548,
+      "step": 8054500
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 3.005352427381678e-05,
+      "loss": 2.1863,
+      "step": 8055000
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 3.0052285687390695e-05,
+      "loss": 2.1506,
+      "step": 8055500
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 3.0051047100964612e-05,
+      "loss": 2.1482,
+      "step": 8056000
+    },
+    {
+      "epoch": 39.91,
+      "learning_rate": 3.004980851453853e-05,
+      "loss": 2.1735,
+      "step": 8056500
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 3.0048569928112446e-05,
+      "loss": 2.1738,
+      "step": 8057000
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 3.0047331341686363e-05,
+      "loss": 2.1549,
+      "step": 8057500
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 3.004609275526028e-05,
+      "loss": 2.1322,
+      "step": 8058000
+    },
+    {
+      "epoch": 39.92,
+      "learning_rate": 3.0044854168834197e-05,
+      "loss": 2.1524,
+      "step": 8058500
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 3.0043615582408114e-05,
+      "loss": 2.141,
+      "step": 8059000
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 3.0042376995982024e-05,
+      "loss": 2.1674,
+      "step": 8059500
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 3.004113840955594e-05,
+      "loss": 2.1547,
+      "step": 8060000
+    },
+    {
+      "epoch": 39.93,
+      "learning_rate": 3.0039899823129858e-05,
+      "loss": 2.1545,
+      "step": 8060500
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 3.003866371387663e-05,
+      "loss": 2.166,
+      "step": 8061000
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 3.0037425127450547e-05,
+      "loss": 2.1595,
+      "step": 8061500
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 3.0036186541024464e-05,
+      "loss": 2.1371,
+      "step": 8062000
+    },
+    {
+      "epoch": 39.94,
+      "learning_rate": 3.0034947954598374e-05,
+      "loss": 2.1685,
+      "step": 8062500
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 3.0033711845345146e-05,
+      "loss": 2.1529,
+      "step": 8063000
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 3.0032475736091915e-05,
+      "loss": 2.1464,
+      "step": 8063500
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 3.0031237149665832e-05,
+      "loss": 2.1639,
+      "step": 8064000
+    },
+    {
+      "epoch": 39.95,
+      "learning_rate": 3.0030001040412597e-05,
+      "loss": 2.1509,
+      "step": 8064500
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 3.0028762453986514e-05,
+      "loss": 2.1675,
+      "step": 8065000
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 3.002752386756043e-05,
+      "loss": 2.1495,
+      "step": 8065500
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 3.0026285281134348e-05,
+      "loss": 2.1633,
+      "step": 8066000
+    },
+    {
+      "epoch": 39.96,
+      "learning_rate": 3.0025046694708265e-05,
+      "loss": 2.1319,
+      "step": 8066500
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 3.0023808108282182e-05,
+      "loss": 2.1646,
+      "step": 8067000
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 3.00225695218561e-05,
+      "loss": 2.1526,
+      "step": 8067500
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 3.0021330935430016e-05,
+      "loss": 2.1123,
+      "step": 8068000
+    },
+    {
+      "epoch": 39.97,
+      "learning_rate": 3.002009482617678e-05,
+      "loss": 2.1465,
+      "step": 8068500
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 3.0018858716923553e-05,
+      "loss": 2.1841,
+      "step": 8069000
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 3.0017620130497464e-05,
+      "loss": 2.1716,
+      "step": 8069500
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 3.001638154407138e-05,
+      "loss": 2.17,
+      "step": 8070000
+    },
+    {
+      "epoch": 39.98,
+      "learning_rate": 3.0015142957645298e-05,
+      "loss": 2.1723,
+      "step": 8070500
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 3.0013904371219214e-05,
+      "loss": 2.1409,
+      "step": 8071000
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 3.001266578479313e-05,
+      "loss": 2.1666,
+      "step": 8071500
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 3.0011427198367048e-05,
+      "loss": 2.1527,
+      "step": 8072000
+    },
+    {
+      "epoch": 39.99,
+      "learning_rate": 3.0010188611940965e-05,
+      "loss": 2.1557,
+      "step": 8072500
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 3.0008950025514882e-05,
+      "loss": 2.1458,
+      "step": 8073000
+    },
+    {
+      "epoch": 40.0,
+      "learning_rate": 3.00077114390888e-05,
+      "loss": 2.1418,
+      "step": 8073500
+    },
+    {
+      "epoch": 40.0,
+      "eval_accuracy": 0.6648354250857901,
+      "eval_accuracy_mlm": 0.6219327105976565,
+      "eval_accuracy_nsp": 0.8670649006310819,
+      "eval_loss": 2.2923583984375,
+      "eval_runtime": 146.917,
+      "eval_samples_per_second": 1735.395,
+      "eval_steps_per_second": 72.313,
+      "step": 8073720
     }
   ],
   "max_steps": 20184300,
   "num_train_epochs": 100,
-  "total_flos": 8.884340001928284e+18,
+  "total_flos": 1.0452273947820423e+19,
   "trial_name": null,
   "trial_params": null
 }