diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6652 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 22090,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.997102761430512e-05,
+      "loss": 2.8287,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.9934812132186514e-05,
+      "loss": 3.0376,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.989859665006791e-05,
+      "loss": 2.8324,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.98623811679493e-05,
+      "loss": 2.7685,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.98261656858307e-05,
+      "loss": 2.5558,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.978995020371209e-05,
+      "loss": 2.5886,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9753734721593485e-05,
+      "loss": 2.6756,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.971751923947488e-05,
+      "loss": 2.5154,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.968130375735627e-05,
+      "loss": 2.5298,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.964508827523767e-05,
+      "loss": 2.227,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.960887279311906e-05,
+      "loss": 2.3811,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.9572657311000456e-05,
+      "loss": 2.3438,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.953644182888185e-05,
+      "loss": 2.2362,
+      "step": 260
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.950022634676324e-05,
+      "loss": 2.2191,
+      "step": 280
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.946401086464464e-05,
+      "loss": 2.3578,
+      "step": 300
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.942779538252604e-05,
+      "loss": 2.3118,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.939157990040743e-05,
+      "loss": 2.3615,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9355364418288824e-05,
+      "loss": 2.2838,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.9319148936170214e-05,
+      "loss": 2.2169,
+      "step": 380
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.928293345405161e-05,
+      "loss": 2.3863,
+      "step": 400
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.924671797193301e-05,
+      "loss": 2.2928,
+      "step": 420
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.92105024898144e-05,
+      "loss": 2.2238,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9174287007695796e-05,
+      "loss": 2.1976,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.9138071525577186e-05,
+      "loss": 2.329,
+      "step": 480
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.910185604345858e-05,
+      "loss": 2.1273,
+      "step": 500
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.906564056133998e-05,
+      "loss": 2.2847,
+      "step": 520
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.902942507922137e-05,
+      "loss": 2.4067,
+      "step": 540
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.899320959710277e-05,
+      "loss": 2.2402,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.895699411498416e-05,
+      "loss": 2.2093,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.8920778632865554e-05,
+      "loss": 2.3391,
+      "step": 600
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.888456315074695e-05,
+      "loss": 2.2032,
+      "step": 620
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.884834766862834e-05,
+      "loss": 2.21,
+      "step": 640
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.881213218650973e-05,
+      "loss": 2.2544,
+      "step": 660
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8775916704391135e-05,
+      "loss": 2.4552,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.8739701222272525e-05,
+      "loss": 2.2468,
+      "step": 700
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.870348574015392e-05,
+      "loss": 2.1652,
+      "step": 720
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.866727025803531e-05,
+      "loss": 2.1679,
+      "step": 740
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.86310547759167e-05,
+      "loss": 2.1369,
+      "step": 760
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.8594839293798106e-05,
+      "loss": 2.2805,
+      "step": 780
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.8558623811679496e-05,
+      "loss": 2.1412,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 3.8522408329560886e-05,
+      "loss": 2.1467,
+      "step": 820
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.8486192847442283e-05,
+      "loss": 2.1684,
+      "step": 840
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.844997736532368e-05,
+      "loss": 2.1887,
+      "step": 860
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.841376188320508e-05,
+      "loss": 2.2834,
+      "step": 880
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.837754640108647e-05,
+      "loss": 2.1439,
+      "step": 900
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 3.834133091896786e-05,
+      "loss": 2.1248,
+      "step": 920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.830511543684926e-05,
+      "loss": 2.347,
+      "step": 940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.826889995473065e-05,
+      "loss": 2.1155,
+      "step": 960
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.823268447261204e-05,
+      "loss": 2.3619,
+      "step": 980
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.819646899049344e-05,
+      "loss": 2.1893,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.816025350837483e-05,
+      "loss": 2.1301,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 3.812403802625623e-05,
+      "loss": 2.0641,
+      "step": 1040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.808782254413762e-05,
+      "loss": 2.1476,
+      "step": 1060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.805160706201901e-05,
+      "loss": 2.2108,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.801539157990041e-05,
+      "loss": 2.2559,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.797917609778181e-05,
+      "loss": 2.1753,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 3.79429606156632e-05,
+      "loss": 2.1629,
+      "step": 1140
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7906745133544594e-05,
+      "loss": 2.2633,
+      "step": 1160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7870529651425984e-05,
+      "loss": 2.2389,
+      "step": 1180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.783431416930738e-05,
+      "loss": 2.1522,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.779809868718878e-05,
+      "loss": 2.2949,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.776188320507017e-05,
+      "loss": 2.1695,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 3.7725667722951565e-05,
+      "loss": 2.3722,
+      "step": 1260
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.7689452240832955e-05,
+      "loss": 2.2207,
+      "step": 1280
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.765323675871435e-05,
+      "loss": 2.1472,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.761702127659575e-05,
+      "loss": 2.1495,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.7582616568583075e-05,
+      "loss": 2.1473,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.7546401086464466e-05,
+      "loss": 1.9812,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 3.7510185604345856e-05,
+      "loss": 2.1141,
+      "step": 1380
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.747397012222726e-05,
+      "loss": 2.1986,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.743775464010865e-05,
+      "loss": 2.1415,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.7401539157990047e-05,
+      "loss": 2.2091,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.736532367587144e-05,
+      "loss": 2.1455,
+      "step": 1460
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 3.7329108193752834e-05,
+      "loss": 2.163,
+      "step": 1480
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.729289271163423e-05,
+      "loss": 2.1876,
+      "step": 1500
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.725848800362155e-05,
+      "loss": 2.1202,
+      "step": 1520
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.722227252150295e-05,
+      "loss": 2.0809,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.718605703938434e-05,
+      "loss": 2.3098,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.7149841557265734e-05,
+      "loss": 2.2058,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 3.711362607514713e-05,
+      "loss": 2.1712,
+      "step": 1600
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.707741059302852e-05,
+      "loss": 2.3473,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.704119511090992e-05,
+      "loss": 2.1506,
+      "step": 1640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.7004979628791315e-05,
+      "loss": 2.1957,
+      "step": 1660
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.6968764146672705e-05,
+      "loss": 2.2387,
+      "step": 1680
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 3.69325486645541e-05,
+      "loss": 2.2433,
+      "step": 1700
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.689633318243549e-05,
+      "loss": 2.1396,
+      "step": 1720
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.686011770031689e-05,
+      "loss": 2.2482,
+      "step": 1740
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.6823902218198286e-05,
+      "loss": 2.1396,
+      "step": 1760
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.6787686736079676e-05,
+      "loss": 2.1512,
+      "step": 1780
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.6751471253961073e-05,
+      "loss": 2.2568,
+      "step": 1800
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 3.6715255771842464e-05,
+      "loss": 2.2791,
+      "step": 1820
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.667904028972386e-05,
+      "loss": 2.1336,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.664282480760526e-05,
+      "loss": 2.0302,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.660660932548665e-05,
+      "loss": 2.3007,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.6570393843368045e-05,
+      "loss": 2.1722,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 3.6534178361249435e-05,
+      "loss": 2.1787,
+      "step": 1920
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.649796287913083e-05,
+      "loss": 2.0759,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.646174739701223e-05,
+      "loss": 2.2239,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.642553191489362e-05,
+      "loss": 2.2151,
+      "step": 1980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.6389316432775016e-05,
+      "loss": 2.0375,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.635310095065641e-05,
+      "loss": 2.1542,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 3.63168854685378e-05,
+      "loss": 2.1623,
+      "step": 2040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.62806699864192e-05,
+      "loss": 2.1761,
+      "step": 2060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.624626527840652e-05,
+      "loss": 2.2656,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.6210049796287916e-05,
+      "loss": 2.2158,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.617383431416931e-05,
+      "loss": 2.1538,
+      "step": 2120
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 3.61376188320507e-05,
+      "loss": 2.0148,
+      "step": 2140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.61014033499321e-05,
+      "loss": 2.181,
+      "step": 2160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.606518786781349e-05,
+      "loss": 2.1822,
+      "step": 2180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.602897238569489e-05,
+      "loss": 2.1637,
+      "step": 2200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.5992756903576284e-05,
+      "loss": 2.1395,
+      "step": 2220
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.5956541421457675e-05,
+      "loss": 2.1034,
+      "step": 2240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 3.592032593933907e-05,
+      "loss": 2.1595,
+      "step": 2260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.588411045722047e-05,
+      "loss": 2.0379,
+      "step": 2280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.584789497510186e-05,
+      "loss": 2.1031,
+      "step": 2300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.5811679492983256e-05,
+      "loss": 2.2266,
+      "step": 2320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.5775464010864646e-05,
+      "loss": 2.0986,
+      "step": 2340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 3.573924852874604e-05,
+      "loss": 2.1674,
+      "step": 2360
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.570303304662744e-05,
+      "loss": 2.3103,
+      "step": 2380
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.566681756450883e-05,
+      "loss": 2.1794,
+      "step": 2400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.563060208239023e-05,
+      "loss": 2.2558,
+      "step": 2420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.559438660027162e-05,
+      "loss": 2.1224,
+      "step": 2440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.5558171118153014e-05,
+      "loss": 2.0814,
+      "step": 2460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 3.552195563603441e-05,
+      "loss": 2.2007,
+      "step": 2480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.54857401539158e-05,
+      "loss": 2.0045,
+      "step": 2500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.54495246717972e-05,
+      "loss": 2.0878,
+      "step": 2520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.541330918967859e-05,
+      "loss": 2.152,
+      "step": 2540
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.5377093707559985e-05,
+      "loss": 2.0813,
+      "step": 2560
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 3.534087822544138e-05,
+      "loss": 2.1433,
+      "step": 2580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.530466274332277e-05,
+      "loss": 2.2122,
+      "step": 2600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.526844726120417e-05,
+      "loss": 2.1284,
+      "step": 2620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5232231779085566e-05,
+      "loss": 2.1024,
+      "step": 2640
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5196016296966956e-05,
+      "loss": 2.0446,
+      "step": 2660
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.515980081484835e-05,
+      "loss": 2.1051,
+      "step": 2680
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 3.5123585332729743e-05,
+      "loss": 2.0625,
+      "step": 2700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.508736985061114e-05,
+      "loss": 2.0042,
+      "step": 2720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.505115436849254e-05,
+      "loss": 2.051,
+      "step": 2740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.501493888637393e-05,
+      "loss": 2.1632,
+      "step": 2760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4978723404255325e-05,
+      "loss": 2.1068,
+      "step": 2780
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 3.4942507922136715e-05,
+      "loss": 2.1309,
+      "step": 2800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.490629244001811e-05,
+      "loss": 2.1194,
+      "step": 2820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.487007695789951e-05,
+      "loss": 2.2082,
+      "step": 2840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.48338614757809e-05,
+      "loss": 2.1216,
+      "step": 2860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4797645993662296e-05,
+      "loss": 2.1243,
+      "step": 2880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.4761430511543686e-05,
+      "loss": 2.0564,
+      "step": 2900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 3.472521502942508e-05,
+      "loss": 2.1302,
+      "step": 2920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.468899954730648e-05,
+      "loss": 2.0772,
+      "step": 2940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.465278406518787e-05,
+      "loss": 2.0272,
+      "step": 2960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.461656858306926e-05,
+      "loss": 2.2952,
+      "step": 2980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4580353100950664e-05,
+      "loss": 2.0506,
+      "step": 3000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 3.4544137618832054e-05,
+      "loss": 2.068,
+      "step": 3020
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4507922136713444e-05,
+      "loss": 2.2147,
+      "step": 3040
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.447170665459484e-05,
+      "loss": 2.1733,
+      "step": 3060
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.443549117247623e-05,
+      "loss": 2.2375,
+      "step": 3080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4399275690357635e-05,
+      "loss": 2.0225,
+      "step": 3100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4363060208239025e-05,
+      "loss": 2.2714,
+      "step": 3120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 3.4326844726120415e-05,
+      "loss": 2.2056,
+      "step": 3140
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.429062924400181e-05,
+      "loss": 2.0419,
+      "step": 3160
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.425441376188321e-05,
+      "loss": 2.0376,
+      "step": 3180
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.42181982797646e-05,
+      "loss": 2.0509,
+      "step": 3200
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.4181982797645996e-05,
+      "loss": 2.1717,
+      "step": 3220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 3.414576731552739e-05,
+      "loss": 2.1081,
+      "step": 3240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.410955183340879e-05,
+      "loss": 2.052,
+      "step": 3260
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.407333635129018e-05,
+      "loss": 2.051,
+      "step": 3280
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.403712086917157e-05,
+      "loss": 2.2021,
+      "step": 3300
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.400090538705297e-05,
+      "loss": 2.164,
+      "step": 3320
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.396468990493436e-05,
+      "loss": 2.0914,
+      "step": 3340
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 3.393028519692169e-05,
+      "loss": 2.3347,
+      "step": 3360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.389406971480308e-05,
+      "loss": 2.094,
+      "step": 3380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.385785423268448e-05,
+      "loss": 2.2954,
+      "step": 3400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.382163875056587e-05,
+      "loss": 2.0015,
+      "step": 3420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.3785423268447265e-05,
+      "loss": 2.1794,
+      "step": 3440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 3.374920778632866e-05,
+      "loss": 2.3331,
+      "step": 3460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.371299230421005e-05,
+      "loss": 2.0371,
+      "step": 3480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.367677682209145e-05,
+      "loss": 2.1625,
+      "step": 3500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.364056133997284e-05,
+      "loss": 2.2299,
+      "step": 3520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.3604345857854236e-05,
+      "loss": 2.1058,
+      "step": 3540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.356813037573563e-05,
+      "loss": 2.0241,
+      "step": 3560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 3.353191489361702e-05,
+      "loss": 2.1641,
+      "step": 3580
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3495699411498414e-05,
+      "loss": 1.976,
+      "step": 3600
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.345948392937982e-05,
+      "loss": 2.1106,
+      "step": 3620
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.342326844726121e-05,
+      "loss": 2.157,
+      "step": 3640
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3387052965142604e-05,
+      "loss": 2.1702,
+      "step": 3660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3350837483023995e-05,
+      "loss": 2.113,
+      "step": 3680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 3.3314622000905385e-05,
+      "loss": 2.112,
+      "step": 3700
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.327840651878679e-05,
+      "loss": 2.1211,
+      "step": 3720
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.324219103666818e-05,
+      "loss": 2.2343,
+      "step": 3740
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.320597555454957e-05,
+      "loss": 2.124,
+      "step": 3760
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.3169760072430966e-05,
+      "loss": 1.9935,
+      "step": 3780
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 3.313354459031236e-05,
+      "loss": 2.1069,
+      "step": 3800
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.309732910819376e-05,
+      "loss": 2.0229,
+      "step": 3820
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.306111362607515e-05,
+      "loss": 2.1129,
+      "step": 3840
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.302489814395654e-05,
+      "loss": 2.1617,
+      "step": 3860
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.298868266183794e-05,
+      "loss": 2.0416,
+      "step": 3880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2952467179719334e-05,
+      "loss": 2.0727,
+      "step": 3900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 3.2916251697600724e-05,
+      "loss": 2.0983,
+      "step": 3920
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.288003621548212e-05,
+      "loss": 1.921,
+      "step": 3940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.284382073336351e-05,
+      "loss": 2.0128,
+      "step": 3960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2807605251244915e-05,
+      "loss": 2.1038,
+      "step": 3980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2771389769126305e-05,
+      "loss": 2.1271,
+      "step": 4000
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 3.2735174287007695e-05,
+      "loss": 2.0482,
+      "step": 4020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.269895880488909e-05,
+      "loss": 2.1031,
+      "step": 4040
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.266274332277048e-05,
+      "loss": 2.0018,
+      "step": 4060
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.262652784065188e-05,
+      "loss": 1.9006,
+      "step": 4080
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2590312358533276e-05,
+      "loss": 2.0833,
+      "step": 4100
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2554096876414667e-05,
+      "loss": 2.1511,
+      "step": 4120
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 3.2517881394296063e-05,
+      "loss": 2.1072,
+      "step": 4140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.248166591217746e-05,
+      "loss": 2.1492,
+      "step": 4160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.244545043005885e-05,
+      "loss": 1.9445,
+      "step": 4180
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.240923494794025e-05,
+      "loss": 1.9307,
+      "step": 4200
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.237301946582164e-05,
+      "loss": 2.0911,
+      "step": 4220
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 3.2336803983703035e-05,
+      "loss": 2.2143,
+      "step": 4240
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.230058850158443e-05,
+      "loss": 2.1662,
+      "step": 4260
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.226437301946582e-05,
+      "loss": 2.1353,
+      "step": 4280
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.222815753734722e-05,
+      "loss": 2.1666,
+      "step": 4300
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.219194205522861e-05,
+      "loss": 2.1496,
+      "step": 4320
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.2155726573110006e-05,
+      "loss": 2.189,
+      "step": 4340
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 3.21195110909914e-05,
+      "loss": 1.9424,
+      "step": 4360
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.208329560887279e-05,
+      "loss": 2.0531,
+      "step": 4380
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.204708012675419e-05,
+      "loss": 1.9528,
+      "step": 4400
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.201086464463559e-05,
+      "loss": 2.1361,
+      "step": 4420
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.197464916251698e-05,
+      "loss": 1.9907,
+      "step": 4440
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 3.1938433680398374e-05,
+      "loss": 1.9543,
+      "step": 4460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.190402897238569e-05,
+      "loss": 2.144,
+      "step": 4480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.186781349026709e-05,
+      "loss": 2.0422,
+      "step": 4500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.183159800814849e-05,
+      "loss": 2.187,
+      "step": 4520
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1795382526029884e-05,
+      "loss": 1.9917,
+      "step": 4540
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1759167043911274e-05,
+      "loss": 2.0603,
+      "step": 4560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 3.1722951561792665e-05,
+      "loss": 2.2549,
+      "step": 4580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.168673607967407e-05,
+      "loss": 2.0071,
+      "step": 4600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.165052059755546e-05,
+      "loss": 2.107,
+      "step": 4620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.161430511543685e-05,
+      "loss": 2.1349,
+      "step": 4640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1578089633318246e-05,
+      "loss": 2.3,
+      "step": 4660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 3.1541874151199636e-05,
+      "loss": 2.0421,
+      "step": 4680
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.150565866908104e-05,
+      "loss": 2.0764,
+      "step": 4700
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.146944318696243e-05,
+      "loss": 1.9746,
+      "step": 4720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.143322770484382e-05,
+      "loss": 2.0114,
+      "step": 4740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.139701222272522e-05,
+      "loss": 2.0057,
+      "step": 4760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.1360796740606614e-05,
+      "loss": 2.066,
+      "step": 4780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 3.132639203259394e-05,
+      "loss": 2.0076,
+      "step": 4800
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.129017655047533e-05,
+      "loss": 2.2607,
+      "step": 4820
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.125396106835673e-05,
+      "loss": 2.0034,
+      "step": 4840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.121774558623812e-05,
+      "loss": 2.1698,
+      "step": 4860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.1181530104119514e-05,
+      "loss": 2.0996,
+      "step": 4880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 3.114531462200091e-05,
+      "loss": 2.0875,
+      "step": 4900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.11090991398823e-05,
+      "loss": 2.1561,
+      "step": 4920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.10728836577637e-05,
+      "loss": 2.1346,
+      "step": 4940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.1036668175645095e-05,
+      "loss": 2.0495,
+      "step": 4960
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.1000452693526485e-05,
+      "loss": 2.0495,
+      "step": 4980
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.096423721140788e-05,
+      "loss": 2.1178,
+      "step": 5000
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 3.092802172928927e-05,
+      "loss": 2.0008,
+      "step": 5020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.089180624717066e-05,
+      "loss": 2.021,
+      "step": 5040
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.0855590765052066e-05,
+      "loss": 2.005,
+      "step": 5060
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.0819375282933457e-05,
+      "loss": 2.1023,
+      "step": 5080
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.0783159800814853e-05,
+      "loss": 2.0493,
+      "step": 5100
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 3.0746944318696244e-05,
+      "loss": 2.0882,
+      "step": 5120
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.071072883657764e-05,
+      "loss": 2.1189,
+      "step": 5140
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.067451335445904e-05,
+      "loss": 2.1325,
+      "step": 5160
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.063829787234043e-05,
+      "loss": 2.0306,
+      "step": 5180
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.060208239022182e-05,
+      "loss": 2.1646,
+      "step": 5200
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.0565866908103215e-05,
+      "loss": 2.0843,
+      "step": 5220
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 3.052965142598461e-05,
+      "loss": 1.968,
+      "step": 5240
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0493435943866005e-05,
+      "loss": 2.2515,
+      "step": 5260
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.04572204617474e-05,
+      "loss": 2.1048,
+      "step": 5280
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0421004979628793e-05,
+      "loss": 2.0571,
+      "step": 5300
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.038478949751019e-05,
+      "loss": 2.1384,
+      "step": 5320
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 3.0348574015391583e-05,
+      "loss": 2.253,
+      "step": 5340
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0312358533272977e-05,
+      "loss": 2.0273,
+      "step": 5360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.027614305115437e-05,
+      "loss": 2.031,
+      "step": 5380
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0239927569035767e-05,
+      "loss": 2.0828,
+      "step": 5400
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.020371208691716e-05,
+      "loss": 2.1536,
+      "step": 5420
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0167496604798554e-05,
+      "loss": 2.0289,
+      "step": 5440
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 3.0131281122679948e-05,
+      "loss": 2.1677,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.009506564056134e-05,
+      "loss": 2.0081,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.005885015844274e-05,
+      "loss": 1.9433,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 3.0022634676324132e-05,
+      "loss": 1.9952,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.9986419194205525e-05,
+      "loss": 1.8982,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 2.995020371208692e-05,
+      "loss": 2.1173,
+      "step": 5560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9913988229968316e-05,
+      "loss": 2.0235,
+      "step": 5580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.987777274784971e-05,
+      "loss": 2.1329,
+      "step": 5600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9841557265731103e-05,
+      "loss": 2.1825,
+      "step": 5620
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9805341783612497e-05,
+      "loss": 2.1083,
+      "step": 5640
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.976912630149389e-05,
+      "loss": 2.1486,
+      "step": 5660
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 2.9732910819375287e-05,
+      "loss": 2.06,
+      "step": 5680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.969669533725668e-05,
+      "loss": 2.1061,
+      "step": 5700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9660479855138074e-05,
+      "loss": 1.9686,
+      "step": 5720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9624264373019468e-05,
+      "loss": 1.9891,
+      "step": 5740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.9588048890900865e-05,
+      "loss": 2.0328,
+      "step": 5760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 2.955183340878226e-05,
+      "loss": 2.0089,
+      "step": 5780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.951742870076958e-05,
+      "loss": 2.194,
+      "step": 5800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9481213218650975e-05,
+      "loss": 2.0237,
+      "step": 5820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9444997736532368e-05,
+      "loss": 2.0917,
+      "step": 5840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9410593028519698e-05,
+      "loss": 2.1599,
+      "step": 5860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.9374377546401088e-05,
+      "loss": 2.0168,
+      "step": 5880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 2.933816206428248e-05,
+      "loss": 2.0528,
+      "step": 5900
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9301946582163875e-05,
+      "loss": 2.0669,
+      "step": 5920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9265731100045275e-05,
+      "loss": 2.0534,
+      "step": 5940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9229515617926666e-05,
+      "loss": 2.1274,
+      "step": 5960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.919511090991399e-05,
+      "loss": 2.2521,
+      "step": 5980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 2.9158895427795385e-05,
+      "loss": 2.1445,
+      "step": 6000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.912267994567678e-05,
+      "loss": 2.1666,
+      "step": 6020
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9086464463558176e-05,
+      "loss": 2.0054,
+      "step": 6040
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.905024898143957e-05,
+      "loss": 1.8451,
+      "step": 6060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.9014033499320963e-05,
+      "loss": 2.1416,
+      "step": 6080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8977818017202356e-05,
+      "loss": 2.0161,
+      "step": 6100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 2.8941602535083753e-05,
+      "loss": 2.0959,
+      "step": 6120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8905387052965147e-05,
+      "loss": 2.0974,
+      "step": 6140
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.886917157084654e-05,
+      "loss": 2.1888,
+      "step": 6160
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8832956088727934e-05,
+      "loss": 2.2963,
+      "step": 6180
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8796740606609324e-05,
+      "loss": 2.1439,
+      "step": 6200
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8760525124490725e-05,
+      "loss": 2.0336,
+      "step": 6220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 2.8726120416478047e-05,
+      "loss": 1.9481,
+      "step": 6240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.868990493435944e-05,
+      "loss": 2.1295,
+      "step": 6260
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8653689452240834e-05,
+      "loss": 2.0927,
+      "step": 6280
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.861747397012223e-05,
+      "loss": 2.1297,
+      "step": 6300
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.8581258488003625e-05,
+      "loss": 2.0766,
+      "step": 6320
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 2.854504300588502e-05,
+      "loss": 2.0061,
+      "step": 6340
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8508827523766412e-05,
+      "loss": 2.0651,
+      "step": 6360
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8472612041647806e-05,
+      "loss": 2.0351,
+      "step": 6380
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8436396559529203e-05,
+      "loss": 2.035,
+      "step": 6400
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8400181077410596e-05,
+      "loss": 2.14,
+      "step": 6420
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.836396559529199e-05,
+      "loss": 1.9856,
+      "step": 6440
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 2.8327750113173383e-05,
+      "loss": 2.1651,
+      "step": 6460
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.829153463105478e-05,
+      "loss": 2.1174,
+      "step": 6480
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8255319148936174e-05,
+      "loss": 2.2192,
+      "step": 6500
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8219103666817567e-05,
+      "loss": 1.9739,
+      "step": 6520
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.818288818469896e-05,
+      "loss": 2.1656,
+      "step": 6540
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 2.8146672702580354e-05,
+      "loss": 2.1554,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.811045722046175e-05,
+      "loss": 2.1915,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8074241738343145e-05,
+      "loss": 2.1186,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.803802625622454e-05,
+      "loss": 2.0831,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.8001810774105932e-05,
+      "loss": 2.0134,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.796559529198733e-05,
+      "loss": 1.9895,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 2.7929379809868723e-05,
+      "loss": 2.132,
+      "step": 6680
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.7893164327750116e-05,
+      "loss": 2.1022,
+      "step": 6700
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.785694884563151e-05,
+      "loss": 2.216,
+      "step": 6720
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.7820733363512907e-05,
+      "loss": 1.864,
+      "step": 6740
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.77845178813943e-05,
+      "loss": 1.9873,
+      "step": 6760
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 2.7748302399275694e-05,
+      "loss": 1.9765,
+      "step": 6780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.7712086917157087e-05,
+      "loss": 2.111,
+      "step": 6800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.767587143503848e-05,
+      "loss": 2.0179,
+      "step": 6820
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.7639655952919878e-05,
+      "loss": 2.0464,
+      "step": 6840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.760344047080127e-05,
+      "loss": 2.065,
+      "step": 6860
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.7567224988682665e-05,
+      "loss": 2.0962,
+      "step": 6880
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 2.753100950656406e-05,
+      "loss": 2.2815,
+      "step": 6900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7494794024445456e-05,
+      "loss": 2.1114,
+      "step": 6920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.745857854232685e-05,
+      "loss": 1.922,
+      "step": 6940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7422363060208243e-05,
+      "loss": 2.1107,
+      "step": 6960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7386147578089636e-05,
+      "loss": 1.9859,
+      "step": 6980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 2.7349932095971026e-05,
+      "loss": 2.0425,
+      "step": 7000
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7313716613852427e-05,
+      "loss": 2.0704,
+      "step": 7020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.727750113173382e-05,
+      "loss": 2.1113,
+      "step": 7040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7241285649615214e-05,
+      "loss": 2.0124,
+      "step": 7060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7205070167496604e-05,
+      "loss": 1.9183,
+      "step": 7080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7168854685378004e-05,
+      "loss": 2.021,
+      "step": 7100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.7132639203259398e-05,
+      "loss": 2.0165,
+      "step": 7120
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.709642372114079e-05,
+      "loss": 2.1237,
+      "step": 7140
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7060208239022182e-05,
+      "loss": 1.9857,
+      "step": 7160
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7023992756903575e-05,
+      "loss": 2.0966,
+      "step": 7180
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.6987777274784976e-05,
+      "loss": 2.2455,
+      "step": 7200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.695156179266637e-05,
+      "loss": 2.0672,
+      "step": 7220
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.691534631054776e-05,
+      "loss": 2.0218,
+      "step": 7240
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6879130828429153e-05,
+      "loss": 2.087,
+      "step": 7260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6842915346310553e-05,
+      "loss": 1.9421,
+      "step": 7280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6806699864191943e-05,
+      "loss": 2.0234,
+      "step": 7300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6770484382073337e-05,
+      "loss": 2.0911,
+      "step": 7320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.673426889995473e-05,
+      "loss": 2.0759,
+      "step": 7340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6698053417836124e-05,
+      "loss": 2.057,
+      "step": 7360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.666183793571752e-05,
+      "loss": 2.0833,
+      "step": 7380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6625622453598915e-05,
+      "loss": 1.9881,
+      "step": 7400
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6589406971480308e-05,
+      "loss": 2.0511,
+      "step": 7420
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6553191489361702e-05,
+      "loss": 2.0798,
+      "step": 7440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.65169760072431e-05,
+      "loss": 2.0665,
+      "step": 7460
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6480760525124492e-05,
+      "loss": 2.0543,
+      "step": 7480
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6444545043005886e-05,
+      "loss": 2.0709,
+      "step": 7500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.640832956088728e-05,
+      "loss": 2.0094,
+      "step": 7520
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.6372114078768673e-05,
+      "loss": 2.2191,
+      "step": 7540
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.633589859665007e-05,
+      "loss": 2.0337,
+      "step": 7560
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6299683114531464e-05,
+      "loss": 2.197,
+      "step": 7580
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6263467632412857e-05,
+      "loss": 2.1088,
+      "step": 7600
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.622725215029425e-05,
+      "loss": 2.2008,
+      "step": 7620
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.6191036668175648e-05,
+      "loss": 2.1119,
+      "step": 7640
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.615482118605704e-05,
+      "loss": 2.0474,
+      "step": 7660
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6118605703938435e-05,
+      "loss": 2.1261,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6082390221819828e-05,
+      "loss": 1.9889,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.6046174739701225e-05,
+      "loss": 2.1041,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.601177003168855e-05,
+      "loss": 2.0248,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.5975554549569945e-05,
+      "loss": 2.2268,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.593933906745134e-05,
+      "loss": 2.2217,
+      "step": 7780
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.590312358533273e-05,
+      "loss": 2.1022,
+      "step": 7800
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.586690810321413e-05,
+      "loss": 1.9611,
+      "step": 7820
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5830692621095523e-05,
+      "loss": 2.1287,
+      "step": 7840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5794477138976913e-05,
+      "loss": 1.9398,
+      "step": 7860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.5758261656858306e-05,
+      "loss": 2.12,
+      "step": 7880
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5722046174739707e-05,
+      "loss": 2.1609,
+      "step": 7900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.56858306926211e-05,
+      "loss": 2.1651,
+      "step": 7920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.564961521050249e-05,
+      "loss": 2.0493,
+      "step": 7940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5613399728383884e-05,
+      "loss": 1.9657,
+      "step": 7960
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5577184246265277e-05,
+      "loss": 2.1616,
+      "step": 7980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 2.5540968764146678e-05,
+      "loss": 2.053,
+      "step": 8000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5504753282028068e-05,
+      "loss": 2.0428,
+      "step": 8020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.546853779990946e-05,
+      "loss": 2.077,
+      "step": 8040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5432322317790855e-05,
+      "loss": 2.1948,
+      "step": 8060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5396106835672255e-05,
+      "loss": 2.1833,
+      "step": 8080
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 2.5359891353553646e-05,
+      "loss": 2.0653,
+      "step": 8100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.532367587143504e-05,
+      "loss": 2.0288,
+      "step": 8120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5287460389316433e-05,
+      "loss": 2.0778,
+      "step": 8140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5251244907197826e-05,
+      "loss": 2.1851,
+      "step": 8160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5215029425079223e-05,
+      "loss": 2.0221,
+      "step": 8180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.5178813942960617e-05,
+      "loss": 2.3303,
+      "step": 8200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 2.514259846084201e-05,
+      "loss": 2.1052,
+      "step": 8220
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5106382978723404e-05,
+      "loss": 2.0322,
+      "step": 8240
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.50701674966048e-05,
+      "loss": 2.1158,
+      "step": 8260
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.5033952014486195e-05,
+      "loss": 1.9968,
+      "step": 8280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.4997736532367588e-05,
+      "loss": 2.2117,
+      "step": 8300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.496152105024898e-05,
+      "loss": 2.0357,
+      "step": 8320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4925305568130375e-05,
+      "loss": 2.0154,
+      "step": 8340
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4889090086011772e-05,
+      "loss": 1.9553,
+      "step": 8360
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4852874603893166e-05,
+      "loss": 2.0916,
+      "step": 8380
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.481665912177456e-05,
+      "loss": 2.2247,
+      "step": 8400
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.4780443639655953e-05,
+      "loss": 2.0868,
+      "step": 8420
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.474422815753735e-05,
+      "loss": 2.0361,
+      "step": 8440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4708012675418743e-05,
+      "loss": 2.1541,
+      "step": 8460
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4671797193300137e-05,
+      "loss": 2.1148,
+      "step": 8480
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.463558171118153e-05,
+      "loss": 2.0958,
+      "step": 8500
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.4599366229062924e-05,
+      "loss": 1.9408,
+      "step": 8520
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.456315074694432e-05,
+      "loss": 2.0801,
+      "step": 8540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4526935264825715e-05,
+      "loss": 2.0304,
+      "step": 8560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4490719782707108e-05,
+      "loss": 2.0228,
+      "step": 8580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4454504300588502e-05,
+      "loss": 2.227,
+      "step": 8600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44182888184699e-05,
+      "loss": 1.9172,
+      "step": 8620
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4382073336351292e-05,
+      "loss": 2.1165,
+      "step": 8640
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4345857854232686e-05,
+      "loss": 2.0441,
+      "step": 8660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.430964237211408e-05,
+      "loss": 1.9735,
+      "step": 8680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4273426889995476e-05,
+      "loss": 2.0116,
+      "step": 8700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.423721140787687e-05,
+      "loss": 2.1257,
+      "step": 8720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4200995925758263e-05,
+      "loss": 2.0603,
+      "step": 8740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.4164780443639657e-05,
+      "loss": 2.0474,
+      "step": 8760
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.412856496152105e-05,
+      "loss": 2.1369,
+      "step": 8780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4092349479402448e-05,
+      "loss": 2.0122,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.405613399728384e-05,
+      "loss": 2.0775,
+      "step": 8820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.4019918515165235e-05,
+      "loss": 2.0533,
+      "step": 8840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3985513807152557e-05,
+      "loss": 2.0739,
+      "step": 8860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.3949298325033958e-05,
+      "loss": 2.0773,
+      "step": 8880
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.3913082842915348e-05,
+      "loss": 2.1548,
+      "step": 8900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.387686736079674e-05,
+      "loss": 1.9734,
+      "step": 8920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.3840651878678135e-05,
+      "loss": 2.044,
+      "step": 8940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.380443639655953e-05,
+      "loss": 2.0268,
+      "step": 8960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.3768220914440926e-05,
+      "loss": 1.921,
+      "step": 8980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 2.373200543232232e-05,
+      "loss": 2.0945,
+      "step": 9000
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.3695789950203713e-05,
+      "loss": 2.007,
+      "step": 9020
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.3659574468085106e-05,
+      "loss": 2.0309,
+      "step": 9040
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.3623358985966503e-05,
+      "loss": 1.9564,
+      "step": 9060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.3587143503847897e-05,
+      "loss": 1.9914,
+      "step": 9080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 2.355092802172929e-05,
+      "loss": 2.0892,
+      "step": 9100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3514712539610684e-05,
+      "loss": 2.035,
+      "step": 9120
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3478497057492077e-05,
+      "loss": 2.1272,
+      "step": 9140
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3442281575373474e-05,
+      "loss": 1.9546,
+      "step": 9160
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3406066093254868e-05,
+      "loss": 2.0108,
+      "step": 9180
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.336985061113626e-05,
+      "loss": 1.9855,
+      "step": 9200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 2.3333635129017655e-05,
+      "loss": 2.0843,
+      "step": 9220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3297419646899052e-05,
+      "loss": 2.0269,
+      "step": 9240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3261204164780446e-05,
+      "loss": 1.9648,
+      "step": 9260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.322498868266184e-05,
+      "loss": 1.9472,
+      "step": 9280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3188773200543233e-05,
+      "loss": 1.9292,
+      "step": 9300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 2.3152557718424626e-05,
+      "loss": 2.0318,
+      "step": 9320
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3116342236306023e-05,
+      "loss": 2.0421,
+      "step": 9340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3080126754187417e-05,
+      "loss": 2.0236,
+      "step": 9360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.304391127206881e-05,
+      "loss": 2.0441,
+      "step": 9380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.3007695789950204e-05,
+      "loss": 1.995,
+      "step": 9400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.29714803078316e-05,
+      "loss": 2.0099,
+      "step": 9420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 2.2935264825712994e-05,
+      "loss": 2.08,
+      "step": 9440
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2899049343594388e-05,
+      "loss": 1.9742,
+      "step": 9460
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.286283386147578e-05,
+      "loss": 2.1018,
+      "step": 9480
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2826618379357175e-05,
+      "loss": 2.0268,
+      "step": 9500
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2790402897238572e-05,
+      "loss": 2.1419,
+      "step": 9520
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 2.2754187415119966e-05,
+      "loss": 2.0197,
+      "step": 9540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.271797193300136e-05,
+      "loss": 2.0867,
+      "step": 9560
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2681756450882753e-05,
+      "loss": 2.0606,
+      "step": 9580
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.264554096876415e-05,
+      "loss": 2.0034,
+      "step": 9600
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2609325486645543e-05,
+      "loss": 2.1645,
+      "step": 9620
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.2573110004526937e-05,
+      "loss": 2.0127,
+      "step": 9640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 2.253689452240833e-05,
+      "loss": 2.2445,
+      "step": 9660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2500679040289724e-05,
+      "loss": 1.8585,
+      "step": 9680
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.246446355817112e-05,
+      "loss": 2.06,
+      "step": 9700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2430058850158444e-05,
+      "loss": 2.0828,
+      "step": 9720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.2393843368039837e-05,
+      "loss": 2.0931,
+      "step": 9740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 2.235762788592123e-05,
+      "loss": 2.0226,
+      "step": 9760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2321412403802628e-05,
+      "loss": 1.9536,
+      "step": 9780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.228519692168402e-05,
+      "loss": 1.9915,
+      "step": 9800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2248981439565415e-05,
+      "loss": 2.055,
+      "step": 9820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.221276595744681e-05,
+      "loss": 2.0929,
+      "step": 9840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.2176550475328205e-05,
+      "loss": 2.1228,
+      "step": 9860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.21403349932096e-05,
+      "loss": 2.1332,
+      "step": 9880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.2104119511090993e-05,
+      "loss": 1.9926,
+      "step": 9900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.2067904028972386e-05,
+      "loss": 1.9974,
+      "step": 9920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.203168854685378e-05,
+      "loss": 2.0638,
+      "step": 9940
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.1995473064735177e-05,
+      "loss": 2.009,
+      "step": 9960
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.195925758261657e-05,
+      "loss": 2.0809,
+      "step": 9980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1923042100497964e-05,
+      "loss": 1.9369,
+      "step": 10000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1886826618379357e-05,
+      "loss": 2.1579,
+      "step": 10020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1850611136260754e-05,
+      "loss": 2.2145,
+      "step": 10040
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1814395654142148e-05,
+      "loss": 2.0343,
+      "step": 10060
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.177818017202354e-05,
+      "loss": 2.0431,
+      "step": 10080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1741964689904935e-05,
+      "loss": 2.0719,
+      "step": 10100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.170574920778633e-05,
+      "loss": 2.196,
+      "step": 10120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1669533725667725e-05,
+      "loss": 2.074,
+      "step": 10140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.163331824354912e-05,
+      "loss": 1.8136,
+      "step": 10160
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1597102761430513e-05,
+      "loss": 2.0011,
+      "step": 10180
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.1560887279311906e-05,
+      "loss": 1.9408,
+      "step": 10200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1524671797193303e-05,
+      "loss": 1.9268,
+      "step": 10220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1488456315074697e-05,
+      "loss": 2.1235,
+      "step": 10240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.145224083295609e-05,
+      "loss": 2.1492,
+      "step": 10260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1416025350837484e-05,
+      "loss": 2.1039,
+      "step": 10280
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1379809868718877e-05,
+      "loss": 1.9475,
+      "step": 10300
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.1343594386600274e-05,
+      "loss": 2.1523,
+      "step": 10320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1307378904481668e-05,
+      "loss": 2.0971,
+      "step": 10340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.127116342236306e-05,
+      "loss": 2.1351,
+      "step": 10360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1234947940244455e-05,
+      "loss": 1.9302,
+      "step": 10380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1198732458125852e-05,
+      "loss": 2.0062,
+      "step": 10400
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1162516976007246e-05,
+      "loss": 2.1423,
+      "step": 10420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.112630149388864e-05,
+      "loss": 2.1026,
+      "step": 10440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1090086011770033e-05,
+      "loss": 2.0975,
+      "step": 10460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1053870529651426e-05,
+      "loss": 2.0484,
+      "step": 10480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.1017655047532823e-05,
+      "loss": 2.0774,
+      "step": 10500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.0981439565414217e-05,
+      "loss": 2.1849,
+      "step": 10520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 2.094522408329561e-05,
+      "loss": 2.1408,
+      "step": 10540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0909008601177004e-05,
+      "loss": 1.9993,
+      "step": 10560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.08727931190584e-05,
+      "loss": 2.1025,
+      "step": 10580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0836577636939794e-05,
+      "loss": 2.0306,
+      "step": 10600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.0800362154821188e-05,
+      "loss": 2.1032,
+      "step": 10620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 2.076414667270258e-05,
+      "loss": 1.9966,
+      "step": 10640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0727931190583975e-05,
+      "loss": 2.0805,
+      "step": 10660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0691715708465372e-05,
+      "loss": 1.961,
+      "step": 10680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0655500226346766e-05,
+      "loss": 2.1456,
+      "step": 10700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.061928474422816e-05,
+      "loss": 2.1213,
+      "step": 10720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0583069262109553e-05,
+      "loss": 2.1327,
+      "step": 10740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.054685377999095e-05,
+      "loss": 2.044,
+      "step": 10760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0510638297872343e-05,
+      "loss": 2.0865,
+      "step": 10780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0474422815753737e-05,
+      "loss": 1.9971,
+      "step": 10800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.043820733363513e-05,
+      "loss": 1.9774,
+      "step": 10820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0401991851516527e-05,
+      "loss": 1.9416,
+      "step": 10840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.036577636939792e-05,
+      "loss": 2.1607,
+      "step": 10860
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0329560887279314e-05,
+      "loss": 2.1186,
+      "step": 10880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0293345405160708e-05,
+      "loss": 2.0231,
+      "step": 10900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.02571299230421e-05,
+      "loss": 2.13,
+      "step": 10920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.02209144409235e-05,
+      "loss": 1.9632,
+      "step": 10940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0184698958804892e-05,
+      "loss": 2.0626,
+      "step": 10960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0148483476686286e-05,
+      "loss": 2.0685,
+      "step": 10980
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.011226799456768e-05,
+      "loss": 2.0651,
+      "step": 11000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0076052512449076e-05,
+      "loss": 2.0802,
+      "step": 11020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.003983703033047e-05,
+      "loss": 2.0314,
+      "step": 11040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0003621548211863e-05,
+      "loss": 2.1656,
+      "step": 11060
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9967406066093257e-05,
+      "loss": 2.0977,
+      "step": 11080
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.993119058397465e-05,
+      "loss": 2.1811,
+      "step": 11100
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9894975101856044e-05,
+      "loss": 2.0119,
+      "step": 11120
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.985875961973744e-05,
+      "loss": 2.1042,
+      "step": 11140
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9822544137618835e-05,
+      "loss": 2.0716,
+      "step": 11160
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.978813942960616e-05,
+      "loss": 2.0984,
+      "step": 11180
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.975192394748755e-05,
+      "loss": 1.9497,
+      "step": 11200
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9715708465368948e-05,
+      "loss": 2.0127,
+      "step": 11220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.967949298325034e-05,
+      "loss": 2.0654,
+      "step": 11240
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9643277501131735e-05,
+      "loss": 2.045,
+      "step": 11260
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.960706201901313e-05,
+      "loss": 2.0206,
+      "step": 11280
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.9570846536894522e-05,
+      "loss": 1.9633,
+      "step": 11300
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.953463105477592e-05,
+      "loss": 1.9748,
+      "step": 11320
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9498415572657313e-05,
+      "loss": 2.0016,
+      "step": 11340
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9462200090538706e-05,
+      "loss": 2.0058,
+      "step": 11360
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.94259846084201e-05,
+      "loss": 2.0268,
+      "step": 11380
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9391579900407426e-05,
+      "loss": 2.091,
+      "step": 11400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.935536441828882e-05,
+      "loss": 2.0768,
+      "step": 11420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9319148936170213e-05,
+      "loss": 2.1023,
+      "step": 11440
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.928293345405161e-05,
+      "loss": 2.134,
+      "step": 11460
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9246717971933003e-05,
+      "loss": 1.8826,
+      "step": 11480
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9210502489814397e-05,
+      "loss": 1.9726,
+      "step": 11500
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.917428700769579e-05,
+      "loss": 2.0458,
+      "step": 11520
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9138071525577187e-05,
+      "loss": 2.0177,
+      "step": 11540
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.910185604345858e-05,
+      "loss": 2.1323,
+      "step": 11560
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9065640561339975e-05,
+      "loss": 1.9898,
+      "step": 11580
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9029425079221368e-05,
+      "loss": 2.0115,
+      "step": 11600
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8993209597102762e-05,
+      "loss": 2.0374,
+      "step": 11620
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.895699411498416e-05,
+      "loss": 1.9915,
+      "step": 11640
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8920778632865552e-05,
+      "loss": 2.0657,
+      "step": 11660
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8884563150746946e-05,
+      "loss": 2.0139,
+      "step": 11680
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.884834766862834e-05,
+      "loss": 2.0556,
+      "step": 11700
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8812132186509736e-05,
+      "loss": 2.0342,
+      "step": 11720
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.877591670439113e-05,
+      "loss": 1.8927,
+      "step": 11740
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8739701222272523e-05,
+      "loss": 2.0362,
+      "step": 11760
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8703485740153917e-05,
+      "loss": 2.1162,
+      "step": 11780
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8667270258035314e-05,
+      "loss": 2.2084,
+      "step": 11800
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8631054775916708e-05,
+      "loss": 1.9562,
+      "step": 11820
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8594839293798098e-05,
+      "loss": 2.0559,
+      "step": 11840
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8558623811679495e-05,
+      "loss": 2.1134,
+      "step": 11860
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8522408329560888e-05,
+      "loss": 1.928,
+      "step": 11880
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8486192847442285e-05,
+      "loss": 2.1533,
+      "step": 11900
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8449977365323675e-05,
+      "loss": 2.0976,
+      "step": 11920
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8413761883205072e-05,
+      "loss": 2.0301,
+      "step": 11940
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8377546401086466e-05,
+      "loss": 1.9375,
+      "step": 11960
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8341330918967863e-05,
+      "loss": 2.004,
+      "step": 11980
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8305115436849253e-05,
+      "loss": 2.0064,
+      "step": 12000
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8268899954730647e-05,
+      "loss": 2.1905,
+      "step": 12020
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8232684472612044e-05,
+      "loss": 1.9467,
+      "step": 12040
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8196468990493437e-05,
+      "loss": 1.963,
+      "step": 12060
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.816025350837483e-05,
+      "loss": 2.0172,
+      "step": 12080
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8124038026256224e-05,
+      "loss": 2.1093,
+      "step": 12100
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.808782254413762e-05,
+      "loss": 1.9967,
+      "step": 12120
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8051607062019015e-05,
+      "loss": 2.0362,
+      "step": 12140
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8015391579900408e-05,
+      "loss": 2.0415,
+      "step": 12160
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.7979176097781802e-05,
+      "loss": 1.9409,
+      "step": 12180
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.79429606156632e-05,
+      "loss": 2.0646,
+      "step": 12200
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7906745133544592e-05,
+      "loss": 2.0637,
+      "step": 12220
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7870529651425986e-05,
+      "loss": 2.1201,
+      "step": 12240
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.783431416930738e-05,
+      "loss": 2.1059,
+      "step": 12260
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.7798098687188773e-05,
+      "loss": 2.0127,
+      "step": 12280
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.776188320507017e-05,
+      "loss": 1.9974,
+      "step": 12300
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7725667722951564e-05,
+      "loss": 2.1898,
+      "step": 12320
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7689452240832957e-05,
+      "loss": 1.9657,
+      "step": 12340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.765323675871435e-05,
+      "loss": 1.9698,
+      "step": 12360
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7617021276595748e-05,
+      "loss": 1.9256,
+      "step": 12380
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.758080579447714e-05,
+      "loss": 1.8693,
+      "step": 12400
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.7544590312358535e-05,
+      "loss": 1.9088,
+      "step": 12420
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.750837483023993e-05,
+      "loss": 2.1958,
+      "step": 12440
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7472159348121322e-05,
+      "loss": 1.9507,
+      "step": 12460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.743594386600272e-05,
+      "loss": 2.0344,
+      "step": 12480
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7399728383884112e-05,
+      "loss": 1.9669,
+      "step": 12500
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.7363512901765506e-05,
+      "loss": 2.0499,
+      "step": 12520
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.73272974196469e-05,
+      "loss": 2.0765,
+      "step": 12540
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7291081937528297e-05,
+      "loss": 2.0399,
+      "step": 12560
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.725667722951562e-05,
+      "loss": 2.1701,
+      "step": 12580
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7220461747397013e-05,
+      "loss": 2.0596,
+      "step": 12600
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.718424626527841e-05,
+      "loss": 1.9806,
+      "step": 12620
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.71480307831598e-05,
+      "loss": 2.0696,
+      "step": 12640
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7111815301041197e-05,
+      "loss": 2.1453,
+      "step": 12660
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.707559981892259e-05,
+      "loss": 2.0051,
+      "step": 12680
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7039384336803987e-05,
+      "loss": 2.079,
+      "step": 12700
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7003168854685378e-05,
+      "loss": 2.054,
+      "step": 12720
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.6966953372566775e-05,
+      "loss": 2.0768,
+      "step": 12740
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6930737890448168e-05,
+      "loss": 2.0489,
+      "step": 12760
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6894522408329565e-05,
+      "loss": 2.1374,
+      "step": 12780
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6858306926210955e-05,
+      "loss": 2.0152,
+      "step": 12800
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.682209144409235e-05,
+      "loss": 2.082,
+      "step": 12820
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.6787686736079675e-05,
+      "loss": 2.0217,
+      "step": 12840
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.675147125396107e-05,
+      "loss": 2.0073,
+      "step": 12860
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6715255771842465e-05,
+      "loss": 2.0909,
+      "step": 12880
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.667904028972386e-05,
+      "loss": 1.961,
+      "step": 12900
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6642824807605252e-05,
+      "loss": 2.1695,
+      "step": 12920
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6606609325486646e-05,
+      "loss": 2.0455,
+      "step": 12940
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.6570393843368043e-05,
+      "loss": 2.1383,
+      "step": 12960
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6534178361249437e-05,
+      "loss": 2.1036,
+      "step": 12980
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.649796287913083e-05,
+      "loss": 2.0603,
+      "step": 13000
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6461747397012224e-05,
+      "loss": 2.0141,
+      "step": 13020
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6425531914893617e-05,
+      "loss": 2.0332,
+      "step": 13040
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6389316432775014e-05,
+      "loss": 2.0475,
+      "step": 13060
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.6353100950656408e-05,
+      "loss": 2.014,
+      "step": 13080
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.63168854685378e-05,
+      "loss": 1.9751,
+      "step": 13100
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6280669986419195e-05,
+      "loss": 1.9468,
+      "step": 13120
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6244454504300592e-05,
+      "loss": 2.0346,
+      "step": 13140
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.6208239022181985e-05,
+      "loss": 2.1157,
+      "step": 13160
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.617202354006338e-05,
+      "loss": 2.0459,
+      "step": 13180
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6135808057944773e-05,
+      "loss": 2.0237,
+      "step": 13200
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6099592575826166e-05,
+      "loss": 1.9888,
+      "step": 13220
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6063377093707563e-05,
+      "loss": 1.9949,
+      "step": 13240
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.6027161611588957e-05,
+      "loss": 2.1008,
+      "step": 13260
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.599094612947035e-05,
+      "loss": 1.9829,
+      "step": 13280
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5954730647351744e-05,
+      "loss": 1.9491,
+      "step": 13300
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.591851516523314e-05,
+      "loss": 1.9661,
+      "step": 13320
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5882299683114534e-05,
+      "loss": 1.9159,
+      "step": 13340
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5846084200995928e-05,
+      "loss": 2.0245,
+      "step": 13360
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.580986871887732e-05,
+      "loss": 2.0905,
+      "step": 13380
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5773653236758715e-05,
+      "loss": 1.9746,
+      "step": 13400
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5737437754640112e-05,
+      "loss": 2.0959,
+      "step": 13420
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5701222272521502e-05,
+      "loss": 1.9769,
+      "step": 13440
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.56650067904029e-05,
+      "loss": 2.1005,
+      "step": 13460
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5628791308284293e-05,
+      "loss": 2.1608,
+      "step": 13480
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5592575826165686e-05,
+      "loss": 2.006,
+      "step": 13500
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.555636034404708e-05,
+      "loss": 2.008,
+      "step": 13520
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5520144861928477e-05,
+      "loss": 2.1034,
+      "step": 13540
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.548392937980987e-05,
+      "loss": 2.0376,
+      "step": 13560
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5447713897691264e-05,
+      "loss": 2.0343,
+      "step": 13580
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5411498415572657e-05,
+      "loss": 2.0762,
+      "step": 13600
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.537528293345405e-05,
+      "loss": 1.9863,
+      "step": 13620
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5339067451335448e-05,
+      "loss": 2.0349,
+      "step": 13640
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.530285196921684e-05,
+      "loss": 1.9007,
+      "step": 13660
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5266636487098235e-05,
+      "loss": 1.9318,
+      "step": 13680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.523042100497963e-05,
+      "loss": 2.0908,
+      "step": 13700
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5194205522861026e-05,
+      "loss": 2.1327,
+      "step": 13720
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5157990040742419e-05,
+      "loss": 2.1275,
+      "step": 13740
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5123585332729742e-05,
+      "loss": 2.0555,
+      "step": 13760
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5087369850611137e-05,
+      "loss": 2.0588,
+      "step": 13780
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.505115436849253e-05,
+      "loss": 2.1312,
+      "step": 13800
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5014938886373926e-05,
+      "loss": 2.0227,
+      "step": 13820
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.497872340425532e-05,
+      "loss": 2.073,
+      "step": 13840
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4942507922136715e-05,
+      "loss": 2.0305,
+      "step": 13860
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4906292440018108e-05,
+      "loss": 2.0507,
+      "step": 13880
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4870076957899504e-05,
+      "loss": 1.9528,
+      "step": 13900
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4833861475780897e-05,
+      "loss": 2.0968,
+      "step": 13920
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4797645993662292e-05,
+      "loss": 2.0055,
+      "step": 13940
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4761430511543686e-05,
+      "loss": 1.8985,
+      "step": 13960
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.472521502942508e-05,
+      "loss": 2.0214,
+      "step": 13980
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4688999547306475e-05,
+      "loss": 2.0261,
+      "step": 14000
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4652784065187868e-05,
+      "loss": 2.0178,
+      "step": 14020
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4616568583069264e-05,
+      "loss": 2.0605,
+      "step": 14040
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4580353100950657e-05,
+      "loss": 1.9876,
+      "step": 14060
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4544137618832052e-05,
+      "loss": 1.8739,
+      "step": 14080
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4507922136713446e-05,
+      "loss": 2.0477,
+      "step": 14100
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4471706654594841e-05,
+      "loss": 1.9751,
+      "step": 14120
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4435491172476235e-05,
+      "loss": 1.9342,
+      "step": 14140
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4399275690357628e-05,
+      "loss": 2.0952,
+      "step": 14160
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4363060208239024e-05,
+      "loss": 2.0246,
+      "step": 14180
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4326844726120417e-05,
+      "loss": 1.9082,
+      "step": 14200
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4290629244001812e-05,
+      "loss": 1.9938,
+      "step": 14220
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4254413761883206e-05,
+      "loss": 2.0018,
+      "step": 14240
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4218198279764601e-05,
+      "loss": 1.9841,
+      "step": 14260
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4181982797645995e-05,
+      "loss": 1.9467,
+      "step": 14280
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.414576731552739e-05,
+      "loss": 2.0645,
+      "step": 14300
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4109551833408784e-05,
+      "loss": 2.1864,
+      "step": 14320
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4073336351290177e-05,
+      "loss": 2.0997,
+      "step": 14340
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4037120869171572e-05,
+      "loss": 1.9941,
+      "step": 14360
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4000905387052966e-05,
+      "loss": 1.9623,
+      "step": 14380
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.3964689904934361e-05,
+      "loss": 2.0867,
+      "step": 14400
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3928474422815755e-05,
+      "loss": 1.9384,
+      "step": 14420
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.389225894069715e-05,
+      "loss": 2.0535,
+      "step": 14440
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3856043458578544e-05,
+      "loss": 2.1574,
+      "step": 14460
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3819827976459939e-05,
+      "loss": 2.0478,
+      "step": 14480
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3783612494341333e-05,
+      "loss": 1.9597,
+      "step": 14500
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.3747397012222728e-05,
+      "loss": 1.9673,
+      "step": 14520
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3711181530104121e-05,
+      "loss": 2.0668,
+      "step": 14540
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3674966047985513e-05,
+      "loss": 1.9192,
+      "step": 14560
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.363875056586691e-05,
+      "loss": 2.0558,
+      "step": 14580
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3602535083748302e-05,
+      "loss": 1.9952,
+      "step": 14600
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.3566319601629699e-05,
+      "loss": 2.1496,
+      "step": 14620
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3530104119511091e-05,
+      "loss": 2.1666,
+      "step": 14640
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3493888637392488e-05,
+      "loss": 2.1297,
+      "step": 14660
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.345767315527388e-05,
+      "loss": 1.9451,
+      "step": 14680
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3421457673155277e-05,
+      "loss": 1.8763,
+      "step": 14700
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3385242191036669e-05,
+      "loss": 2.0955,
+      "step": 14720
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.3349026708918062e-05,
+      "loss": 2.1524,
+      "step": 14740
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3312811226799457e-05,
+      "loss": 1.9623,
+      "step": 14760
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3276595744680851e-05,
+      "loss": 2.0334,
+      "step": 14780
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3240380262562246e-05,
+      "loss": 2.055,
+      "step": 14800
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.320416478044364e-05,
+      "loss": 2.0647,
+      "step": 14820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.3167949298325035e-05,
+      "loss": 1.9503,
+      "step": 14840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3131733816206429e-05,
+      "loss": 2.3149,
+      "step": 14860
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3095518334087824e-05,
+      "loss": 1.9999,
+      "step": 14880
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3059302851969217e-05,
+      "loss": 2.0607,
+      "step": 14900
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.3023087369850613e-05,
+      "loss": 2.0617,
+      "step": 14920
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2986871887732006e-05,
+      "loss": 1.9725,
+      "step": 14940
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.29506564056134e-05,
+      "loss": 2.1148,
+      "step": 14960
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2914440923494795e-05,
+      "loss": 2.1168,
+      "step": 14980
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2878225441376189e-05,
+      "loss": 2.0361,
+      "step": 15000
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2842009959257584e-05,
+      "loss": 2.0631,
+      "step": 15020
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2805794477138977e-05,
+      "loss": 2.2278,
+      "step": 15040
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2769578995020373e-05,
+      "loss": 1.9343,
+      "step": 15060
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2733363512901766e-05,
+      "loss": 2.0086,
+      "step": 15080
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2697148030783161e-05,
+      "loss": 1.9276,
+      "step": 15100
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2660932548664555e-05,
+      "loss": 1.9663,
+      "step": 15120
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2624717066545949e-05,
+      "loss": 2.0225,
+      "step": 15140
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2588501584427344e-05,
+      "loss": 1.8871,
+      "step": 15160
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.2552286102308737e-05,
+      "loss": 2.0613,
+      "step": 15180
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2516070620190133e-05,
+      "loss": 2.0857,
+      "step": 15200
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2479855138071526e-05,
+      "loss": 1.9739,
+      "step": 15220
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2443639655952922e-05,
+      "loss": 2.0949,
+      "step": 15240
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.2407424173834315e-05,
+      "loss": 1.9926,
+      "step": 15260
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.237120869171571e-05,
+      "loss": 2.0389,
+      "step": 15280
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2334993209597104e-05,
+      "loss": 2.1397,
+      "step": 15300
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2298777727478497e-05,
+      "loss": 2.047,
+      "step": 15320
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2262562245359893e-05,
+      "loss": 1.9892,
+      "step": 15340
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2226346763241286e-05,
+      "loss": 2.0924,
+      "step": 15360
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2190131281122682e-05,
+      "loss": 1.9841,
+      "step": 15380
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.2153915799004075e-05,
+      "loss": 1.9659,
+      "step": 15400
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.211770031688547e-05,
+      "loss": 1.9502,
+      "step": 15420
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2081484834766864e-05,
+      "loss": 2.1378,
+      "step": 15440
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.204526935264826e-05,
+      "loss": 2.0291,
+      "step": 15460
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.2009053870529653e-05,
+      "loss": 2.1513,
+      "step": 15480
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1972838388411048e-05,
+      "loss": 2.0587,
+      "step": 15500
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1936622906292442e-05,
+      "loss": 1.9707,
+      "step": 15520
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1900407424173835e-05,
+      "loss": 2.0306,
+      "step": 15540
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.186419194205523e-05,
+      "loss": 2.1524,
+      "step": 15560
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1827976459936624e-05,
+      "loss": 1.9775,
+      "step": 15580
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.179176097781802e-05,
+      "loss": 1.9971,
+      "step": 15600
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1755545495699411e-05,
+      "loss": 2.1624,
+      "step": 15620
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1719330013580808e-05,
+      "loss": 1.9798,
+      "step": 15640
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.16831145314622e-05,
+      "loss": 2.0609,
+      "step": 15660
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1646899049343597e-05,
+      "loss": 1.9897,
+      "step": 15680
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1610683567224989e-05,
+      "loss": 1.8303,
+      "step": 15700
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1574468085106382e-05,
+      "loss": 2.0673,
+      "step": 15720
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1538252602987778e-05,
+      "loss": 2.0646,
+      "step": 15740
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1502037120869171e-05,
+      "loss": 2.0393,
+      "step": 15760
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1465821638750566e-05,
+      "loss": 1.9769,
+      "step": 15780
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.142960615663196e-05,
+      "loss": 1.8212,
+      "step": 15800
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1393390674513355e-05,
+      "loss": 2.2035,
+      "step": 15820
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1357175192394749e-05,
+      "loss": 2.0192,
+      "step": 15840
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1320959710276144e-05,
+      "loss": 2.0477,
+      "step": 15860
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1284744228157538e-05,
+      "loss": 2.0584,
+      "step": 15880
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1248528746038933e-05,
+      "loss": 2.0377,
+      "step": 15900
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1212313263920326e-05,
+      "loss": 1.9808,
+      "step": 15920
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.117609778180172e-05,
+      "loss": 1.994,
+      "step": 15940
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1139882299683115e-05,
+      "loss": 1.9633,
+      "step": 15960
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1103666817564509e-05,
+      "loss": 2.0782,
+      "step": 15980
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1067451335445904e-05,
+      "loss": 2.0446,
+      "step": 16000
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1031235853327298e-05,
+      "loss": 1.9386,
+      "step": 16020
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0995020371208693e-05,
+      "loss": 2.0138,
+      "step": 16040
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0958804889090086e-05,
+      "loss": 2.1022,
+      "step": 16060
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0922589406971482e-05,
+      "loss": 2.0243,
+      "step": 16080
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0886373924852875e-05,
+      "loss": 2.1387,
+      "step": 16100
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0850158442734269e-05,
+      "loss": 1.9336,
+      "step": 16120
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0813942960615664e-05,
+      "loss": 2.1144,
+      "step": 16140
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0777727478497058e-05,
+      "loss": 1.9973,
+      "step": 16160
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0741511996378453e-05,
+      "loss": 1.9627,
+      "step": 16180
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0705296514259846e-05,
+      "loss": 1.9629,
+      "step": 16200
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0669081032141242e-05,
+      "loss": 1.9753,
+      "step": 16220
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0632865550022635e-05,
+      "loss": 2.0677,
+      "step": 16240
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.059665006790403e-05,
+      "loss": 2.0087,
+      "step": 16260
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0560434585785424e-05,
+      "loss": 2.0612,
+      "step": 16280
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0524219103666818e-05,
+      "loss": 2.071,
+      "step": 16300
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0488003621548213e-05,
+      "loss": 2.1715,
+      "step": 16320
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0451788139429607e-05,
+      "loss": 2.0065,
+      "step": 16340
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0415572657311002e-05,
+      "loss": 2.0538,
+      "step": 16360
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0379357175192395e-05,
+      "loss": 1.9638,
+      "step": 16380
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.034314169307379e-05,
+      "loss": 1.9035,
+      "step": 16400
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0306926210955184e-05,
+      "loss": 1.9809,
+      "step": 16420
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.027071072883658e-05,
+      "loss": 1.9515,
+      "step": 16440
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0234495246717973e-05,
+      "loss": 1.9427,
+      "step": 16460
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0198279764599368e-05,
+      "loss": 2.0155,
+      "step": 16480
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0162064282480762e-05,
+      "loss": 2.0799,
+      "step": 16500
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0125848800362155e-05,
+      "loss": 1.9325,
+      "step": 16520
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.008963331824355e-05,
+      "loss": 1.8791,
+      "step": 16540
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0053417836124944e-05,
+      "loss": 1.9997,
+      "step": 16560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.001720235400634e-05,
+      "loss": 1.9296,
+      "step": 16580
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.980986871887733e-06,
+      "loss": 2.0251,
+      "step": 16600
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 9.944771389769127e-06,
+      "loss": 1.916,
+      "step": 16620
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.908555907650522e-06,
+      "loss": 1.8849,
+      "step": 16640
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.872340425531915e-06,
+      "loss": 2.0681,
+      "step": 16660
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.83612494341331e-06,
+      "loss": 2.0353,
+      "step": 16680
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.799909461294704e-06,
+      "loss": 2.039,
+      "step": 16700
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 9.7636939791761e-06,
+      "loss": 1.9525,
+      "step": 16720
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.727478497057493e-06,
+      "loss": 1.9731,
+      "step": 16740
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.691263014938888e-06,
+      "loss": 2.0152,
+      "step": 16760
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.655047532820282e-06,
+      "loss": 2.0452,
+      "step": 16780
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.620642824807606e-06,
+      "loss": 2.0566,
+      "step": 16800
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.58623811679493e-06,
+      "loss": 1.9455,
+      "step": 16820
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 9.550022634676324e-06,
+      "loss": 2.0453,
+      "step": 16840
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.51380715255772e-06,
+      "loss": 2.0898,
+      "step": 16860
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.477591670439113e-06,
+      "loss": 2.0093,
+      "step": 16880
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.441376188320508e-06,
+      "loss": 2.0142,
+      "step": 16900
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.405160706201902e-06,
+      "loss": 2.0167,
+      "step": 16920
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 9.368945224083297e-06,
+      "loss": 1.9674,
+      "step": 16940
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.33272974196469e-06,
+      "loss": 1.9232,
+      "step": 16960
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.296514259846084e-06,
+      "loss": 2.1215,
+      "step": 16980
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.26029877772748e-06,
+      "loss": 2.143,
+      "step": 17000
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.224083295608873e-06,
+      "loss": 1.9697,
+      "step": 17020
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.187867813490268e-06,
+      "loss": 2.0715,
+      "step": 17040
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 9.151652331371662e-06,
+      "loss": 1.999,
+      "step": 17060
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.115436849253057e-06,
+      "loss": 2.1776,
+      "step": 17080
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.07922136713445e-06,
+      "loss": 2.048,
+      "step": 17100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.043005885015846e-06,
+      "loss": 1.9566,
+      "step": 17120
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.00679040289724e-06,
+      "loss": 1.9839,
+      "step": 17140
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 8.970574920778635e-06,
+      "loss": 1.9067,
+      "step": 17160
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.934359438660027e-06,
+      "loss": 1.9615,
+      "step": 17180
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.898143956541422e-06,
+      "loss": 2.0626,
+      "step": 17200
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.861928474422816e-06,
+      "loss": 1.9505,
+      "step": 17220
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.82571299230421e-06,
+      "loss": 1.9661,
+      "step": 17240
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.789497510185604e-06,
+      "loss": 1.9831,
+      "step": 17260
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 8.753282028067e-06,
+      "loss": 2.0268,
+      "step": 17280
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.717066545948393e-06,
+      "loss": 2.0192,
+      "step": 17300
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.680851063829788e-06,
+      "loss": 2.0519,
+      "step": 17320
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.644635581711182e-06,
+      "loss": 2.0572,
+      "step": 17340
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.608420099592577e-06,
+      "loss": 2.1422,
+      "step": 17360
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 8.57220461747397e-06,
+      "loss": 1.9839,
+      "step": 17380
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.535989135355364e-06,
+      "loss": 1.9993,
+      "step": 17400
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.501584427342689e-06,
+      "loss": 1.9822,
+      "step": 17420
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.465368945224084e-06,
+      "loss": 2.0014,
+      "step": 17440
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.429153463105478e-06,
+      "loss": 2.0362,
+      "step": 17460
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.392937980986873e-06,
+      "loss": 1.978,
+      "step": 17480
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 8.356722498868266e-06,
+      "loss": 2.1017,
+      "step": 17500
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.320507016749662e-06,
+      "loss": 2.0356,
+      "step": 17520
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.284291534631055e-06,
+      "loss": 1.9816,
+      "step": 17540
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.248076052512449e-06,
+      "loss": 2.0248,
+      "step": 17560
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.211860570393844e-06,
+      "loss": 1.9248,
+      "step": 17580
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 8.175645088275238e-06,
+      "loss": 1.9648,
+      "step": 17600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.139429606156633e-06,
+      "loss": 1.8626,
+      "step": 17620
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.103214124038026e-06,
+      "loss": 2.1176,
+      "step": 17640
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.066998641919422e-06,
+      "loss": 2.0206,
+      "step": 17660
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 8.030783159800815e-06,
+      "loss": 2.0182,
+      "step": 17680
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.99456767768221e-06,
+      "loss": 1.9659,
+      "step": 17700
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.958352195563604e-06,
+      "loss": 2.0059,
+      "step": 17720
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.922136713445e-06,
+      "loss": 1.9191,
+      "step": 17740
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.885921231326393e-06,
+      "loss": 2.0412,
+      "step": 17760
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.849705749207786e-06,
+      "loss": 2.1335,
+      "step": 17780
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.813490267089182e-06,
+      "loss": 1.9905,
+      "step": 17800
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.777274784970575e-06,
+      "loss": 2.1627,
+      "step": 17820
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.74105930285197e-06,
+      "loss": 2.0587,
+      "step": 17840
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.704843820733364e-06,
+      "loss": 1.9432,
+      "step": 17860
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.66862833861476e-06,
+      "loss": 1.95,
+      "step": 17880
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.632412856496153e-06,
+      "loss": 2.0322,
+      "step": 17900
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.596197374377547e-06,
+      "loss": 2.0037,
+      "step": 17920
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.559981892258942e-06,
+      "loss": 1.9361,
+      "step": 17940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.523766410140335e-06,
+      "loss": 2.1615,
+      "step": 17960
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.48755092802173e-06,
+      "loss": 2.0098,
+      "step": 17980
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.451335445903124e-06,
+      "loss": 2.0331,
+      "step": 18000
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.415119963784519e-06,
+      "loss": 2.0847,
+      "step": 18020
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.378904481665913e-06,
+      "loss": 1.9978,
+      "step": 18040
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.342688999547307e-06,
+      "loss": 1.8751,
+      "step": 18060
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.306473517428702e-06,
+      "loss": 2.2345,
+      "step": 18080
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.270258035310096e-06,
+      "loss": 1.925,
+      "step": 18100
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.234042553191491e-06,
+      "loss": 2.0296,
+      "step": 18120
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.197827071072883e-06,
+      "loss": 1.994,
+      "step": 18140
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.161611588954278e-06,
+      "loss": 1.9312,
+      "step": 18160
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.125396106835672e-06,
+      "loss": 1.9623,
+      "step": 18180
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.089180624717067e-06,
+      "loss": 2.0018,
+      "step": 18200
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.052965142598461e-06,
+      "loss": 1.988,
+      "step": 18220
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.016749660479855e-06,
+      "loss": 2.0251,
+      "step": 18240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 6.98053417836125e-06,
+      "loss": 2.0355,
+      "step": 18260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.944318696242644e-06,
+      "loss": 1.9063,
+      "step": 18280
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.908103214124039e-06,
+      "loss": 1.8837,
+      "step": 18300
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.871887732005433e-06,
+      "loss": 2.0012,
+      "step": 18320
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.835672249886827e-06,
+      "loss": 2.0049,
+      "step": 18340
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.799456767768221e-06,
+      "loss": 1.9033,
+      "step": 18360
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 6.7632412856496154e-06,
+      "loss": 1.8857,
+      "step": 18380
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.72702580353101e-06,
+      "loss": 1.9256,
+      "step": 18400
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.690810321412404e-06,
+      "loss": 2.0365,
+      "step": 18420
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.654594839293799e-06,
+      "loss": 1.9982,
+      "step": 18440
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.618379357175193e-06,
+      "loss": 1.9691,
+      "step": 18460
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.5821638750565875e-06,
+      "loss": 1.9749,
+      "step": 18480
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 6.545948392937982e-06,
+      "loss": 2.0684,
+      "step": 18500
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.509732910819376e-06,
+      "loss": 1.9444,
+      "step": 18520
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.47351742870077e-06,
+      "loss": 2.0301,
+      "step": 18540
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.437301946582164e-06,
+      "loss": 2.1235,
+      "step": 18560
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.401086464463559e-06,
+      "loss": 2.02,
+      "step": 18580
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 6.364870982344953e-06,
+      "loss": 2.027,
+      "step": 18600
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.3286555002263475e-06,
+      "loss": 2.0242,
+      "step": 18620
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.294250792213672e-06,
+      "loss": 2.0671,
+      "step": 18640
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.258035310095066e-06,
+      "loss": 2.0188,
+      "step": 18660
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.221819827976461e-06,
+      "loss": 2.1455,
+      "step": 18680
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.185604345857855e-06,
+      "loss": 1.9436,
+      "step": 18700
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 6.149388863739249e-06,
+      "loss": 2.0486,
+      "step": 18720
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.113173381620643e-06,
+      "loss": 2.0407,
+      "step": 18740
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.0769578995020376e-06,
+      "loss": 2.146,
+      "step": 18760
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.040742417383432e-06,
+      "loss": 1.9871,
+      "step": 18780
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 6.004526935264826e-06,
+      "loss": 2.0523,
+      "step": 18800
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 5.968311453146221e-06,
+      "loss": 2.0115,
+      "step": 18820
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.932095971027615e-06,
+      "loss": 2.0613,
+      "step": 18840
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.89588048890901e-06,
+      "loss": 2.0066,
+      "step": 18860
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.859665006790404e-06,
+      "loss": 2.0445,
+      "step": 18880
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.8234495246717984e-06,
+      "loss": 1.9291,
+      "step": 18900
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.787234042553191e-06,
+      "loss": 1.9039,
+      "step": 18920
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 5.7510185604345856e-06,
+      "loss": 2.0281,
+      "step": 18940
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.71480307831598e-06,
+      "loss": 2.0617,
+      "step": 18960
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.678587596197374e-06,
+      "loss": 1.8976,
+      "step": 18980
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.642372114078769e-06,
+      "loss": 1.917,
+      "step": 19000
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.606156631960163e-06,
+      "loss": 2.092,
+      "step": 19020
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 5.569941149841558e-06,
+      "loss": 1.9123,
+      "step": 19040
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.533725667722952e-06,
+      "loss": 2.0138,
+      "step": 19060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.499320959710277e-06,
+      "loss": 2.07,
+      "step": 19080
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.463105477591672e-06,
+      "loss": 2.1708,
+      "step": 19100
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.426889995473064e-06,
+      "loss": 2.0515,
+      "step": 19120
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.390674513354459e-06,
+      "loss": 2.0282,
+      "step": 19140
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 5.354459031235853e-06,
+      "loss": 2.0866,
+      "step": 19160
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.318243549117248e-06,
+      "loss": 2.0123,
+      "step": 19180
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.282028066998642e-06,
+      "loss": 1.998,
+      "step": 19200
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.2458125848800365e-06,
+      "loss": 2.0993,
+      "step": 19220
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.209597102761431e-06,
+      "loss": 2.029,
+      "step": 19240
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 5.173381620642825e-06,
+      "loss": 2.056,
+      "step": 19260
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.13716613852422e-06,
+      "loss": 2.0467,
+      "step": 19280
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.100950656405613e-06,
+      "loss": 2.0471,
+      "step": 19300
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.064735174287008e-06,
+      "loss": 2.0587,
+      "step": 19320
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 5.028519692168402e-06,
+      "loss": 2.1116,
+      "step": 19340
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.9923042100497965e-06,
+      "loss": 1.9219,
+      "step": 19360
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.956088727931191e-06,
+      "loss": 1.9429,
+      "step": 19380
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.919873245812585e-06,
+      "loss": 2.1182,
+      "step": 19400
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.88365776369398e-06,
+      "loss": 2.0713,
+      "step": 19420
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.847442281575374e-06,
+      "loss": 1.9771,
+      "step": 19440
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.811226799456768e-06,
+      "loss": 1.9952,
+      "step": 19460
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.775011317338162e-06,
+      "loss": 2.0599,
+      "step": 19480
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.7387958352195565e-06,
+      "loss": 2.0133,
+      "step": 19500
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.702580353100951e-06,
+      "loss": 2.017,
+      "step": 19520
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.666364870982345e-06,
+      "loss": 2.1722,
+      "step": 19540
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.63014938886374e-06,
+      "loss": 1.983,
+      "step": 19560
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.593933906745134e-06,
+      "loss": 2.2004,
+      "step": 19580
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.557718424626529e-06,
+      "loss": 1.9737,
+      "step": 19600
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.521502942507923e-06,
+      "loss": 2.0671,
+      "step": 19620
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.485287460389317e-06,
+      "loss": 1.8798,
+      "step": 19640
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.449071978270711e-06,
+      "loss": 1.8859,
+      "step": 19660
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.412856496152105e-06,
+      "loss": 2.0654,
+      "step": 19680
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.3766410140335e-06,
+      "loss": 2.0789,
+      "step": 19700
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.340425531914894e-06,
+      "loss": 2.2518,
+      "step": 19720
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.304210049796289e-06,
+      "loss": 2.0008,
+      "step": 19740
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.267994567677682e-06,
+      "loss": 1.9164,
+      "step": 19760
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.231779085559077e-06,
+      "loss": 2.0358,
+      "step": 19780
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.195563603440471e-06,
+      "loss": 2.0897,
+      "step": 19800
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.1593481213218654e-06,
+      "loss": 2.0521,
+      "step": 19820
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.12313263920326e-06,
+      "loss": 2.0629,
+      "step": 19840
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.086917157084654e-06,
+      "loss": 1.9844,
+      "step": 19860
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.050701674966049e-06,
+      "loss": 2.0107,
+      "step": 19880
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.014486192847443e-06,
+      "loss": 2.124,
+      "step": 19900
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.9782707107288375e-06,
+      "loss": 1.9956,
+      "step": 19920
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.942055228610231e-06,
+      "loss": 2.0532,
+      "step": 19940
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.9058397464916255e-06,
+      "loss": 2.2484,
+      "step": 19960
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.86962426437302e-06,
+      "loss": 1.9962,
+      "step": 19980
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.833408782254414e-06,
+      "loss": 2.1203,
+      "step": 20000
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7971933001358087e-06,
+      "loss": 2.0752,
+      "step": 20020
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.7609778180172023e-06,
+      "loss": 1.9413,
+      "step": 20040
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.7247623358985967e-06,
+      "loss": 1.977,
+      "step": 20060
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.688546853779991e-06,
+      "loss": 2.0783,
+      "step": 20080
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.6523313716613855e-06,
+      "loss": 2.2848,
+      "step": 20100
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.61611588954278e-06,
+      "loss": 2.0972,
+      "step": 20120
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.579900407424174e-06,
+      "loss": 2.0313,
+      "step": 20140
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5436849253055683e-06,
+      "loss": 2.0227,
+      "step": 20160
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.5074694431869627e-06,
+      "loss": 2.0386,
+      "step": 20180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.471253961068357e-06,
+      "loss": 2.0474,
+      "step": 20200
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.4350384789497515e-06,
+      "loss": 2.0421,
+      "step": 20220
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.3988229968311455e-06,
+      "loss": 2.1014,
+      "step": 20240
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.36260751471254e-06,
+      "loss": 1.8945,
+      "step": 20260
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.3263920325939343e-06,
+      "loss": 2.0787,
+      "step": 20280
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2901765504753288e-06,
+      "loss": 1.8478,
+      "step": 20300
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.253961068356723e-06,
+      "loss": 2.1288,
+      "step": 20320
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.2177455862381167e-06,
+      "loss": 2.0716,
+      "step": 20340
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.181530104119511e-06,
+      "loss": 1.9161,
+      "step": 20360
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1453146220009056e-06,
+      "loss": 2.1212,
+      "step": 20380
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.1090991398823e-06,
+      "loss": 1.9724,
+      "step": 20400
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0728836577636944e-06,
+      "loss": 2.0512,
+      "step": 20420
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0366681756450884e-06,
+      "loss": 2.1767,
+      "step": 20440
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.0004526935264828e-06,
+      "loss": 1.9733,
+      "step": 20460
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 2.964237211407877e-06,
+      "loss": 2.1791,
+      "step": 20480
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.9280217292892716e-06,
+      "loss": 2.011,
+      "step": 20500
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.891806247170666e-06,
+      "loss": 2.0423,
+      "step": 20520
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.85740153915799e-06,
+      "loss": 1.9699,
+      "step": 20540
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.822996831145315e-06,
+      "loss": 2.082,
+      "step": 20560
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 2.7867813490267092e-06,
+      "loss": 2.0362,
+      "step": 20580
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.7505658669081036e-06,
+      "loss": 2.0117,
+      "step": 20600
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.7143503847894976e-06,
+      "loss": 1.8959,
+      "step": 20620
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.678134902670892e-06,
+      "loss": 2.1008,
+      "step": 20640
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.6419194205522864e-06,
+      "loss": 2.1399,
+      "step": 20660
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.605703938433681e-06,
+      "loss": 1.9281,
+      "step": 20680
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 2.5694884563150753e-06,
+      "loss": 2.1764,
+      "step": 20700
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.533272974196469e-06,
+      "loss": 1.9605,
+      "step": 20720
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4970574920778632e-06,
+      "loss": 2.1025,
+      "step": 20740
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.4608420099592577e-06,
+      "loss": 1.9564,
+      "step": 20760
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.424626527840652e-06,
+      "loss": 1.8858,
+      "step": 20780
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 2.3884110457220465e-06,
+      "loss": 1.8728,
+      "step": 20800
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.352195563603441e-06,
+      "loss": 1.9324,
+      "step": 20820
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.315980081484835e-06,
+      "loss": 1.9937,
+      "step": 20840
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2797645993662293e-06,
+      "loss": 2.1515,
+      "step": 20860
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2435491172476233e-06,
+      "loss": 2.0864,
+      "step": 20880
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.2073336351290177e-06,
+      "loss": 2.1283,
+      "step": 20900
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 2.171118153010412e-06,
+      "loss": 1.9856,
+      "step": 20920
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.1349026708918065e-06,
+      "loss": 1.972,
+      "step": 20940
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.098687188773201e-06,
+      "loss": 2.0295,
+      "step": 20960
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.062471706654595e-06,
+      "loss": 2.0955,
+      "step": 20980
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 2.0262562245359893e-06,
+      "loss": 1.9522,
+      "step": 21000
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9900407424173837e-06,
+      "loss": 1.973,
+      "step": 21020
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.9538252602987777e-06,
+      "loss": 2.0737,
+      "step": 21040
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.917609778180172e-06,
+      "loss": 1.9752,
+      "step": 21060
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8813942960615663e-06,
+      "loss": 1.9911,
+      "step": 21080
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8451788139429607e-06,
+      "loss": 2.0052,
+      "step": 21100
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8089633318243552e-06,
+      "loss": 2.0729,
+      "step": 21120
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.7727478497057494e-06,
+      "loss": 2.0738,
+      "step": 21140
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7365323675871438e-06,
+      "loss": 2.2317,
+      "step": 21160
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.7003168854685378e-06,
+      "loss": 2.0938,
+      "step": 21180
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6641014033499322e-06,
+      "loss": 2.0353,
+      "step": 21200
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.6278859212313266e-06,
+      "loss": 1.9552,
+      "step": 21220
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5916704391127208e-06,
+      "loss": 2.0546,
+      "step": 21240
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.5554549569941152e-06,
+      "loss": 2.0315,
+      "step": 21260
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.5192394748755094e-06,
+      "loss": 2.173,
+      "step": 21280
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4830239927569038e-06,
+      "loss": 2.0392,
+      "step": 21300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4468085106382978e-06,
+      "loss": 1.9997,
+      "step": 21320
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4105930285196922e-06,
+      "loss": 1.8135,
+      "step": 21340
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.3743775464010866e-06,
+      "loss": 1.8827,
+      "step": 21360
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3381620642824808e-06,
+      "loss": 2.0313,
+      "step": 21380
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3019465821638752e-06,
+      "loss": 2.0068,
+      "step": 21400
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2657311000452694e-06,
+      "loss": 2.0113,
+      "step": 21420
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2295156179266638e-06,
+      "loss": 2.1259,
+      "step": 21440
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.193300135808058e-06,
+      "loss": 1.9679,
+      "step": 21460
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.1570846536894522e-06,
+      "loss": 1.9823,
+      "step": 21480
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.1208691715708466e-06,
+      "loss": 2.0744,
+      "step": 21500
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0846536894522408e-06,
+      "loss": 1.9816,
+      "step": 21520
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0484382073336353e-06,
+      "loss": 2.0606,
+      "step": 21540
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.0122227252150295e-06,
+      "loss": 1.9649,
+      "step": 21560
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 9.760072430964237e-07,
+      "loss": 1.9769,
+      "step": 21580
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.397917609778182e-07,
+      "loss": 2.0627,
+      "step": 21600
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 9.035762788592124e-07,
+      "loss": 1.9936,
+      "step": 21620
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.673607967406067e-07,
+      "loss": 1.9619,
+      "step": 21640
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 8.31145314622001e-07,
+      "loss": 2.0361,
+      "step": 21660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.949298325033952e-07,
+      "loss": 1.8695,
+      "step": 21680
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 7.587143503847895e-07,
+      "loss": 2.1968,
+      "step": 21700
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 7.224988682661839e-07,
+      "loss": 2.0466,
+      "step": 21720
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.862833861475782e-07,
+      "loss": 1.9676,
+      "step": 21740
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.500679040289724e-07,
+      "loss": 2.0507,
+      "step": 21760
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.138524219103667e-07,
+      "loss": 2.0409,
+      "step": 21780
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.77636939791761e-07,
+      "loss": 2.0149,
+      "step": 21800
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.414214576731553e-07,
+      "loss": 2.0,
+      "step": 21820
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.052059755545496e-07,
+      "loss": 2.0153,
+      "step": 21840
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.6899049343594394e-07,
+      "loss": 2.0283,
+      "step": 21860
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 4.327750113173382e-07,
+      "loss": 1.9389,
+      "step": 21880
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.965595291987325e-07,
+      "loss": 2.0778,
+      "step": 21900
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.603440470801268e-07,
+      "loss": 1.9154,
+      "step": 21920
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.2412856496152105e-07,
+      "loss": 1.8948,
+      "step": 21940
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.8972385694884566e-07,
+      "loss": 2.0993,
+      "step": 21960
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.5350837483023997e-07,
+      "loss": 1.9066,
+      "step": 21980
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 2.1729289271163425e-07,
+      "loss": 1.974,
+      "step": 22000
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8107741059302852e-07,
+      "loss": 2.018,
+      "step": 22020
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.4486192847442283e-07,
+      "loss": 2.0365,
+      "step": 22040
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.0864644635581712e-07,
+      "loss": 2.0005,
+      "step": 22060
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 7.243096423721142e-08,
+      "loss": 1.8369,
+      "step": 22080
+    },
+    {
+      "epoch": 2.0,
+      "step": 22090,
+      "total_flos": 2.890306476268585e+17,
+      "train_loss": 2.0700602635478154,
+      "train_runtime": 9050.4463,
+      "train_samples_per_second": 14.644,
+      "train_steps_per_second": 2.441
+    }
+  ],
+  "logging_steps": 20,
+  "max_steps": 22090,
+  "num_train_epochs": 2,
+  "save_steps": 5000,
+  "total_flos": 2.890306476268585e+17,
+  "trial_name": null,
+  "trial_params": null
+}